I wonder if this finding hold across programming languages. I suspect the conclusions would be different for people programming in APL instead of Python, for example.
It surely helps. Take for example History, if I space out for whatever reason in the class and botch my examination on the Aztecs, that doesn't preclude me doing great in the classes about the Roman empire. History is like a sparse suburb, where a single house catching fire does not necessarily propagate and destroy the others. On the other side Math its like a skyscrapper, one floor build upon the other. If someone living in a lower floor botches a home renovation in his unit and damages a pillar, it risks taking down the whole building. In math if you botch a prerequisite, you'll ending messing your learning of subsequent concepts 95% of the time. So it is more unforgiving to people with problems like ADHD, that struggle to keep unbroken focus and discipline.
After I started my PhD program in CS/AI (late 90's), one of the faculty told me that the GRE verbal score was thing on the application most predictive of student success in the program. Of course, most applicants had perfect or near perfect GRE math scores, so there was no predictive power there. But the application also included the GRE subject test, transcripts, and letters of recommendation.
The GRE math bar for excellence was quite low and so a perfect score was only the 92nd percentile or so. It’s not surprising that other metrics correlated better with success
I also seem to recall from The Making of the Atomic Bomb that the theoretical physicists of the era were noted as having the highest verbal intelligence of any group of scientists. That struck me since I would not have expected it.
EDIT: I found the quote, from chapter 6:
Psychometricians have closely questioned American scientists of this first modern generation, curious to know what kind of men they were—there were few women among them—and from what backgrounds they emerged. Small liberal arts colleges in the Middle West and on the Pacific coast, one study found, were most productive of scientists then (by contrast, New England in the same period excelled at the manufacture of lawyers).
...
Theoretical physicists averaged the highest verbal IQs among all scientists studied, clustering around 170, almost 20 percent higher than the experimentalists. Theoreticians also averaged the highest spatial IQs, experimentalists ranking second.
In order to solve what are essentially "word problems" -- i.e. reading a verbal description of a problem and translating it to math -- you have to have very good reading comprehension.
Perhaps survivorship bias explain part of it. Many of those scientists arrived at US fleeing their home countries in WWII. Migrating under those circumstances is tricky, lacking the necessary support network or failing to successfully negotiate an encounter with border patrol may be the difference between reaching a safe place and being dragged to rot in the trenches. Good verbal intelligence is very helpful in this process. Perhaps the same way good verbal intelligence is over-represented in those who escaped the war, the trait of being taciturn was over-represented in all the scientists that got stuck and died in WWII[1].
Makes sense that verbal ability would line up more with success in CS, especially when math scores are already high across the board. A lot of programming leans on language-type skills: reading and understanding code, navigating docs, naming things clearly, writing maintainable logic etc.
The field probably does itself a disservice by overemphasising math. That framing can push people away who might actually do really well, especially those strong in reasoning, abstraction, or communication. Linked study is a good reminder to rethink how we present programming imo.
It does. The majority of what's popular and gets attention in programming is things like languages and frameworks which do the best smooth talking. The whole point of CS is to focus on the aspects of programming that aren't just vogue trends. With any field or aspect of modern life, language skills are the gateway to privilege and software engineering is no exception. But here it's respect for the math that stops it all from spiraling into a bunch of bs. You can't fool the math people.
Probably true. Math used to be written in prose style before we invented symbols to compress the prose. Most of the symbols are very recent just in the last 200 years or so.
Geometry only benefits from visual reasoning in 3d and lower, and there are a lot of dimensions above that!
You can see visual reasoning as a little cheat computation, you can run math problems through your sense-determining brain, which is what brains are really good at (robots struggle with our levels of dexterity). But the fact remains that you can only visualize in low dimensions, and there are infinitely many dimensions.
Note: You can reduce many problems to 3d, but also many problems in 3d have configuration spaces with much higher dimension, so there's some nuance.
I have found a fairly interesting correlation between people who are good at learning programming and people who are good at English spelling bees. Something about holding a lot of anecdotes and esoteric rule exceptions when performing an otherwise algorithmic process.
Makes sense to me; I've always been a naturally good English speller. My brain just knows how to store and retrieve this type of data, which has a high correlation with e.g. CLI interfaces and their idiosyncratic command structures.
I can’t decide if you’ve cheekily included the classic homophonic there-error. This isn’t technically an issue with English spelling, but I think you intended to communicate that CLI interfaces have idiosyncratic command structures, not that they are idiosyncratic structures themselves. If I’ve been taken in by pedant-bait, I apologize.
Lol, nope, just a case of my fingers deciding they know what I'm typing, or possibly autocorrect "fixing" something that was already correct. (I don't know why but the Android GBoard has been introducing crazy numbers of wrong homophones recently.)
I'm a terrible speller; it's taken me ten years of typing "ammend" to learn its proper spelling. It also sort of goes against the "programmers are lazy" meme: why memorize what a computer can detect and correct?
Are you a teacher and/or someone that runs spelling bees or otherwise someone that has done the proper statistics to distinguish this correlation with generalised intelligence. Just wondering where you get this anecdata from?
This totally makes sense to me. I've always been a very good / fast reader, which has been incredibly useful in my programming career. I had a good SAT math score (710) but got a perfect score (800) on the SAT verbal (in the late 90s).
I remember when I first started working on my Master's project on wireless sensor networks, my advisor sat me down and said "I think I know a good project for you. I want you to print out the source code for TinyOS, study it for a week, and come back to me when you think you know enough to make these changes." This was a sort of formative experience for me, and ever since when joining a new project I made sure to take the time to read through the code to understand how things fit together.
Early, strong reader here. Off-the-charts spatial reasoning, as measured by tests.
Terrible at math, I hate it and feel dyslexic trying to read most mathematical writing. I excelled at it in elementary school, then quickly came to feel frustratingly stupid at it as it became less about algorithms (more on that in a bit...) and all about equations and abstract stuff with unknown applications.
However, programming was natural and easy to pick up. I've repeatedly had to take more time convincing myself I actually understand some supposedly "hard" thing, like pointers or recursion, than it took to learn them in the first place, because they were in fact very easy to understand so I kept second-guessing myself—"I must not get it, because that was easy". I've been the go-to guy for "hard" and "low-level" problems basically everywhere I've worked.
What I've noticed is that when I must read math, the only way I can get any headway is to turn everything into steps, through which some example values may pass and affect one another. I have to turn it all into algorithms. Algorithms, I can get. Attempts to express meaning through equations and proofs, no, I have to painstakingly turn every single boundary between every symbol into a step and "walk through" it to have any hope of understanding it, and once I do, this new understanding only barely illuminates the original representation.
I think programming clicked for me because, as typically encountered and taught, it's very heavy on algorithms and very light on other varieties of mathematical presentation. Plus, having so very much more context available about what variables represent and what routines do, than a jumble of letters and symbols. FFS, if we say Perl is line noise, what's mathematical writing? Straight gibberish from a brain-wrecked Cthulhu cultist? Perl's the clearest thing in the world by comparison!
... where I do run into trouble is languages with "mathy" syntax, where the idiomatic style favors single-letter variables and accomplishing-things-by-asserting-equality. I can't read Haskell to save my life. Put the same supposedly-tricky concepts (monads, type classes) in any more-ordinary language, and it's easy, but when I tried to learn them using Haskell, I couldn't get anywhere at all. Shit, it takes me forever just to understand fizzbuzz-level programs written in Haskell.
Same but I am literally math dyslexic (formal diagnosis of dyscalculia).
It prevented me from having a CS degree, I was unable to complete the math courses, but as far as actual programming and "software engineering" goes (design, etc) it's never hindered me. I can work out the logic and I let the computer do the math.
Edit: I'm downvoted below zero for this comment. I don't know what people are so offended by?
Yeah, to be clear, I'm not diagnosed with anything like that, and am only likening the experience to what I imagine it's like for dyslexics of humans language as a ready metaphor—meanwhile, the concepts aren't hard for me, like, a lot of people with no diagnosis of anything fall off the math-wagon right around the time operations on fractions are introduced, but the concepts have always posed no trouble to me and I breezed through early potential trip-ups like that. The style of presentation, specifically, is what gives me such a hard time and is what makes trying to approach even fairly easy "real" mathematics so hard for me.
> It prevented me from having a CS degree, I was unable to complete the math courses, but as far as actual programming and "software engineering" goes (design, etc) it's never hindered me. I can work out the logic and I let the computer do the math.
This is what's wild to me: I have a long, successful career in a "STEM" field that's allegedly math-heavy, while being practically incapable of working with math. Like, it's never even been slightly a problem. I can't relate at all to characterizations of programming as heavy on math. It's never been my experience of it, and at this rate, probably never will be. If it were, I'd for-sure be in a different job.
They probably suspect you to be a slow coder who writes code with bad abstractions because they percieve the maths skills as how one does things in software engineering.
Very interesting, as I am nearly the opposite. Bad spatial reasoning though pretty good at most other things. In school, geometry was fun and interesting - I enjoyed proofs. And I majored in English in college with a CS minor, so I think there must be a variety of in-roads for programming.
I think one "verbal" skill that has served me well is fast reading. When I had to read a 300 page novel once a week you learn how to skim for key elements, which is immensively useful for getting up to speed in a new codebase/language or locating a bug.
Also checking in as someone with strong language skills. When I was a kid in school a million years ago I was actually rather weak in math and crapped out around trig, but excelled in English, Spanish and C. Now I'm a senior web dev, and I still can't do much math beyond the basics required for everyday living (though 18 years in food service means I'm really quick at estimating percentages to within delta)
When I tell you that I was a server and bartender it should all become very clear. Though your guess isn't far off, it's a skill that's served me well since I've picked up breadmaking as a hobby
Oh wow, are you me - I had an almost identical SAT experience. Oddly enough compared to another poster in this thread, I love functional programming and Haskell and studying things like dependent types in a programming context has helped me patch up my crappy understanding of "actual math", proofs, etc.
> I've always been a very good / fast reader, which has been incredibly useful in my programming career. I had a good SAT math score (710) but got a perfect score (800) on the SAT verbal (in the late 90s).
This is me and also many of the CS students in my cohort, and AFAIK something that universities actively selected for, and students also self-selected around in an era of RTFM/MUDs/IRC before LLMs or youtube. The best programmers I've worked with are still always very linguistically brained.. polyglots even when they didn't have to be, or with a long track record of engaging with difficult literature. If nothing else.. just very witty in that certain way that's meta-cognitive, meta-linguistic.
This is still true I think but it's much harder see out in the wild due to the degree/career trajectory popularity. Plus as long we're optimizing for leet-coding even though built-from-scratch algorithms are a very rare need compared to skills with good design/exposition.. naturally the math-brain is favored.
Remember people whining about weak ecosystems? That's a lot of opportunity to build things from scratch instead of making another - for real this time - IRC clone with Electron.
I came here to comment the same -- also ready to share my SAT scores from the 1980s. I was also much highter on the verbal side and I have had an awesome software engineering career using many, many languages (often together) on large, professional projects at large companies anybody here would recognize.
This mirrors my personal experience. My writing skills have been above average and my parents' presumption was that I would pursue some humanistic studies. I ended up studying software engineering and made a career out of it of almost three decades and counting. Meanwhile my wife who always had a "math brain" has struggled to learn to program at a decent competency level.
Interestingly, I became better at learning human languages after learning a couple of programming languages. I was good at math at the beginning. I guess programming kinda bridged the gap.
How much math is in a typical CS program these days? Calc 1-3 (maybe 3, varied by school), Linear Algebra, Statistics (inconsistent across programs), Discrete were pretty much it when I was in school 25+ years or so ago. That's only 4-6 courses depending on the university, though some where the CS dept was more strongly associated with an engineering college might have added Diff Eq and others. (I got interested in CS education and reviewed a lot of curricula in the US at the time, this is from memory.)
Some schools like MIT might have required more, but on average what I wrote was about it. Has it increased since then? Based on the new hires I've seen the last decade I'd have guessed the math requirements were mostly the same.
I'd argue none of that math is really necessary. While I have used most of my classes at least once, it was never a barrier to advancement in my career. Hell you could say the same thing about any of the theory. Like yea it's cool I know what a "merkle tree" is but it ultimately is a distraction from most of the skills you need to work with git.
Anyway, both computation and math are grouped under "apriori" knowledge. Any semantic distinction is ultimately silly. But we could just as easily be teaching programming as a craft in the context of the real world—I think this is closer to how it's done outside the US. I am not at all convinced the American style is what people ought to be paying for.
When I did my CS degree in New Zealand there were just two mandatory maths papers - statistics and discrete mathematics. Would be wrong to say I didn't get anything from them - but I'm not fumbling around truth tables or poisson distributions all that often either. Everything else was pretty standard: intro to programming, DSA, low level programming, compilers and networks. What I do find kind of mind blowing is comparing my lectures with the ones from MIT and CM (on YouTube) where they can't go more than a few seconds without jumping into math. Ultimately I'm left unconvinced I was deprived of anything important as a typical software engineer.
It was my grades in math that ultimately failed me out of my undergraduate CS program. My university had: Calculus 1, Calculus 2, Linear Algebra, Vector Geometry, Multi-Variable Calculus, Applied Combinatorics, Discrete Math, Differential Equations and maybe more that I don't remember. So many that CS majors could take one more math class and get a minor.
Yeah, I never thought this made sense, but so many people did; and, I always hear people on Slashdot talking about how programming IS math. None of that has been my personal experience, and I'm coming up on 21 years as software engineer. Discrete was the ONLY math course that I really enjoyed and did well in the first time around. For me, this always made sense.
I can count the times I've ever applied math past approximately high school algebra 1, on one hand. Period, in private life, in hobbies, at work. I'm not sure I've ever used any "college level" math, for anything at all.
I've, and other programmers I've known, gotten excited on the very few occasions anything even slightly mathematically-tricky came up, precisely because it almost never happens.
This makes sense for my path to math. In high school I was bad at math and good at learning languages. Then I started learning Python and realized that it was just like learning a language. Then at some point I realized math notation was just another language for expressing the kinds of things you could express in Python. Now I'm in a job where I do math every day and I read math textbooks for fun.
It helped that Python was meant to resemble natural language. I had learned C++ and Perl before but they never stuck, because I never made the connection to language. Ironically, since Perl was designed by a linguist!
I don't think it's fair to attribute anything to anything. Stuff comes from all over the place. In other words, attributing programming prowess to math was a mistake, and we are making the same mistake again attributing it to language.
---
Just one more:
Kurt Vonnegut: --consider himself in competition with a world's champion. And this is one reason good writers are unlikely to come from an English Department. It's because the English Department teaches you good taste too early.
I think his main point is when we put something on a pedestal, we actually limit people, whether that be math or language.
My experience is that they spit out reasonably looking solutions but then they don't even parse/compile.
They are OK to create small spinets of code and completion.
Anything past that they suck.
It's actually hilarious that AI "solved" bullshiting and and artistic fields much better and faster than say reasoning fields like math or programming.
It's the supreme irony. Even 5 years ago the status quo was saying artistic fields were completely safe from the AI apocalypse.
I disagree that the current generation of AI has "solved" artistic fields any more than it's solved math or programming.
Just as an LLM may be good at spitting out code that looks plausible but fails to work, diffusion models are good at spitting out art that looks shiny but is lacking in any real creativity or artistic expression.
> "looks shiny but is lacking in any real creativity or artistic expression."
My experience with that is that artistic milieus now sometimes even explicitly admit that the difference is who created the art.
"Human that suffered and created something" => high quality art
"The exact same thing but by a machine" => soulless claptrap
It's not about the end result.
A lot could be written about this but it's completely socially unacceptable.
Whether an analogous thing will happen with beautiful mathematical proofs or physical theories remains to be seen. I for one am curious, but as far as art is concerned, in my view it's done.
Truly great art, the kind that expands the field of artistry and makes people think, requires creativity; if you make something that's just a rehashing of existing art, that's not truly creative, it's boring and derivative.
This has nothing to do with whether a human or AI created the art, and I don't think it's controversial to say that AI-generated art is derivative; the models are literally trained to mimic existing artwork.
Creativity in AI art production is a fancy term for temperature that adds no semantic value.
Your "creativity" is just "high temperature" novel art done by the right person/entity.
This was something already obvious to anyone paying attention. Innovation from the "wrong people" was just "sophomoric", derivative or another euphemism, but the same thing from the right person would be a work of genius.
I would say the fact they do better than anyone could have imagined using just language points in the same direction, okay is very much up for debate imo.
It's like people enjoy extrapolating their surprise when it comes to LLMs, and I don't think it's very helpful.
100% agree, I've been saying this for years. I'm terrible with arithmetic but great with symbols and relations. Recursion is also fundamentally linguistic, and although our internal "stacks" for processing it naturally are quite small, language remains the easiest demonstration of recursion in our daily lives.
Oddly, I also use spatial intuition when thinking about stuff like stacks and the shape of data structures.
Recursion itself is simply a conjecture. Nothing fundamental about it unless you believe Chomsky, but his is a speculative claim, not empirical per se.
I don't really know what you mean by "conjecture", but I thought apriori was implied by positing it as a linguistic construct. "Fundamental" doesn't imply empiricism at all. All of apriori knowledge for a language is a set of all sets of coherent statements: the outer set represents a set of implied axioms required to make the statements cohere. Recursion just broadens the complexity of the statements you can express, but it's fundamentally a concept that arises from language and can be evaluated for coherency (like all other apriori concepts).
Edit: added a definition of apriori knowledge.
Edit2: to put this another way, nobody is arguing that recursion doesn't exist. Or that it is empirically-derived. No, it's a useful construct to show certain relations.
Edit3: added a sentence
Edit4: The extent to which our own grammars are inherently recursive vs this being culture or technology is irrelevant to identifying the concept of recursion as an apriori, linguistic concept.
Edit5: i suppose you might also be referring to the idea that we naturally process recursion. I mean, we clearly, evidently do; whether or not that's inherent to being human is a separate question entirely. Hell in the free software world there's a whole recursive acronym meme that taps into some part of our brain and tickles it.
It kinda is empirically true that human language is recursive. Every human language ever discovered is recursive, except, supposedly, for one: Pirahã. And Pirahã has mainly been described by one researcher whose results are controversial.
Language has an inherently recursive structure: I saw the man who saw the man who saw the man who saw the man who saw the man who... While our brains have practical limits to how deeply such things can actually be nested, language has a recursive tree-like aspect to it.
Yes, but “language is fundamentally recursive” doesn’t mean the same thing as “recursion is fundamentally linguistic”. Language is just one example of a recursive structure.
I am also unsure whether recursion is fundamentally linguistic, but I thought that “language remains the easiest demonstration of recursion in our daily lives” to be useful. If I ever write another essay about recursion, I'll now consider starting with a linguistic example before diving into recursive functions or data structures.
Language has nothing that corresponds to a recursive function, so that is a bad example. You can write a sentence that could correspond to a call to a recursive function, but its not the same thing as a recursive function.
If recursion was just writing the function 10 times like you did in language then people wouldn't struggle with it.
Recursive functions are just a subset of all possibly recursive concepts. In the case of human spoken language, the recursion exists in our characterization of the grammar. You could just as easily frame this in "iterative" terms just like you can make any recursive function iterative, but that's less convenient for analysis.
So in this case, "recursive function" would be "clause" or something like that; I'm no linguist. But clauses can embed clauses which can embed further clauses, etc.
I think your usage of recursive functions is just high-level logic—you're describing an inductive proof. We also frame a lot of our social games as recursive processes. But these are conscious processes that we can evaluate consciously; the recursion in spoken language is largely unconscious and very shallow.
> In the case of human spoken language, the recursion exists in our characterization of the grammar
But people are constructing sentences, not grammars. When you construct a grammar you can add a recursive part to it, that is true, just like in a programming language, but constructing grammars is not what people mean with language skills.
A sentence can't be recursive since languages in themselves has no concept of applying a concept, for that you need an interpretation of the language references. For example, you can have a recursive function written in a programming language that doesn't have a recursive grammar, the concepts are different things.
There are two ways that recursion intersects with language that are relevant here:
1. Our spoken and especially written grammar is recursive. We do handle this unconsciously. This is not related to our ability to reason about recursion at a high level, and recursive grammars are not necessary to do so. This is not a skill in the normal sense and we have only (very) limited ability to improve our capacity to interpret deeply nested grammars. However, this is still a useful illustration of what recursion IS, which is why I brought it up.
2. Language also introduces the ability to semantically reason about recursiveness. This is still a linguistic thing—you need a symbol and relations among symbols in order for recursion to be meaningful—but this is a skill and is likely very related to linguistic skill. This is the part that really helps you to program: ultimately, you're just reasoning about symbols and looking for incoherency.
Can you come up with some conception of recursion that doesn't involve symbols referring to themselves, directly or indirectly? Ie what is left of recursion when you remove the linguistic component?
This is just a guess on my part, but I'd also bet that writing inductive proofs (or proofs in general) require more of the language brain than just doing math problems.
I agree that there's enormous value in carving out mathematics from other linguistic reasoning, but I don't see defining as something as mathematic rather than linguistic is generally useful. You use the same skills to look for incoherency in both situations, but human language is generally expected to be incoherent on some level.
Besides, a lot of what people mean when they say they're bad at math is that they're bad at arithmetic, which is honestly understandable.
If you define recursion as a symbol referencing itself, either directly or indirectly, and if you define language as a system of relating symbols to each other, recursion is a linguistic concept, it is a concept that describes a relationship between symbols. There are good reasons to define each concept differently, but if you identify recursion empirically, recursion won't "actually" exist outside of the description of the process. It's our characterization of the process that reveals the recursive structure, even if that characterization doesn't actually exist outside of language.
> If you define recursion as a symbol referencing itself, either directly or indirectly, and if you define language as a system of relating symbols to each other, recursion is a linguistic concept
But that isn't what we mean with recursive function. We don't call this recursive:
> We don't call this recursive... it's just incrementing x
That's not a recursive function as it's written, but you could certainly consider it a form of symbolic recursion. This just isn't a very useful characterization in an iterative/imperative context. You could frame incrementing as recursive, though—this is just peano axioms/church encoding.
I can't speak for other forms of FP, but symbol operators make communicating about haskell very annoying. Outside of that FP seems to be doing fine, IMO.
To be clear, the symbols themselves don't bother me so much as trying to refer to them in spoken english. I have no particular beef with the use of symbols in code, which can be quite readable.
Good code doesn’t just solve a problem, it solves it in a way that’s readable and modular.
I think the problem-solving part of coding requires math skills, while the organization part requires writing skills. The organization part affects the problem-solving part, because if you write messy code (that you can’t reread once you forget or extend without rewriting) you’ll quickly get overwhelmed.
Writing large math proofs also requires organization skills, since you’ll refer to earlier sections of your proof and may have to modify it when you encounter issues. But to me, math seems to have more “big steps”: sudden insights that can’t be derived from writing (“how did you discover this?”), and concepts that are intrinsically complicated so one can’t really explain them no matter how well they can write. Whereas programming has more “small steps”: even someone who’s not smart (but has grit) can write an impressive program, if they write one component at a time and there aren’t too many components that rely on each other.
Most coding doesn't need much of any math past boolean logic and very basic set operations. I'm much more likely to spend my time studying DB and interface schemas to understand how something works than doing a lot of mathy fiddling. Sure, some people write game engines and such, but even much of 3D graphics doesn't need anything more complicated than the first half of a linear algebra course.
It's funny, reading the post you're replying to, I basically read it as
> I don't need math, I need <math, but by another name>
My teenage daughter used to complain about math, and I spent some time trying to explain to her that we use math every day... EVERY day. Now, when I see her do something that was math (even if it's not obvious it was math), I say "Math... every day". I say that a lot.
Also, yes, my daughter finds me annoying. But also funny; but likely not for the math thing.
DBs aren't math. In math it's perfectly ok for computation to take infinite time, you can just assume it already completed, but DBs don't work like that.
Now we're getting into the "define maths" part of the discussion which is always where these discussions die. It can be argued that turning a kettle on and boiling some water is "maths" or it can be as narrow as "everything above basic arithmetic is logic, not maths."
So how much of programming is maths? Before we answer that, let's answer: How much of maths is actually maths? Because first we define maths, and then we define programming based on whatever that is, but until we have that first concrete definition this discussion cannot occur.
I will add that "it is taught by the maths department in college" is a flimsy argument, and frankly one the Physics department in particular would mock.
I think for this discussion, "math is stuff you'd learn in a math department" is a pretty useful definition, even if it's not a very good one. There's a lot of math involved in the design and manufacture of the kettle, electrical grid, and water utilities, but a person's ability to put a kettle on isn't going to be improved by math classes. In that way, programming probably is a bit mathy, but good programming is more like good technical writing than it is like math.
> I think for this discussion, "math is stuff you'd learn in a math department" is a pretty useful definition
That means that definition shifts over time. For example, courses on numerical analysis, graph algorithms, programming, and on compilers used to be part of “what you’d learn in a math department”.
It likely also even today will show geographical variation.
Overlap between programming and math is quite small, e.g. halting problem is solvable in programming (by flow analysis) but isn't solvable in math. Programming deals only with practical problems and can be completely guided by practical considerations, while math requires abstract outlandish skills - exact opposite. Why talk about math at all if it's already well known that programming is engineering?
It's definitely a gray area. Is a DAG traversal algo "math", or is it more computer-sciencey? What if you do it in SQL? Certainly there's a mix of more or less concentrated logic/math vs glue code, and most of that is very dependent on the domain you're working in.
I find this distinction useful in the abstract, that one can engage different parts of the brain for different components of development. This probably explains why a well-written DSL can be so powerful in the right context.
Computer science is math in the same way that physics is philosophy, in that computer science certainly started as math, just like the natural sciences used to be subdisciplines of philosophy.
But it's hardly a useful grouping any more. You can study and do well in computer science with minimal knowledge of most of the core mathematical subjects.
While graph theory certainly crosses over into math, you can cover most of the parts of it relevant to most computer science as a discussion of algorithms that would not be the natural way of dealing with them for most mathematicians.
> You can study and do well in computer science with minimal knowledge of most of the core mathematical subjects.
You will fail at Theoretical Computer Science without mathematical proficiency. Go read some textbooks and papers in theoretical CS. It is a subfield of mathematics. Theorems and proofs. Rigorous and difficult mathematics.
I've read plenty of theoretical computer science papers over the last 30+ years, and while some of it requires "rigorous and difficult mathematics" that is by no means universal.
I wrote my MSc thesis on the use of statistical methods for reducing error rates for OCR, and most of the papers in my literature review hardly required more than basic arithmetic and algebra.
So I stand by my statement.
Sure, there are subsets of computer science where you need more maths, just like in any field there are sub fields where you will need to understand other subjects as well, but that does not alter what I claimed.
EDIT:
Some authors are quicker to pull out the maths than others, and frankly in a lot of CS papers maths is used to obscure lack of rigor rather than to provide it. E.g the problem I ran across when writing my thesis was that once you unpacked the limited math into code you'd often reveal unstated assumptions that were less than obvious if you just read their formulas.
Is it possible your mental model of what CS is more aligned with software engineering rather than actual CS? Could you share some examples of what you consider to be CS but lacks any mathematical relation?
I agree is not a useful grouping in practice. I'm just interested in what makes you think like you do.
I did categorically not claim, nor even suggest, that any CS "lacks any mathematical relation".
What I claimed was that in computer science we often discuss things in terms that would not be the natural way of dealing with it in maths. We do that because our focus is different, and our abstractions are different.
It doesn't mean it's not math. It means it's not useful to insist that it isn't a different field, and its obtuse when people insist it's all the same.
> Most coding doesn't need much of any math past boolean logic and very basic set operations
Coding IS math.
Not "coding uses math", I mean it is math.
Mathematicians do not deal in objects, but in relations among objects; they are free to replace some object by others so long as the relations remain unchanged. Content to them is irrelevant; they are interested in form only.
- Poincare[0]
I don't know how you code, but I don't think I'm aware of code that can't be reasonably explained as forming relationships between objects. The face we can trace a program seems to necessitate this.
But that doesnt necessarily mean successful programmers are good at conventional math. This is why certain people in the department are identified as "math people".
I'm not sure why you'd think I disagree. It seems you understood I argued that it's unhelpful to make the distinction between math and "conventional" math
By that same logic you could also say that language is math. In fact I think your quote kind of disproves your point because the content/state of a program is super important in coding more than the form.
Coding used to be very close to pure math (many early computer science classes were taught in the Math Department in universities) but it has been so far abstracted from that to the point that it is its own thing and is as close to math as any other subject is.
> By that same logic you could also say that language is math
Not quite, but the inverse is true. The language to math direction doesn't work because a lack of formalism. I can state incomprehensible sentences or words. (There's an advantage to that in some cases!) but when you do that with code you get errors and even you do it with math its just that there's no compiler or interpreter that tells at you
I think you misunderstand what "paradox" means. While it can mean "self-contradictory" it can also mean "contrary to one's expectation." Math uses both, but in very different contexts.
The contradiction is used in proof formulation, specifically to invalidate some claim. I don't think this is what you're implying.
The latter is what it contextually sounds like you're stating; things like the Banach-Tarksi Paradox. There's no self-contradiction in that, but it is an unexpected result and points to the need to refine certain things like the ZFC set theory.
I'd also stress that there are true statements which cannot be proven through axiomatic systems. The Halting Problem is an example of what Godel proved. But that's not contradictory, even if unexpected or frustrating.
I don't think that quote really supports coding and math being equivalent. To me, the quote provides an abstraction of math through a structuralist perspective. Language can also be viewed through this abstraction. I think coding could share the abstraction, but that doesn't make the three of these fields equivalent.
Well a lot of people did wildly misunderstand the Poincare quote. To me is is obviously about abstraction and I think this is true for any mathematician. I thought it would also be natural for programmers considering we use "object" quite similarly, if not identically. So... maybe it is or maybe this is the joke.
That sounds like you're backing off from your original claim, probably because it is impossible to defend.
That you can use mathematics to describe code doesn't seem very different from using math to describe gravity, or the projected winner in an election, or how sound waves propagate.
Isn't the primary purpose of math to describe the world around us?
Then it shouldn't be surprising that it can also be used to describe programming.
In the real world, however, software engineering has nothing to do with mathematical abstractions 99% of the time
A programmer constructs a function from some data type to another while a mathematician constructs a function from witnesses of some proposition to another?
Though interpreting a CRUD app as a theorem (or collection of theorems) doesn’t result in an interesting theorem, and interpreting a typical theorem as a program… well, sometimes the result would be a useful program, but often it wouldn’t be.
Interpreting a CRUD apps (or fragments of them) as theorems is interesting (given a programming language and culture that doesn't suck)! e.g. if you have a function `A => ZIO[Any,Nothing,B]`, then you have reasonable certainty that barring catastrophic events like the machine going OOM or encountering a hardware failure (essentially, things that happen outside of the programming model), that given an A, you can run some IO operation that will produce a B and will not throw an exception or return an error. If you have an `A => B`, then you know that given an A, you can make a B. Sounds simple enough but in practice this is extremely useful!
It's not the type of thing that gets mathematicians excited, but from an engineering perspective, such theorems are great. You can often blindly code your way through things by just following the type signatures and having a vague sense of what you want to accomplish.
It's actually the halting problem that I find is not relevant to practical programming; in practice, CRUD apps are basically a trivial loop around a dispatcher into a bunch of simple functions operating on bounded data. The hard parts have been neatly tidied away into databases and operating systems (which for practical purposes, you can usually import as "axioms").
> It's not the type of thing that gets mathematicians excited
Says who? I've certainly seen mathematicians get excited about these kinds of things. Frequently they study Programming Languages and will talk your ear off about Category Theory.
> You can often blindly code your way through things by just following the type signatures and having a vague sense of what you want to accomplish.
Sounds like math to me. A simple and imprecise math, but still math via Poincare's description.
> in practice, CRUD apps are basically a trivial loop around a dispatcher into a bunch of simple functions operating on bounded data
In common settings. But those settings also change. You may see those uncommon settings as not practical or useful but I'd say that studying those uncommon settings is necessary for them to become practical and useful (presumably with additional benefits that the current paradigm doesn't have).
I think we're in agreement. My comment about the halting problem was meant to refer to Rice's theorem (the name was slipping my mind), which I occasionally see people use to justify the idea that you can't prove interesting facts about real-world programs. In practice, real-world programming involves constantly proving small, useful theorems. Your useful theorem (e.g. `Map[User,Seq[Account]] => Map[User,NetWorth]`) is probably not that interesting to even the category theorists, but that's fine, and there's plenty you can learn from the theorists about how to factor the proof well (e.g. as _.map(_.map(_.balance).sum)).
> Isn't the primary purpose of math to describe the world around us?
No, that's Physics[0]. I joke that "Physics is the subset of mathematics that reflects the observable world." This is also a jab at String Theorists[1].
Physicists use math, but that doesn't mean it is math. It's not the only language at their disposal nor do they use all of math.
> software engineering has nothing to do with mathematical abstractions 99% of the time
I'd argue that 100% of the time it has to do with mathematical abstractions. Please read the Poincare quote again. Take a moment to digest his meaning. Determine what an "object" means. What he means by "[content] is irrelevant" and why only form matters. I'll give you a lead: a class object isn't the only type of object in programming, nor is a type object. :)
[0] Technically a specific (class of) physics, but the physics that any reasonable reader knows I'm referencing. But hey, I'll be a tad pedantic.
[1] String Theory is untestable, therefore doesn't really reflect the observable world. Even if all observable consequences could be explained through this theory it would still be indistinguishable from any other alternative theory which could do so. But we're getting too meta and this joke is rarely enjoyed outside mathematician and physicist communities.
Going on a total tangent, if you'll forgive me, and I ask purely as a curious outsider: do you think math could have ever come into being if it weren't to fill the human need of describing and categorizing the world?
What would have been the very beginning of math, the first human thought, or word or action, that could be called "math"? Are you able to picture this?
> do you think math could have ever come into being if it weren't to fill the human need of describing and categorizing the world?
I'm a bit confused. What exactly is the counterfactual[0] here? If it is hyper-specific to categorizing and describing then I think yes, those creatures could still invent math.
But my confusion is because I'm having a difficult time thinking where such things aren't also necessary consequences of just being a living being in general. I cannot think of a single creature that does not also have some world model, even if that model is very poor. My cat understands physics and math, even though her understandings are quite naive (also Wittgenstein[1] is quite wrong. I can understand my cat, even if not completely and even though she has a much harder time understanding me). More naive than say the Greeks, but they were also significantly more naive than your average math undergrad and I wouldn't say the Greeks "didn't do math".
It necessitates a threshold value and I'm not sure that this is useful framing. At least until we have a mutual understanding of what threshold we're concerned with. Frankly, we often place these contrived thresholds/barriers in continuous processes. They can be helpful but they also lead to a lot of confusion.
> What would have been the very beginning of math
This too is hard to describe. Mull over the Poincare quote a bit. There's many thresholds we could pick from.
I could say when the some of the Greeks got tired of arguing with people who were just pulling shit out of their asses, but that'd ignore many times other civilizations independently did the same.
I could say when the first conscious creature arose (I don't know when this was). It needed to understand itself (an object) and its relationship to others. Other creatures, other things, other... objects.
I could also say the first living creature. As I said above, even a bad world model has some understanding that there are objects and relationships between them.
I could also say it always was. But then we get into a "tree falls in a forest and no one is around to hear it" type of thing (also with the prior one). Acoustic vibrations is a fine definition, but so is "what one hears".
I'd more put the line closer to "Greeks" (and probably conscious). The reason for this is formalization, and I think this is a sufficient point where there's near universal agreement. In quotes because I'll accept any point in time that can qualify with the intended distinction, which is really hard to pin-point. I'm certainly not a historian nor remotely qualified to point to a reasonable time lol. But this also seems to be a point in history often referenced as being near "the birth" and frankly I'm more interested in other questions/topics than really getting to the bottom of this one. It also seems unprovable, and I'm okay with that. I'm not so certain it matters when that happened.
To clarify, I do not think life itself necessitates this type of formalization though. I'm unsure what conditions are necessary for this to happen (as an ML researcher I am concerned with this question though), but it does seem the be a natural consequence of a sufficient level of intelligence.
I'll put it this way, if we meet an alien creature I would be astonished if they did not have math. I have no reason to believe that their math would look remotely similar to ours, and I do think there would be difficulties in communicating, but if we both understand Poincare's meaning then it'll surely make that process easier.
Sorry, I know that was long and probably confusing. I just don't have a great answer. Certainly I don't know the answer either. So all I can give are some of my thoughts.
>Even if all observable consequences could be explained through this theory it would still be indistinguishable from any other alternative theory which could do so.
That's not unique, all quantitative theories allow small modifications. Then you select parsimonious theory.
"There are those who tell us that any choice from among theoretically-equivalent alternatives is merely a question of taste. These are the people who bring up the Strong Church-Turing Thesis in discussions of programming languages meant for use by humans. They are malicious idiots. The only punishment which could stand a chance at reforming these miscreants into decent people would be a year or two at hard labor. And not just any kind of hard labor: specifically, carrying out long division using Roman numerals." — Stanislav Datskovskiy
Coding is math in the sense that coding is a subset of math.
Mathematics is a very extensive field, and covers a vast amount of subjects.
For the same reason it can't be said that mathematics is equivalent to coding, as there are many things in mathematics that are not relevant for coding.
However, by far the most interesting parts of coding are definitely related to mathematics.
Working with different objects doesn't make it any less of math. Just because you can derive calculus from set theory (analogous to assembly or even binary here) doesn't make calculus "not math".
Math is about abstractions and relations. See the Poincare quote again.
Plus, the Programming Languages people would like to have a word with you. Two actually: Category Theory. But really, if you get them started they won't shut up. That's either a great time or a terrible time, but I think for most it is the latter.
Wheeler: It from bit. Otherwise put, every it—every particle, every field of force, even the space-time continuum itself—derives its function, its meaning, its very existence entirely—even if in some contexts indirectly—from the apparatus-elicited answers to yes-or-no questions, binary choices, bits. It from bit symbolizes the idea that every item of the physical world has at bottom—at a very deep bottom, in most instances—an immaterial source and explanation; that which we call reality arises in the last analysis from the posing of yes–no questions and the registering of equipment-evoked responses; in short, that all things physical are information-theoretic in origin and that this is a participatory universe.
That is not in contention with what I said, now is it?
Wheeler is arguing that if a tree falls in the forest, and there is nobody to hear it, that it still makes a sound because there are things that interact with the sound. But if the tree fell in a forest and there was nothing else in the universe then there is no sound because there is no observation.
It helps to read the whole thing[0] and to understand the context of the discussion. This is meta-physics and a deep discussion into what the nature of reality is. Ian Hacker has a good introduction to the subject but I find develop grave misunderstandings when they also do not have the strong math and physics background necessary to parse the words. Even people who understand the silliness of "The Secret" and that an observer need not be human often believe that this necessitates a multi-verse. A wildly convoluted solution to the problem of entropy not being invertible. Or closer to computer terms, a solution that insists that P = NP. There is information lost.
If you wanna argue that there's no difference between the word cup and a cup itself because there is no word without the observer who has the language, then yeah.
Programming is an expression of logic, which is absolutely mathematics.
But then we also have to think about naming variables and classes, structuring our code so that it is more readable by other developers, and so on. That's less about formal reasoning and more about communication.
There is an engineering aspect to programming (prototyping, architecture, optimization, etc). It's a combination of mathematics and engineering. Software Engineering.
Of course they do, but that part is not "mathematics". It is communication. If they use the English language to write the proof that doesn't mean English is also mathematics.
Structuring sentences and naming variables so that it is easier for other people to understand is less about formal mathematical reasoning, and more about communication.
You could name a variable x, y, or Banana, but it doesn't change the logic.
Neither is it "the point" in programming. You should be concerned with communication and have every right to get upset when someone is being needlessly convoluted but that's as much of a point in programming as it is in math, physics, or any domain.
I mean the reason we get mad at this is because it is someone destroying "society" in some sense. Even if that society is your team or just the set of programmers. It would be a pretty dick move were I to just use a word that significantly diverged from conventional meaning and expected you to mull it over. Similarly if I drop a unknown out of context math equation. It would be meaningless.
And I'm on your side, really! I strongly advocate for documenting. And let's be real, the conclusion of your argument more strongly argues for documentation than good variable names. Because variable names are much more constrained and much more easily misinterpreted considering how any word has multiple definitions. Surrounding code is often insufficient to derive necessary contextualization.
The presumably-mathematicians who wrote my Algorithms textbook (CLRS) didn't seem to think giving the variables in their algorithms useful names. They just use i, x, etc. all over the place and don't even consider giving things actual names. This makes the book much less accessible than it would be if they treated it more like most people write code.
That's how it is in applied mathematics. That's a hardcore computer science textbook. If your job is computer science research, inventing new algorithms, that is the kind of book you will get used to. There are better options for practical learning.
I do agree with your point btw, but I did want to note that there are good conventions around symbols. The brevity is heavily influenced by the medium. Variable names sucked when you had punch cards. It's still burdensome to write long names when using paper, chalkboard, whiteboard, or any system that doesn't have autocomplete.
In general, lower case letters are used as constants, excluding x,y,z,t,i,j,k (sometimes u,v,w). It isn't a hard rule, but strong preference to begin at the beginning of the alphabet for these. Capital letters usually are held for things like Variable Sets (like random variables). Greek letters need context for constants or variables. BB and Cal typefaces for sets (e.g. Real Numbers, Integers). And much more.
I think a lot of the difficulty in it is that these "rules" or patterns are generally learned through usage and often not explicitly stated. But learning them can really help read unfamiliar topics and is why "notation abuse" leads to confusion. But after all, math is all about abstraction so technically any symbol will do, but no doubt some are (significantly) better than others for communicating.
There are two hard things in Computer Science:
- Cache Invalidation
- Naming Things
- Off-by-One Errors
The murky world of software patents would like a word about that.
For me, coding also feels artistic at times. But we see artistic beauty in mathematics all the time, so no wonder. But often I look at ugly code, and maybe that is subjective, but fixing that code makes it also feel prettier.
It is clear all developers have their owm artistic style, and probably that is why there is so much disagreement in the industry. Maybe we are lacking the pure mathemetical language to describe our intent in a more beutiful and precise way that is more clearly the-right-way. As in how we find beauty in simple physics equations.
Arguably writing a novel is math, if you use the right definition of math. But sometimes its more helpful to use more informal definitions that capture what people mean then what is technically accurate.
No, not always. Quite a lot of high-level code doesn't require any math at all. It doesn't take math to perform CRUD operations, which account for a lot of programming work. Sure, the underlying machine code is all based on math, but the higher level programming doesn't need to involve a single math equation for it to be useful. Let's see where the goalposts move now...
>All of that code is a series of logical statements and expressions. Mathematical logic.
Which code? The machine code that underlies everything? Or the lines of simple high-level CRUD that don't even need a single number or logic statement to function? Not all programming has to be mathematical or even logical at a high enough level, and I say this as someone who's been coding assembly language for 40 years.
Storing a string in some abstraction is not a mathematical operation. I'm done with this thread, it's going way too far down too many rabbit holes. The quarks that make up the matter that the computers are made of are pure "math". There, now I've moved the goalposts.
That is a narrow perspective of mathematics and computer science.
Assigning a variable is a mathematical operation. It is a logical state transition. This is formalized in Programming Language Theory.
We are not talking about quarks. We are talking about lines of code which are objectively mathematical statements, no matter how non-mathematical they seem to you subjectively.
You start with a practical problem to solve, and a computing machine that can perform logical operations on data. Your job is to figure out a correct sequence of logical operations that will solve the problem. The program that you have written is a mathematical structure. It is mathematics.
You confuse map with the territory. By the same logic you can say quarks are mathematics, because they are modeled by a theory that has some mathematics in it.
That may be true to some extent, but I think you are missing the point. Quarks are physical in nature, and code is logical in nature. Programs themselves are formal systems. Code isn’t just modelled by mathematics, it is defined by mathematics.
In the case of code, we can argue that the map is the territory.
Code is logical if you define logic as reasoning in general, broader than mathematics, and since it runs in physical environment, it now interacts with a messy world. Code is defined by business requirements, and there's no mathematics for that.
The discussion started about whether human needs math skills to write code. That's what I mean when I say programming isn't mathematics. Meaning of code is defined by human, how do you intend code to be defined by mathematics? The human first imagines mathematical formulas, then by some unknown process they become code? I don't think anybody does it like that. You start with requirements, intuitively guess architecture, then decompose it, it works more like Cauchy problem (but less numeric, more conceptual, you have an owl, now draw it), but I don't think anybody models it like that.
>Once the code is actually written, that exists as a formal logical system, defined by mathematics
I still think that's not code, but your favorite model of code. For spellchecker language is defined by mathematics too: it splits text into words by whitespace, then for each word not found in dictionary it selects best matches and sorts them by relevance. Oh and characters are stored as numbers.
> The discussion started about whether human needs math skills to write code.
Writing code IS a math skill. When writing code you are writing logic in a formal system. Logic is mathematics.
You may be thinking that mathematics is just like doing arithmetic or solving equations. It is way deeper and broader than that.
> I still think that's not code, but your favourite model of code
Code is not just modelled through mathematics, it is actually defined by mathematics. It is fundamentally a mathematical construct, grounded in formal semantics.
Our mathematical models of reality (i.e. Physics) is not the same as reality. So yes, by the same logic "quarks" are mathematics, but only in the same way that a "cup" is English. The quark was a great example, considering we can't observe it and purely rely on mathematical models, but the same argument would still hold true for a bowling ball. Physics is our description of reality, not reality itself. Our description of reality highly leverages the language of math, as that language was developed to enforce rules of consistency. Much the same way we design our programming languages, which is why programming language people study so much more math than your typical CS person.
If you're going to accuse someone of confusing the map with the territory, you really should make sure you aren't making the same error.
How math helps with programming languages? What math says about zero based indexes? How do you prevent off by one errors? How do you prevent buffer overflows? It's ergonomics problems.
It is hard to answer because of exactly what ndriscoll said[0]
> It's like trying to argue about the distinction between U(1), the complex numbers with magnitude 1, and the unit circle, and getting upset when the mathematicians say "those are 3 names for the same thing". Or saying that writing C is programming but writing in a functional language like Scala or Haskell (or Lean) is not.
As ndriscoll suggests, it is tautological. I mean look at what I said. I really need you to hear it. I said that coding is math. So what I hear is "How programming languages helps with programming languages?" Why are you expecting me to hear anything different?
> What math says about zero based indexes?
Start at 0? Start at 1? Who cares, it is the same thing. The natural numbers, non-negative integers, integers, even integers, who cares? They're the same thing. And who cares about indexing at 0 or 1 in programming? That's always been a silly argument that's inconsequential.
> How do you prevent off by one errors?
By not being off by one? What's the question? Like being confused about if you start at 0 or start at 1 and how to get the right bound? It is a shift from one to the other, but they are isomorphic. We can perfectly map. But I really don't get the question. You can formalize these relationships with equations you know. I know it isn't "cool" but you can grab a pen and paper (or a whiteboard) and write down your program structure if you are often falling for these mistakes. This seems more about the difficulties of keeping track of a lot of things in your head all at once.
> How do you prevent buffer overflows?
By not going over your bounds? I'm so confused. I mean you are asking something like "if f(x) = inf when x > 10, how does math help you prevent the output of the function from being infinite?"
Maybe what will help is seeing what some of the Programming Languages people do and why they like Haskell[1].
Or maybe check out Bartosz Milewski[2,3]. His blog[2], is titled "Bartosz Milewski's Programming Cafe: Category Theory, Haskell, Concurrency, C++". It may look very mathy, and you'd be right(!), but it is all about programming! Go check out his Category Theory Course[3], it is _for programmers_.
Don't trust me, go look at papers published in programming language conferences [4]. You'll find plenty of papers that are VERY mathy as well as plenty that are not. It really depends on the topic and what is the best language for the problems they're solving. But you'll certainly find some of the answers you're looking for.
Seriously, don't trust me, verify these things yourself. Google them. Ask an LLM. I don't know what to tell you because these are verifiable things (i.e. my claims are falsifiable!). The only thing you need to do is look.
Nonmathematical problems are difficult to answer when you try to find mathematical answers. It should be obvious, why it doesn't work.
Mentioning Scala is ironic, it's very light on math spik and to begin with was created to unify object oriented with functional programming, which is mathematically meaningless, because both are Turing complete and thus equivalent, tautological.
>So what I hear is "How programming languages helps with programming languages?"
Oh, right, in mathematics axiom is argument, but in reality it isn't. In programming you should assert what you assume, otherwise your assumptions can be wrong due to divergence from reality, but there no reality in mathematics, only fantasy, so you can't understand this with mathematics alone.
Yes it does. Just because the objects you are working with aren't numbers doesn't mean it isn't math. In fact, that's my entire point. It is why I quoted Poincare in the first place. He didn't say "numbers" he said "objects".
Sorry but you are wrong. "CREATE, READ, UPDATE, DELETE" operations are not mathematical in nature. It doesn't matter what the data is or how esoteric you want to get with it - a programmer doesn't need any math at all for simple CRUD. You're trying to move the goalposts all the way back to the 1800s to win a pointless internet argument.
CREATE, READ, UPDATE, DELETE are fundamentally mathematical in nature.
In a typical implementation these are database operations. That involves relational algebra operations, state transitions, boolean logic.
The READ part can be a very complex SQL query (composed of many algebraic operations) but even the simplest query (SELECT * from t where ID = 1) is filtering a set based on a predicate. That is mathematics.
No one is moving goalposts. Set theory and logic are at the foundations of mathematics.
It's almost the argument of programming vs computer science coming out here.
This is math:
{x | x.id = 1}
OTOH, a SQL query is a SQL query.
This thread is hilarious though. It's like
- Cashier: here is your change.
- Customer: you did math!
- Cashier, no, I gave you change.
- Customer: that IS math!
- Cashier: You mean, I used math to give you change?
- Customer: No, giving change doesn't use math, it IS math!!!!" [2]
= D
Moving along.. FWIW, indeed SQL was created to model set theory (relational algebra and tuple calculus), the close relationship is no accident of course [0][1])
> No one is moving goalposts
I feel too they are.
First goal post:
> Coding IS math. >> No, not always. Quite a lot of high-level code doesn't require any math at all. It doesn't take math to perform CRUD operations
Second goal post:
> CREATE, READ, UPDATE, DELETE are fundamentally mathematical in nature
CRUD is closely related to SQL, and SQL is closely related to various mathematics. Are they identical and therefore equivalent? No - because your database is not going to like it when you write "{x | x.id = 1}", and the Oracle DB might not like something that you can write for your Postgres DB.
The problem you're running into is that people who have some "mathematical maturity" don't get bogged down in notation, so it's difficult for them to see the distinction you're trying to draw between e.g. `{ x∈S | x.id = 1}` and `select x from S where x.id = 1`[0]. You say "a SQL query is a SQL query" and they just think "yes, which is also obviously a mathematical expression".
Computer programs are proofs[1]. This is intuitively and also formally true. You would agree writing proofs is doing math, yeah? Then obviously writing a computer program is also doing math.
Like I have a degree in math and have been a software engineer for over a decade. I do not know what distinction people are trying to get at. It's like trying to argue about the distinction between U(1), the complex numbers with magnitude 1, and the unit circle, and getting upset when the mathematicians say "those are 3 names for the same thing". Or saying that writing C is programming but writing in a functional language like Scala or Haskell (or Lean) is not.
[0] Modulo details like NULLs and multi-set semantics, but surely that's not the distinction?
A better analogy is if someone got upset that someone else said "the set of natural numbers is the set of real numbers." One is a subset of course, and when that is highlighted, the response is "yeah, by 'is', I actually mean subset", therefore indeed: the set of natural numbers is the set of real numbers.
This is an interesting example: "Or saying that writing C is programming but writing in a functional language like Scala or Haskell (or Lean) is not."
The first part is everything we need to look at. Are we saying that writing C is equivalent and equal to the entirety of all programming? That if you're programming then you are writing C code. No, there is an implied "is a form of" in there. Given the other clarifications and that so many people are claiming to be mathematicians, I would have expected the precision to say exactly "C is a form of programming" rather than "C is programming."
Turns out, the analogy of saying "the set of reals is the set of naturals" is more fitting compared to sets that are actually equal.
I understand you have strong opinions, I just don't understand why.
In math are highly concerned with structures and abstraction. We have things called operators. They aren't just addition and multiplication. We also use those words to describe different operations. They have things like groups, rings, fields, and algebras. Yes, plural.
The purpose of these things is to create logical frameworks. It matters not what the operators are. Nor does it matter what objects we operate on. Poincaré is explicitly saying this.
The part you're not understanding is the abstraction. This is what math is about. It is also why the Programming Language people are deeper in the math side and love Category Theory (I have a few friends who's PL dissertations are more math heavy than some math dissertations I've seen). It's no surprise. What's a function? How do you abstract functions? How do you define structure? These are shared critical questions. PL people are more concerned with types but I wouldn't say that makes them any less of a mathematician than a set theorist.
We can perfectly describe the CRUD operations with set theory. Do most programmers need concern themselves with this? Absolutely not. But is it something people designing those operations and systems is thinking about? Yes.
I'd encourage you to learn some set theory, abstract algebra, and maybe a bit of cat theory. It'll make these things pretty clear. But I'd caution about having strong opinions on topics you aren't intimately familiar with. Especially when arguing with those that are. Frankly, CRUD is a terrible example. I'm confident a quick google search (or asking a GPT) would quickly point you to relational algebra. It's discussed in most database classes. It certainly was in the one I was an assistant for.
You could say that math is behind the structure of a leaf, but the farmer doesn't care about that. Keep moving the goalpost all you want, I don't have time to argue these pointless things, and I stopped reading after the first sentence of your comment. I'm done here.
> I understand you have strong opinions, I just don't understand why.
As someone that is "on the other side of the fence", and getting flamed for it, maybe I can shed some light since I have the other perspective as well. IMO, The reason for not seeing eye to eye is (for example) akin to saying "word problems are math". (Thinking of a grade school word problems for common reference). Yes, they are readily mapped to mathematical models that can solve the word problem & perhaps almost indistinguishably so. Though no - word problems are not math. Word problems are a series of phrases and words. That's where the talking past each other comes in... Different interpretations of "word problems are math", or "code is math". It's seemingly not clear whether we are talking about logical 'implies', 'element of', or 'equals'.
Which goes to "We can perfectly describe the CRUD operations with set theory.", we all agree there. That is not readily conveyed though when writing things like "code is math".
> That's where the talking past each other comes in...
Then sorry, but that's your own damn fault. I was clear about my definition and quoted a famous mathematician to give some authority, to not be "trust me even though I'm some rando". The way to respond to that is not "you're wrong, trust me, I'm some rando".
Yes, I agree we're misunderstanding each other because we're using different definitions but "you've" rejected "mine" without defining "yours" and expecting everyone to understand. Of course that'll lead to confusion. You can reject the mathematicians definition of math, but you sure gotta say more than "trust me" and it's a pretty wild thing to do, especially as non mathematicians.
The problem here is one side who's dabbled in cooking says "a chef makes desserts" and chefs are responding "we do a lot more than that". Maybe there's a few chefs that go "yeah all I do is dessert" and everyone points to that while ignoring the second part of their sentence is "but that's just my speciality." Wouldn't you think that conversation is insane? The only reason it's obviously so is because we all know what a chef is and agree on the definition. But who is better qualified to define the chef's job? The chef or consumer?
The way I'm responding I'd more characterize as: "wait, if what you are saying is true, then this other thing should be true too, but it does not seem to be. That would indicate what you are saying is not true."
In another thread, you characterized my response as stating: " ¬(A ↦ B) ⟹ ¬(B ↦ A)" (and this is a great example of language not being math, but math being language!). That was not at all my claim.
My claim is "I believe you are saying 'A = B'. It appears that 'B != A', therefore 'A != B'." My only claims are
(1) I believe you are writing to convey that you mean Math IS Language in the sense they are equal, identical, interchangeable, and fully equivalent, and bi-directionally so
(2) that: B != A
The only results can either be:
- "yeah, because B != A, the statement A = B is not true"
- Your claim (1) is false, I'm not actually saying "A = B"
- Your claim (2) is false, "B = A" is in fact true. I would find that to be an interesting assertion and would have loved to explore more why you think that.
That is a good analogy, except programming languages are formal languages, not natural languages.
> they are readily mapped to mathematical models
With code we are not talking about something that is mapped to mathematical models. Code is not modelled by mathematics, it is defined by mathematics (denotational semantics and operational semantics).
"Code is math" is true in a theoretical sense. Practically, coding doesn't always feel like what is commonly thought of as "doing mathematics", but it is essentially mathematical.
A while back I made this[0] as a quick demo of generic batching for CRUD requests. It's very much practically oriented (it's for performance optimization while making the code nicely reusable/not muddying up the business logic), but also felt quite a bit like the same sort of things I did in my math degree.
Actually I'm starting to wonder whether the thing that made university math easy for me was I quickly developed a good internal "type checker"/"type inference engine" in my head, and that helped make the next steps of proofs seem straightforward.
Mathematics is not "equations". Most of mathematics is not related with calculus either.
For anything CRUD, you put the data into a database. The underlying database is relational, or a key->store.
If it's relational, that's one branch of mathematics. If it's another kind of database, it's another branch of mathematics. Mathematics is extensive, and covers more things you can imagine at a glance.
The main difference between writing mathematics and programming, and this applies to any form of programming, is that in mathematics writing a formal proof is amazingly close to: you write the program, and you are also the compiler, and you are the CPU, performing all operations yourself. With programming you only have to write the program.
Source: On one hand I have studied pure mathematics (not the simplified applied mathematics that are taught in engineering, which is mostly equations), on the other hand I have been working as a software developer for over 15 years.
This is vague and doesn't mean anything. People can't even agree what 'objects' are and people did a lot of programming before the term 'object' was invented.
Programming is about is fundamentally about instructions and data. Yin and yang of two things that are completely different, even if people can occasionally mix them and conflate them.
Category theory is often jokingly called the study of dots and arrows. What's a dot? Anything. What's an arrow? A relationship.
I'm surprised I hit a nerve with so many people. I'm quoting someone who's considered one of the greatest mathematicians. Obviously I don't know the backgrounds of people but it seems like programmers have strong opinions on what math is that disagrees with what mathematicians say they do.
Now it's not math it's 'category theory' and that's dots and arrows and dots are anything.
At what point does the abstract description just not offer any useful insight anymore?
I'm surprised I hit a nerve with so many people.
A lot of people had very good explanations for why you're pretty far off in trying to say two things are the same.
Programming is much more like building a machine than any sort of straight math. There is state, there is interactivity, there is performance, there is IO and there are real world implications to all of it.
Saying they are the same is like saying gardening is plumbing just because you sometimes use a hose.
That's a wild claim considering Category Theory is a branch of mathematics
| Category theory is a general theory of mathematical structures and their relations.
- https://en.wikipedia.org/wiki/Category_theory
It is necessary that you provide an alternative definition as to what "category theory" is, though I suspect it will make many category theorists and mathematicians upset.
That's a wild claim considering Category Theory is a branch of mathematics
It's not a wild claim since you misquoted me.
A lot of non-mathematicians disagreed with mathematicians.
Mathematicians can claim whatever they want, when it comes to programming, programmers understand it better and they're trying to explain to you why this is nonsense. Vanderbilt claims to be "the harvard of the south" but wouldn't you know it, harvard doesn't claim to be "the vanderbilt of the north".
Show me programming languages designed by mathematicians to be 'mathematically pure' and I'll show you a language that hasn't been used to ship software that people want to use.
Boolean logic and set theory are very important to computer logic, and graph theory is not far behind. But you can also learn Boolean logic in the philosophy dept. Which I unfortunately learned the hard way by taking Philosophy 101 after already having the CS class on logic. Took a semester to go over what we did for the midterm. Got a lot of naps that semester.
Probability is math and there you have caching, hashing, etc. Then there are permutations, combinations, etc. with a lot of usecases in software. Distributing graphs across nodes? More math.
> Whereas programming has more “small steps”: even someone who’s not smart (but has grit) can write an impressive program, if they write one component at a time and there aren’t too many components that rely on each other.
... In my experience, learning to write one component at a time (and try the code, and make sure it works before proceeding) is itself a skill that many struggle to develop. Similarly for avoiding unnecessary dependencies between components. Oh, and also being able to analyze the problem and identify separable components.
One of the most frustrating things about teaching programming, for me, is the constant insistence from other teachers that you have to maintain an "absolutely everyone can learn to program" attitude at all times. Many people who start to learn programming have misguided or confused reasons for doing so and - to say the least - could make much more effective use of their time developing other skills. (It's not a question of elitism; I'd surely flounder at some tasks that others find natural.)
I dislike "everyone can learn to program" because it conflates many, many levels of skills and expertise.
I very much think many people could learn the more advanced Excel Formulas, Power Automate and even simple Bash/PowerShell scripting to make their work more effective. I've met quite a few folks who had been intimidated out of trying who could do it.
On the other hand, how many people on this site could bootstrap a linux kernel on either very new or very old hardware? I know there are some, but they are certainly not the majority. I certainly won't be the first person to get linux and doom to run on a quantum computer.
But that is similar to other professions. Everyone with a largely functioning body can learn to turn a few planks and some metal parts into a functional shed door with some basic tools or to put up a decent brick wall that won't topple over in a month.
That doesn't mean everyone is qualified to pour concrete for a dam or a bridge foundation, or to re-do some historical work in original style.
> Everyone with a largely functioning body can learn to turn a few planks and some metal parts into a functional shed door
It's shocking how little physical and spatial ability some people have - that is definitely not true. Sometimes it might be a personal discount or lack of confidence, but this remains true regardless of the cause.
> That doesn't mean everyone is qualified to pour concrete for a dam or a bridge foundation, or to re-do some historical work in original style.
Exactly!
I think statements like that are more concerned with philosophy than reality. Any discussion surrounding topics like this typically ends up being a discussion around definitions.
I believe the vast majority of human beings are capable of learning how to program in the most extreme elementary sense of the word. As in, outside of severe disabilities or complete and utter inaccessibility to circumstances in which one could learn program, then I think the remaining population of people could learn to program to some degree. Obviously, not everyone will learn to program due to a near infinite number of reasons.
I would argue it's like music. Anyone can make 'music.' Just make a sound -- any sound. The difference between noise and music is subjective. I would not argue that everyone could be the next Lovelace, Turning, Ritchie, Thompson, Torvalds, etc..
Now, for my jaded opinion, I think a lot of the "everyone can learn to program" talk does not come from a place of desire to share the gift of knowledge and joy of programming. I think it's more of a subtle way to encourage people to go into programming so that they may be hired by mega corps. In order to keep the Capitalist machine running. It's like the National Hockey League's slogan, "Hockey is for everyone." That is just a fancy way of saying, "everyone's money can be spent on the NHL."
I mean - not "absolutely everyone" can learn to do just about anything. There's a wide distribution of intelligence, aptitude/ability and desire amongst folks amongst all of the various things that you might learn (not to mention learning disabilities, impairments, etc).
I might be capable of learning advanced accounting, but that sounds like torture to me and I'll be damned if I'll ever take that on! I'm sure programming feels like that to a wide variety of people, and I don't see any need for us to try to pretend otherwise - outside of a bizarre ideological desire for equivalent outcomes from disparate groups.
I guess it depends on what is meant by "everyone can learn to program". It's a bit like saying everyone can learn to write, or to do math, or play tennis, etc.
I'm sure everyone is capable of learning some basic level of programming, just as they are able to learn a basic (high school) level of any subject. However, not everyone is going to have the aptitude to take that to an advanced professional level, however hard they try. We're not all cut out to be artists, or writers, or doctors, or scientists, or developers, etc.
Personally I've always considered a solid grasp of algebra to be the minimum bar for being able to program, at least for anything that isn't utterly trivial. Being able to take a word problem and turn it into a system of equations and solve it is a pretty close analog to being able to take some sort of problem or business requirement and turn it into code.
And the sad truth is that a huge percentage of the population struggle with just arithmetic, let alone algebra.
I recall reading about a professor at uni which carefully designed his tests throughout the year so he could determine if a student had grasped a certain programming topic. He also tracked effort, through handins and such.
After collecting data for a few semesters he concluded his students could be clearly divided into three categories: those who just "got" programming, those who understood it after working hard, and a small group that just didn't grasp regardless of effort.
What do you mean by math? Abstractions or calculations? The effort it takes me to do long division in my head matches the effort it takes me to follow some obtuse spaghetti code. See, I can get good and fast at long division in my head, but I may never understand the fundamental theorem of calculus. Some people are really good at mucking around garbage code (they have no choice, they get paid to), but what part of programming did they get good at? Obviously, some part of it, but nothing to write home about. Whenever I sense that I'm just getting practice at doing the equivalent of mental long division at work, that's when I always seek a new job. No amount of money is worth falling behind like that.
I'm thinking of "computation", "intuition", and "organization".
Computation is following an algorithm. e.g. long division or computing a derivative.
Intuition, AKA brilliance, is finding a non-obvious solution. Think "solving an NP problem without brute force"*. e.g. solving an integral (in a form that hasn't already been memorized) or discovering an interesting proof.
Organization is recording information in a way that a) is easy for you to recall later on (and get insights from) and b) is digestible by others**. e.g. explaining how to compute a derivative, solve an integral, or anything else.
Math, programming, and writing each require all skills. The kind of math taught in school (e.g. long division) and your boring jobs are primarily computation. I believe advanced math (e.g. calculus) is primarily intuition; it requires some organization because big theories are broken into smaller steps, but seems to mostly involve smart people "banging their head against the wall" to solve problems that are still quite unclear***. Programming is primarily organization. It requires some intuition (I think this is why some people seemingly can't learn to code), but in contrast to math, most programs can be broken into many relatively-simple features. IMO implementing all the features and interactions between them without creating a buggy, verbose, and unmaintainable codebase is programming's real challenge. Writing is also primarily organization, but finding interesting ideas requires intuition, and worldbuilding requires computation (even in fiction, there must be some coherence or people won't like your work).
> Some people are really good at mucking around garbage code (they have no choice, they get paid to), but what part of programming did they get good at? Obviously, some part of it, but nothing to write home about.
I agree that work you find boring should be avoided, and I also try to avoid working with it. But some people really seem to like working on esoteric code, and I think there are some skills (beyond computation) developed from it, that even apply when working with good code. Building a mental model of a spaghetti codebase involves organization, and if the codebase uses "genius hacks", intuition. Moreover, the same techniques to discern that two code segments in completely different locations are tightly coupled, may also discern that two seemingly-separate ideas have some connection, leading to an "intuitive" discovery. There's an MIT lecture somewhere that describes how a smart student found interesting work in a factory, and I think ended up optimizing the factory; the lesson was that you can gain some amount of knowledge and growth from pretty much any experience, and sometimes there's a lot of opportunity where you'd least expect it.
* Or maybe it is just brute force but people with this skill ("geniuses") do it very fast.
** These are kind of two separate skills but they're similar. Moreover, b) is more important because it's necessary for problems too large for one person to solve, and it implies a).
*** And whatever method solves these problems doesn't seem to be simplification, because many theories and proofs were initially written down very obtuse, then simplified later.
The problem with doing things right the first time is some people look at it and just say, “well of course it should work that way.” Yes but did you think of doing it that way?
Its very rare imo that computational problems emerge fully formed & ready to be tackled like proofs.
Usually even deciding what the problem is is in part an art, requires an act of narrativization, to shape and form concepts of origin, movement, and destination.
A good problem solver has a very wide range of abstract ideas and concepts and concrete tools they can use to model and explain problem, solution, & destination. Sometimes raw computational intellect can arrive at stunningly good proposals, can see brilliant paths through. But more often, my gut tells me it's about having a breadth of exposure, to different techniques and tools, and being someone who can both see a vast number of ways to tackle a situation, and being able to see tradeoffs in approaches, being able to weight long and short term impacts.
> Its very rare imo that computational problems emerge fully formed & ready to be tackled like proofs.
In my generation, the perfect example is Python's Timsort. It is an modest improvement upon prior sorting algorithms, but it has come to dominate. And, frankly, in terms of computer science history, it was discovered very late. The paper was written in 1993, but the first major, high-impact open source implementation was not written until 2003. Ref: https://en.wikipedia.org/wiki/Timsort
It has been reimplemented in a wide variety of languages today. I look forward to the next iteration: WolfgangSort or FatimaSort or XiaomiSort or whatever.
The a capital example of an exception that prove the rule.
I absolutely value & have huge respect for the deeply computational works that advance us all along!
But this is an exceedingly rare event. Most development is more glue work than advancing computional fundamentals. Very very very little of the industry is paid to work on honing data structures so generally.
+1. Excellent description of how the skills relate. I often read, listen to poetry when stuck in a math or programming problem. ... or just talk to the rubber duck. :-)
I'm not a mathematician, but I figure mathematicians aim for clean, composable abstractions the same way programmers do. Something complicated, not just complex in its interactions with other things, seems more useful as a bespoke tool (e.g. in a proof) than as a general purpose object?
> Whereas programming has more “small steps”: even someone who’s not smart (but has grit) can write an impressive program, if they write one component at a time and there aren’t too many components that rely on each other.
This is well put. I often wonder if a merely average working memory might be a benefit to (or at least may place a lower bound on the output quality of) a programmer tasked with writing maintainable code. You cannot possibly deliver working spaghetti if you can't recall what you wrote three minutes ago.
> You cannot possibly write or deliver spaghetti, working or otherwise, if you're not capable of remembering what you wrote three minutes ago.
Forth programmers make a similar point. Forth is stack based; you typically use stack operations rather than local variables. This is ok when your 'words' (analogous to functions/procedures) have short and simple definitions, but code can quickly become unreadable if they don't. In this way, the language strongly nudges the programmer toward developing composable words with simple definitions.
(Of course, Forth sees little use today, and it hasn't won over the masses with its approach, but the broader point stands.)
> Good code doesn’t just solve a problem, it solves it in a way that’s readable and modular.
Strongly disagree.
There are plenty of cases where non-modular code is more readable and faster than modular code (littered, presumably, with invocations of modular logic).
There are also countless cases, particularly in low-level languages or languages with manual memory management, where the best solution -- or the correct solution -- is far from readable.
Readability is anyways in the eye of the beholder (My code is readable. Isn't yours?) and takes a back seat to formal requirements.
Just as a matter of personal style, I prefer long, well-organized functions that are minimally modular. In my experience, context changes -- file boundaries, new function contexts, etc -- are the single greatest source of complexity and bugs, once one has shipped a piece of code and needs to maintain it or build on it. Modular code, by multiplying those, tends to obscure complexity and augment organizational problems by making them harder to reason about and fix. Longer functions certainty feel less readable initially but I'd wager they produce better, clearer mental models, leading to better solutions.
I think the important concept that "readable and modular" is trying to get at is how easy is it to continue working on the code in future. There's definitely codebases that are easier to work on than others, even when the domain is the same.
I'd say that readability, which often boils down to consistency, and modularity are ways to do this, but they aren't the only ways. And as you say, sometimes there's a need for "unreadable" code, so not everything can be easy.
A lot of people in this type of threads always makes the same mistake: confusing what math is with branches of math, or rather, ways in which math is used. The way the education system is built certainly contributes to this.
I've always found the car metaphor to work very good to understand this: A car is a machine that can transport itself to point A to B (some other rules apply). There are different types of cars, but you certainly haven't understood the definition of you say that something is not a car because is not a Volvo, or because it doesn't look like a Ford, when it's clearly able to transport itself.
Math is the study of ideal objects and the way they behave or are related to each others. We have many branches of mathematics because people have invented so many objects and rules to play with them. Programming is nothing if not this very definition. The fact that you don't have to "use math" when programming is not really addressing the point, it's like saying a car is not a car because it has no discernible brand.
"Time is not a thing in math" is not understanding what math is. Time is another ideal object following certain rules under a given domain. Programming is coming up with objects of different size, with different characteristics, with interact at different points in time, i.e. following certain rules.
I tell people all the time that the single greatest tool an aspiring lawyer can have is a background with programming, as the analytical and algorithmic mindset is FAR more important than being "good at public speaking" or any of the other base skills often cited as desirable for lawyers. I've also said the second greatest tool an aspiring lawyer can have (in my personal opinion) is a significant background in foreign language learning, as that is a skill that is closely related to programming, though a bit abstracted and coming from a different angle. I'm going to see if I can use this article to support that.
The article is extremely misleading, I dare even say almost malignant.
The study itself claims:
- fluid reasoning and working-memory capacity explained 34%
- language aptitude (17%)
- resting-state EEG power in beta and low-gamma bands (10%)
- numeracy (2%)
They take math skills to equal numeracy. The study itself implies this too. I disagree on a fundamental level with that. Math skills align much more closely to fluid reasoning than to numeracy.
This is largely true, but also, the dynamic shifts depending on how learners engage with peers as they move from understanding to synthesis.
In small peer groups (“pods”) that debug and learn together, communication becomes a core skill—and that can actually change how math skills are applied and developed. Language doesn’t just support learning; it reshapes the process.
I feel like I'm using language part of the brain for almost everything. My performance at any task (except purely manual and automatic ones) drops to 0-15% when I can hear someone talking. Alternatively if I manage to focus on the task I immediately stop understanding what's being said.
I'm very curious what this math pretest looked like, whether it was "proper" high level math or, like, computing some trig problems. Folks who have aptitude with algebra or number theory or topology, I'd expect that to be correlated, but not to rote computational math.
I don't find this too surprising. The study itself was primarily just testing a students ability to identify syntax and remember what various functions do. I wouldn't expect math proficiency to help much in this area vs I would very much expect language to.
It'd be interesting to see correlations (language brain vs math brain) for how easy or hard it is for people to solve new problems with language after they already know the basics.
464 comments
[ 3.2 ms ] story [ 312 ms ] threadEDIT: I found the quote, from chapter 6:
Psychometricians have closely questioned American scientists of this first modern generation, curious to know what kind of men they were—there were few women among them—and from what backgrounds they emerged. Small liberal arts colleges in the Middle West and on the Pacific coast, one study found, were most productive of scientists then (by contrast, New England in the same period excelled at the manufacture of lawyers).
...
Theoretical physicists averaged the highest verbal IQs among all scientists studied, clustering around 170, almost 20 percent higher than the experimentalists. Theoreticians also averaged the highest spatial IQs, experimentalists ranking second.
[1] https://worldscientific.com/worldscibooks/10.1142/q0436#t=ab...
The field probably does itself a disservice by overemphasising math. That framing can push people away who might actually do really well, especially those strong in reasoning, abstraction, or communication. Linked study is a good reminder to rethink how we present programming imo.
You can see visual reasoning as a little cheat computation, you can run math problems through your sense-determining brain, which is what brains are really good at (robots struggle with our levels of dexterity). But the fact remains that you can only visualize in low dimensions, and there are infinitely many dimensions.
Note: You can reduce many problems to 3d, but also many problems in 3d have configuration spaces with much higher dimension, so there's some nuance.
I remember when I first started working on my Master's project on wireless sensor networks, my advisor sat me down and said "I think I know a good project for you. I want you to print out the source code for TinyOS, study it for a week, and come back to me when you think you know enough to make these changes." This was a sort of formative experience for me, and ever since when joining a new project I made sure to take the time to read through the code to understand how things fit together.
Terrible at math, I hate it and feel dyslexic trying to read most mathematical writing. I excelled at it in elementary school, then quickly came to feel frustratingly stupid at it as it became less about algorithms (more on that in a bit...) and all about equations and abstract stuff with unknown applications.
However, programming was natural and easy to pick up. I've repeatedly had to take more time convincing myself I actually understand some supposedly "hard" thing, like pointers or recursion, than it took to learn them in the first place, because they were in fact very easy to understand so I kept second-guessing myself—"I must not get it, because that was easy". I've been the go-to guy for "hard" and "low-level" problems basically everywhere I've worked.
What I've noticed is that when I must read math, the only way I can get any headway is to turn everything into steps, through which some example values may pass and affect one another. I have to turn it all into algorithms. Algorithms, I can get. Attempts to express meaning through equations and proofs, no, I have to painstakingly turn every single boundary between every symbol into a step and "walk through" it to have any hope of understanding it, and once I do, this new understanding only barely illuminates the original representation.
I think programming clicked for me because, as typically encountered and taught, it's very heavy on algorithms and very light on other varieties of mathematical presentation. Plus, having so very much more context available about what variables represent and what routines do, than a jumble of letters and symbols. FFS, if we say Perl is line noise, what's mathematical writing? Straight gibberish from a brain-wrecked Cthulhu cultist? Perl's the clearest thing in the world by comparison!
... where I do run into trouble is languages with "mathy" syntax, where the idiomatic style favors single-letter variables and accomplishing-things-by-asserting-equality. I can't read Haskell to save my life. Put the same supposedly-tricky concepts (monads, type classes) in any more-ordinary language, and it's easy, but when I tried to learn them using Haskell, I couldn't get anywhere at all. Shit, it takes me forever just to understand fizzbuzz-level programs written in Haskell.
It prevented me from having a CS degree, I was unable to complete the math courses, but as far as actual programming and "software engineering" goes (design, etc) it's never hindered me. I can work out the logic and I let the computer do the math.
Edit: I'm downvoted below zero for this comment. I don't know what people are so offended by?
> It prevented me from having a CS degree, I was unable to complete the math courses, but as far as actual programming and "software engineering" goes (design, etc) it's never hindered me. I can work out the logic and I let the computer do the math.
This is what's wild to me: I have a long, successful career in a "STEM" field that's allegedly math-heavy, while being practically incapable of working with math. Like, it's never even been slightly a problem. I can't relate at all to characterizations of programming as heavy on math. It's never been my experience of it, and at this rate, probably never will be. If it were, I'd for-sure be in a different job.
I think one "verbal" skill that has served me well is fast reading. When I had to read a 300 page novel once a week you learn how to skim for key elements, which is immensively useful for getting up to speed in a new codebase/language or locating a bug.
This is me and also many of the CS students in my cohort, and AFAIK something that universities actively selected for, and students also self-selected around in an era of RTFM/MUDs/IRC before LLMs or youtube. The best programmers I've worked with are still always very linguistically brained.. polyglots even when they didn't have to be, or with a long track record of engaging with difficult literature. If nothing else.. just very witty in that certain way that's meta-cognitive, meta-linguistic.
This is still true I think but it's much harder see out in the wild due to the degree/career trajectory popularity. Plus as long we're optimizing for leet-coding even though built-from-scratch algorithms are a very rare need compared to skills with good design/exposition.. naturally the math-brain is favored.
Some schools like MIT might have required more, but on average what I wrote was about it. Has it increased since then? Based on the new hires I've seen the last decade I'd have guessed the math requirements were mostly the same.
Anyway, both computation and math are grouped under "apriori" knowledge. Any semantic distinction is ultimately silly. But we could just as easily be teaching programming as a craft in the context of the real world—I think this is closer to how it's done outside the US. I am not at all convinced the American style is what people ought to be paying for.
Yeah, I never thought this made sense, but so many people did; and, I always hear people on Slashdot talking about how programming IS math. None of that has been my personal experience, and I'm coming up on 21 years as software engineer. Discrete was the ONLY math course that I really enjoyed and did well in the first time around. For me, this always made sense.
I can count the times I've ever applied math past approximately high school algebra 1, on one hand. Period, in private life, in hobbies, at work. I'm not sure I've ever used any "college level" math, for anything at all.
I've, and other programmers I've known, gotten excited on the very few occasions anything even slightly mathematically-tricky came up, precisely because it almost never happens.
You chose CS but really wanted Software Engineering. Discrete Mathematics was all you needed for that.
It helped that Python was meant to resemble natural language. I had learned C++ and Perl before but they never stuck, because I never made the connection to language. Ironically, since Perl was designed by a linguist!
Kurt Vonnegut: See, I came up through a chemistry department.
Charlie Rose: Yeah, right.
Kurt Vonnegut: And so I wrote and there was nobody there to tell me whether it was any good or not. I was just making my soul grow, writing stories.
There's some stuff about his opinion on training for writing that could be relevant:
https://charlierose.com/videos/25437
I don't think it's fair to attribute anything to anything. Stuff comes from all over the place. In other words, attributing programming prowess to math was a mistake, and we are making the same mistake again attributing it to language.
---
Just one more:
Kurt Vonnegut: --consider himself in competition with a world's champion. And this is one reason good writers are unlikely to come from an English Department. It's because the English Department teaches you good taste too early.
I think his main point is when we put something on a pedestal, we actually limit people, whether that be math or language.
https://www.youtube.com/watch?v=ROyr83rib3Q
My experience is that they spit out reasonably looking solutions but then they don't even parse/compile.
They are OK to create small spinets of code and completion.
Anything past that they suck.
It's actually hilarious that AI "solved" bullshiting and and artistic fields much better and faster than say reasoning fields like math or programming.
It's the supreme irony. Even 5 years ago the status quo was saying artistic fields were completely safe from the AI apocalypse.
Just as an LLM may be good at spitting out code that looks plausible but fails to work, diffusion models are good at spitting out art that looks shiny but is lacking in any real creativity or artistic expression.
My experience with that is that artistic milieus now sometimes even explicitly admit that the difference is who created the art.
"Human that suffered and created something" => high quality art
"The exact same thing but by a machine" => soulless claptrap
It's not about the end result.
A lot could be written about this but it's completely socially unacceptable.
Whether an analogous thing will happen with beautiful mathematical proofs or physical theories remains to be seen. I for one am curious, but as far as art is concerned, in my view it's done.
This has nothing to do with whether a human or AI created the art, and I don't think it's controversial to say that AI-generated art is derivative; the models are literally trained to mimic existing artwork.
Your "creativity" is just "high temperature" novel art done by the right person/entity.
This was something already obvious to anyone paying attention. Innovation from the "wrong people" was just "sophomoric", derivative or another euphemism, but the same thing from the right person would be a work of genius.
It's like people enjoy extrapolating their surprise when it comes to LLMs, and I don't think it's very helpful.
Oddly, I also use spatial intuition when thinking about stuff like stacks and the shape of data structures.
You sure about that? How about inductive proofs?
I would just say that language is more familiar to most. Mathematics are also languages, but more formal and foreign to most.
Edit: added a definition of apriori knowledge.
Edit2: to put this another way, nobody is arguing that recursion doesn't exist. Or that it is empirically-derived. No, it's a useful construct to show certain relations.
Edit3: added a sentence
Edit4: The extent to which our own grammars are inherently recursive vs this being culture or technology is irrelevant to identifying the concept of recursion as an apriori, linguistic concept.
Edit5: i suppose you might also be referring to the idea that we naturally process recursion. I mean, we clearly, evidently do; whether or not that's inherent to being human is a separate question entirely. Hell in the free software world there's a whole recursive acronym meme that taps into some part of our brain and tickles it.
With that argument everything is fundamentally linguistic since everything is communicated using a language.
Can you come up with a more reasonable argument?
If recursion was just writing the function 10 times like you did in language then people wouldn't struggle with it.
So in this case, "recursive function" would be "clause" or something like that; I'm no linguist. But clauses can embed clauses which can embed further clauses, etc.
I think your usage of recursive functions is just high-level logic—you're describing an inductive proof. We also frame a lot of our social games as recursive processes. But these are conscious processes that we can evaluate consciously; the recursion in spoken language is largely unconscious and very shallow.
But people are constructing sentences, not grammars. When you construct a grammar you can add a recursive part to it, that is true, just like in a programming language, but constructing grammars is not what people mean with language skills.
A sentence can't be recursive since languages in themselves has no concept of applying a concept, for that you need an interpretation of the language references. For example, you can have a recursive function written in a programming language that doesn't have a recursive grammar, the concepts are different things.
1. Our spoken and especially written grammar is recursive. We do handle this unconsciously. This is not related to our ability to reason about recursion at a high level, and recursive grammars are not necessary to do so. This is not a skill in the normal sense and we have only (very) limited ability to improve our capacity to interpret deeply nested grammars. However, this is still a useful illustration of what recursion IS, which is why I brought it up.
2. Language also introduces the ability to semantically reason about recursiveness. This is still a linguistic thing—you need a symbol and relations among symbols in order for recursion to be meaningful—but this is a skill and is likely very related to linguistic skill. This is the part that really helps you to program: ultimately, you're just reasoning about symbols and looking for incoherency.
What does this mean exactly?
> What does this mean exactly?
What does this mean exactly?
- Defining recursion is linguistic
- Defining a function recursively is mathmatic
Besides, a lot of what people mean when they say they're bad at math is that they're bad at arithmetic, which is honestly understandable.
But that isn't what we mean with recursive function. We don't call this recursive:
Its just incrementing x.That's not a recursive function as it's written, but you could certainly consider it a form of symbolic recursion. This just isn't a very useful characterization in an iterative/imperative context. You could frame incrementing as recursive, though—this is just peano axioms/church encoding.
I've always wondered why FP isn't more popular. I concluded it's because most folks don't like thinking like abstract math.
I think the problem-solving part of coding requires math skills, while the organization part requires writing skills. The organization part affects the problem-solving part, because if you write messy code (that you can’t reread once you forget or extend without rewriting) you’ll quickly get overwhelmed.
Writing large math proofs also requires organization skills, since you’ll refer to earlier sections of your proof and may have to modify it when you encounter issues. But to me, math seems to have more “big steps”: sudden insights that can’t be derived from writing (“how did you discover this?”), and concepts that are intrinsically complicated so one can’t really explain them no matter how well they can write. Whereas programming has more “small steps”: even someone who’s not smart (but has grit) can write an impressive program, if they write one component at a time and there aren’t too many components that rely on each other.
Just like in math.
BTW relational DBs are math.
It's funny, reading the post you're replying to, I basically read it as
> I don't need math, I need <math, but by another name>
My teenage daughter used to complain about math, and I spent some time trying to explain to her that we use math every day... EVERY day. Now, when I see her do something that was math (even if it's not obvious it was math), I say "Math... every day". I say that a lot.
Also, yes, my daughter finds me annoying. But also funny; but likely not for the math thing.
https://news.ycombinator.com/item?id=43872687
So how much of programming is maths? Before we answer that, let's answer: How much of maths is actually maths? Because first we define maths, and then we define programming based on whatever that is, but until we have that first concrete definition this discussion cannot occur.
I will add that "it is taught by the maths department in college" is a flimsy argument, and frankly one the Physics department in particular would mock.
That means that definition shifts over time. For example, courses on numerical analysis, graph algorithms, programming, and on compilers used to be part of “what you’d learn in a math department”.
It likely also even today will show geographical variation.
I find this distinction useful in the abstract, that one can engage different parts of the brain for different components of development. This probably explains why a well-written DSL can be so powerful in the right context.
Firstly, computer science is math.
Secondly, I remember covering graphs in a discrete math course back when I was in college.
> What if you do it in SQL?
SQL is more-or-less a poor implementation of relational algebra. Ie, math.
But it's hardly a useful grouping any more. You can study and do well in computer science with minimal knowledge of most of the core mathematical subjects.
While graph theory certainly crosses over into math, you can cover most of the parts of it relevant to most computer science as a discussion of algorithms that would not be the natural way of dealing with them for most mathematicians.
You will fail at Theoretical Computer Science without mathematical proficiency. Go read some textbooks and papers in theoretical CS. It is a subfield of mathematics. Theorems and proofs. Rigorous and difficult mathematics.
https://en.wikipedia.org/wiki/Theoretical_computer_science
I wrote my MSc thesis on the use of statistical methods for reducing error rates for OCR, and most of the papers in my literature review hardly required more than basic arithmetic and algebra.
So I stand by my statement.
Sure, there are subsets of computer science where you need more maths, just like in any field there are sub fields where you will need to understand other subjects as well, but that does not alter what I claimed.
EDIT:
Some authors are quicker to pull out the maths than others, and frankly in a lot of CS papers maths is used to obscure lack of rigor rather than to provide it. E.g the problem I ran across when writing my thesis was that once you unpacked the limited math into code you'd often reveal unstated assumptions that were less than obvious if you just read their formulas.
I agree is not a useful grouping in practice. I'm just interested in what makes you think like you do.
What I claimed was that in computer science we often discuss things in terms that would not be the natural way of dealing with it in maths. We do that because our focus is different, and our abstractions are different.
It doesn't mean it's not math. It means it's not useful to insist that it isn't a different field, and its obtuse when people insist it's all the same.
Not "coding uses math", I mean it is math.
I don't know how you code, but I don't think I'm aware of code that can't be reasonably explained as forming relationships between objects. The face we can trace a program seems to necessitate this.[0] https://philosophy.stackexchange.com/questions/22440/what-di...
But I'll refer you to a longer conversation if it helps https://news.ycombinator.com/item?id=43872687
Coding used to be very close to pure math (many early computer science classes were taught in the Math Department in universities) but it has been so far abstracted from that to the point that it is its own thing and is as close to math as any other subject is.
since you can express paradoxes with match, perhaps not that different.
The contradiction is used in proof formulation, specifically to invalidate some claim. I don't think this is what you're implying.
The latter is what it contextually sounds like you're stating; things like the Banach-Tarksi Paradox. There's no self-contradiction in that, but it is an unexpected result and points to the need to refine certain things like the ZFC set theory.
I'd also stress that there are true statements which cannot be proven through axiomatic systems. The Halting Problem is an example of what Godel proved. But that's not contradictory, even if unexpected or frustrating.
> I mean equivalent in the way mathematicians do
That sounds like you're backing off from your original claim, probably because it is impossible to defend.
That you can use mathematics to describe code doesn't seem very different from using math to describe gravity, or the projected winner in an election, or how sound waves propagate.
Isn't the primary purpose of math to describe the world around us?
Then it shouldn't be surprising that it can also be used to describe programming.
In the real world, however, software engineering has nothing to do with mathematical abstractions 99% of the time
Though interpreting a CRUD app as a theorem (or collection of theorems) doesn’t result in an interesting theorem, and interpreting a typical theorem as a program… well, sometimes the result would be a useful program, but often it wouldn’t be.
It's not the type of thing that gets mathematicians excited, but from an engineering perspective, such theorems are great. You can often blindly code your way through things by just following the type signatures and having a vague sense of what you want to accomplish.
It's actually the halting problem that I find is not relevant to practical programming; in practice, CRUD apps are basically a trivial loop around a dispatcher into a bunch of simple functions operating on bounded data. The hard parts have been neatly tidied away into databases and operating systems (which for practical purposes, you can usually import as "axioms").
Physicists use math, but that doesn't mean it is math. It's not the only language at their disposal nor do they use all of math.
I'd argue that 100% of the time it has to do with mathematical abstractions. Please read the Poincare quote again. Take a moment to digest his meaning. Determine what an "object" means. What he means by "[content] is irrelevant" and why only form matters. I'll give you a lead: a class object isn't the only type of object in programming, nor is a type object. :)[0] Technically a specific (class of) physics, but the physics that any reasonable reader knows I'm referencing. But hey, I'll be a tad pedantic.
[1] String Theory is untestable, therefore doesn't really reflect the observable world. Even if all observable consequences could be explained through this theory it would still be indistinguishable from any other alternative theory which could do so. But we're getting too meta and this joke is rarely enjoyed outside mathematician and physicist communities.
Going on a total tangent, if you'll forgive me, and I ask purely as a curious outsider: do you think math could have ever come into being if it weren't to fill the human need of describing and categorizing the world?
What would have been the very beginning of math, the first human thought, or word or action, that could be called "math"? Are you able to picture this?
But my confusion is because I'm having a difficult time thinking where such things aren't also necessary consequences of just being a living being in general. I cannot think of a single creature that does not also have some world model, even if that model is very poor. My cat understands physics and math, even though her understandings are quite naive (also Wittgenstein[1] is quite wrong. I can understand my cat, even if not completely and even though she has a much harder time understanding me). More naive than say the Greeks, but they were also significantly more naive than your average math undergrad and I wouldn't say the Greeks "didn't do math".
It necessitates a threshold value and I'm not sure that this is useful framing. At least until we have a mutual understanding of what threshold we're concerned with. Frankly, we often place these contrived thresholds/barriers in continuous processes. They can be helpful but they also lead to a lot of confusion.
This too is hard to describe. Mull over the Poincare quote a bit. There's many thresholds we could pick from.I could say when the some of the Greeks got tired of arguing with people who were just pulling shit out of their asses, but that'd ignore many times other civilizations independently did the same.
I could say when the first conscious creature arose (I don't know when this was). It needed to understand itself (an object) and its relationship to others. Other creatures, other things, other... objects.
I could also say the first living creature. As I said above, even a bad world model has some understanding that there are objects and relationships between them.
I could also say it always was. But then we get into a "tree falls in a forest and no one is around to hear it" type of thing (also with the prior one). Acoustic vibrations is a fine definition, but so is "what one hears".
I'd more put the line closer to "Greeks" (and probably conscious). The reason for this is formalization, and I think this is a sufficient point where there's near universal agreement. In quotes because I'll accept any point in time that can qualify with the intended distinction, which is really hard to pin-point. I'm certainly not a historian nor remotely qualified to point to a reasonable time lol. But this also seems to be a point in history often referenced as being near "the birth" and frankly I'm more interested in other questions/topics than really getting to the bottom of this one. It also seems unprovable, and I'm okay with that. I'm not so certain it matters when that happened.
To clarify, I do not think life itself necessitates this type of formalization though. I'm unsure what conditions are necessary for this to happen (as an ML researcher I am concerned with this question though), but it does seem the be a natural consequence of a sufficient level of intelligence.
I'll put it this way, if we meet an alien creature I would be astonished if they did not have math. I have no reason to believe that their math would look remotely similar to ours, and I do think there would be difficulties in communicating, but if we both understand Poincare's meaning then it'll surely make that process easier.
Sorry, I know that was long and probably confusing. I just don't have a great answer. Certainly I don't know the answer either. So all I can give are some of my thoughts.
[0] GoblinSlayer ↗ >Even if all observable consequences could be explained through this theory it would still be indistinguishable from any other alternative theory which could do so.
That's not unique, all quantitative theories allow small modifications. Then you select parsimonious theory.
We didn't:
Mathematics is a very extensive field, and covers a vast amount of subjects.
For the same reason it can't be said that mathematics is equivalent to coding, as there are many things in mathematics that are not relevant for coding.
However, by far the most interesting parts of coding are definitely related to mathematics.
Math is about abstractions and relations. See the Poincare quote again.
Plus, the Programming Languages people would like to have a word with you. Two actually: Category Theory. But really, if you get them started they won't shut up. That's either a great time or a terrible time, but I think for most it is the latter.
Wheeler is arguing that if a tree falls in the forest, and there is nobody to hear it, that it still makes a sound because there are things that interact with the sound. But if the tree fell in a forest and there was nothing else in the universe then there is no sound because there is no observation.
It helps to read the whole thing[0] and to understand the context of the discussion. This is meta-physics and a deep discussion into what the nature of reality is. Ian Hacker has a good introduction to the subject but I find develop grave misunderstandings when they also do not have the strong math and physics background necessary to parse the words. Even people who understand the silliness of "The Secret" and that an observer need not be human often believe that this necessitates a multi-verse. A wildly convoluted solution to the problem of entropy not being invertible. Or closer to computer terms, a solution that insists that P = NP. There is information lost.
If you wanna argue that there's no difference between the word cup and a cup itself because there is no word without the observer who has the language, then yeah.
[0] https://historyofinformation.com/detail.php?id=5041
Programming is an expression of logic, which is absolutely mathematics.
But then we also have to think about naming variables and classes, structuring our code so that it is more readable by other developers, and so on. That's less about formal reasoning and more about communication.
There is an engineering aspect to programming (prototyping, architecture, optimization, etc). It's a combination of mathematics and engineering. Software Engineering.
Structuring sentences and naming variables so that it is easier for other people to understand is less about formal mathematical reasoning, and more about communication.
You could name a variable x, y, or Banana, but it doesn't change the logic.
I mean the reason we get mad at this is because it is someone destroying "society" in some sense. Even if that society is your team or just the set of programmers. It would be a pretty dick move were I to just use a word that significantly diverged from conventional meaning and expected you to mull it over. Similarly if I drop a unknown out of context math equation. It would be meaningless.
And I'm on your side, really! I strongly advocate for documenting. And let's be real, the conclusion of your argument more strongly argues for documentation than good variable names. Because variable names are much more constrained and much more easily misinterpreted considering how any word has multiple definitions. Surrounding code is often insufficient to derive necessary contextualization.
https://news.ycombinator.com/item?id=43874738
Mathematicians do have to deal with difficulties in naming things.
In general, lower case letters are used as constants, excluding x,y,z,t,i,j,k (sometimes u,v,w). It isn't a hard rule, but strong preference to begin at the beginning of the alphabet for these. Capital letters usually are held for things like Variable Sets (like random variables). Greek letters need context for constants or variables. BB and Cal typefaces for sets (e.g. Real Numbers, Integers). And much more.
I think a lot of the difficulty in it is that these "rules" or patterns are generally learned through usage and often not explicitly stated. But learning them can really help read unfamiliar topics and is why "notation abuse" leads to confusion. But after all, math is all about abstraction so technically any symbol will do, but no doubt some are (significantly) better than others for communicating.
and also philosophy.
The murky world of software patents would like a word about that.
For me, coding also feels artistic at times. But we see artistic beauty in mathematics all the time, so no wonder. But often I look at ugly code, and maybe that is subjective, but fixing that code makes it also feel prettier.
It is clear all developers have their owm artistic style, and probably that is why there is so much disagreement in the industry. Maybe we are lacking the pure mathemetical language to describe our intent in a more beutiful and precise way that is more clearly the-right-way. As in how we find beauty in simple physics equations.
> Not "coding uses math", I mean it is math.
Arguably writing a novel is math, if you use the right definition of math. But sometimes its more helpful to use more informal definitions that capture what people mean then what is technically accurate.
No, not always. Quite a lot of high-level code doesn't require any math at all. It doesn't take math to perform CRUD operations, which account for a lot of programming work. Sure, the underlying machine code is all based on math, but the higher level programming doesn't need to involve a single math equation for it to be useful. Let's see where the goalposts move now...
But the CRUD logic is so basic and boring, so obvious, that it doesn't require any thought.
Which code? The machine code that underlies everything? Or the lines of simple high-level CRUD that don't even need a single number or logic statement to function? Not all programming has to be mathematical or even logical at a high enough level, and I say this as someone who's been coding assembly language for 40 years.
Those lines ARE mathematical logical statements.
Each line defines logical operations that are executed by the computer.
Same for high level or low level programming. It's all logic.
Assigning a variable is a mathematical operation. It is a logical state transition. This is formalized in Programming Language Theory.
We are not talking about quarks. We are talking about lines of code which are objectively mathematical statements, no matter how non-mathematical they seem to you subjectively.
You start with a practical problem to solve, and a computing machine that can perform logical operations on data. Your job is to figure out a correct sequence of logical operations that will solve the problem. The program that you have written is a mathematical structure. It is mathematics.
https://en.wikipedia.org/wiki/Semantics_(computer_science)
In the case of code, we can argue that the map is the territory.
Those business requirements are inputs to the process of writing the code.
Once the code is actually written, that exists as a formal logical system, defined by mathematics, not business requirements.
>Once the code is actually written, that exists as a formal logical system, defined by mathematics
I still think that's not code, but your favorite model of code. For spellchecker language is defined by mathematics too: it splits text into words by whitespace, then for each word not found in dictionary it selects best matches and sorts them by relevance. Oh and characters are stored as numbers.
Writing code IS a math skill. When writing code you are writing logic in a formal system. Logic is mathematics.
You may be thinking that mathematics is just like doing arithmetic or solving equations. It is way deeper and broader than that.
> I still think that's not code, but your favourite model of code
Code is not just modelled through mathematics, it is actually defined by mathematics. It is fundamentally a mathematical construct, grounded in formal semantics.
https://en.wikipedia.org/wiki/Semantics_(computer_science)
Code is not modelled mathematically, it is defined mathematically.
It exists as an abstraction which is fully defined by operational semantics and denotational semantics, not modelled or approximated.
In the counter example of a quark, that exists in nature and is modelled by mathematics, but not defined by mathematics.
https://en.wikipedia.org/wiki/Formal_language
https://en.wikipedia.org/wiki/Semantics_(computer_science)
If you're going to accuse someone of confusing the map with the territory, you really should make sure you aren't making the same error.
Maybe what will help is seeing what some of the Programming Languages people do and why they like Haskell[1].
Or maybe check out Bartosz Milewski[2,3]. His blog[2], is titled "Bartosz Milewski's Programming Cafe: Category Theory, Haskell, Concurrency, C++". It may look very mathy, and you'd be right(!), but it is all about programming! Go check out his Category Theory Course[3], it is _for programmers_.
Don't trust me, go look at papers published in programming language conferences [4]. You'll find plenty of papers that are VERY mathy as well as plenty that are not. It really depends on the topic and what is the best language for the problems they're solving. But you'll certainly find some of the answers you're looking for.
Seriously, don't trust me, verify these things yourself. Google them. Ask an LLM. I don't know what to tell you because these are verifiable things (i.e. my claims are falsifiable!). The only thing you need to do is look.
[0] https://news.ycombinator.com/item?id=43882197
[1] https://excessivelyadequate.com/posts/isomorphisms.html
[2] https://bartoszmilewski.com/
[3] https://www.youtube.com/watch?v=I8LbkfSSR58&list=PLbgaMIhjbm...
[4] https://www.sigplan.org/Conferences/
Mentioning Scala is ironic, it's very light on math spik and to begin with was created to unify object oriented with functional programming, which is mathematically meaningless, because both are Turing complete and thus equivalent, tautological.
>So what I hear is "How programming languages helps with programming languages?"
Oh, right, in mathematics axiom is argument, but in reality it isn't. In programming you should assert what you assume, otherwise your assumptions can be wrong due to divergence from reality, but there no reality in mathematics, only fantasy, so you can't understand this with mathematics alone.
No. Code is an abstraction. It exists as a logical structure, grounded in mathematical logic.
https://news.ycombinator.com/item?id=43888917
In a typical implementation these are database operations. That involves relational algebra operations, state transitions, boolean logic.
The READ part can be a very complex SQL query (composed of many algebraic operations) but even the simplest query (SELECT * from t where ID = 1) is filtering a set based on a predicate. That is mathematics.
No one is moving goalposts. Set theory and logic are at the foundations of mathematics.
This is math:
{x | x.id = 1}
OTOH, a SQL query is a SQL query.
This thread is hilarious though. It's like
- Cashier: here is your change.
- Customer: you did math!
- Cashier, no, I gave you change.
- Customer: that IS math!
- Cashier: You mean, I used math to give you change?
- Customer: No, giving change doesn't use math, it IS math!!!!" [2]
= D
Moving along.. FWIW, indeed SQL was created to model set theory (relational algebra and tuple calculus), the close relationship is no accident of course [0][1])
> No one is moving goalposts
I feel too they are.
First goal post:
> Coding IS math. >> No, not always. Quite a lot of high-level code doesn't require any math at all. It doesn't take math to perform CRUD operations
Second goal post:
> CREATE, READ, UPDATE, DELETE are fundamentally mathematical in nature
CRUD is closely related to SQL, and SQL is closely related to various mathematics. Are they identical and therefore equivalent? No - because your database is not going to like it when you write "{x | x.id = 1}", and the Oracle DB might not like something that you can write for your Postgres DB.
[0] https://simpleprogrammer.com/mastering-sql/
[1] https://en.wikipedia.org/wiki/SQL
[2] To quote: """Not "coding uses math", I mean it is math""" @ https://news.ycombinator.com/item?id=43872771
Goes back to this ridiculous proposition:
- Cashier: You mean, I used math to give you change?
- Customer: No, giving change doesn't use math, it IS math!!!!" [2]
The proposition is that "code IS math", not defined by, not uses, not inspired by, not relies on, not modeled after, but IS.
Computer programs are proofs[1]. This is intuitively and also formally true. You would agree writing proofs is doing math, yeah? Then obviously writing a computer program is also doing math.
Like I have a degree in math and have been a software engineer for over a decade. I do not know what distinction people are trying to get at. It's like trying to argue about the distinction between U(1), the complex numbers with magnitude 1, and the unit circle, and getting upset when the mathematicians say "those are 3 names for the same thing". Or saying that writing C is programming but writing in a functional language like Scala or Haskell (or Lean) is not.
[0] Modulo details like NULLs and multi-set semantics, but surely that's not the distinction?
[1] Up to isomorphism
This is an interesting example: "Or saying that writing C is programming but writing in a functional language like Scala or Haskell (or Lean) is not."
The first part is everything we need to look at. Are we saying that writing C is equivalent and equal to the entirety of all programming? That if you're programming then you are writing C code. No, there is an implied "is a form of" in there. Given the other clarifications and that so many people are claiming to be mathematicians, I would have expected the precision to say exactly "C is a form of programming" rather than "C is programming."
Turns out, the analogy of saying "the set of reals is the set of naturals" is more fitting compared to sets that are actually equal.
In math are highly concerned with structures and abstraction. We have things called operators. They aren't just addition and multiplication. We also use those words to describe different operations. They have things like groups, rings, fields, and algebras. Yes, plural.
The purpose of these things is to create logical frameworks. It matters not what the operators are. Nor does it matter what objects we operate on. Poincaré is explicitly saying this.
The part you're not understanding is the abstraction. This is what math is about. It is also why the Programming Language people are deeper in the math side and love Category Theory (I have a few friends who's PL dissertations are more math heavy than some math dissertations I've seen). It's no surprise. What's a function? How do you abstract functions? How do you define structure? These are shared critical questions. PL people are more concerned with types but I wouldn't say that makes them any less of a mathematician than a set theorist.
We can perfectly describe the CRUD operations with set theory. Do most programmers need concern themselves with this? Absolutely not. But is it something people designing those operations and systems is thinking about? Yes.
I'd encourage you to learn some set theory, abstract algebra, and maybe a bit of cat theory. It'll make these things pretty clear. But I'd caution about having strong opinions on topics you aren't intimately familiar with. Especially when arguing with those that are. Frankly, CRUD is a terrible example. I'm confident a quick google search (or asking a GPT) would quickly point you to relational algebra. It's discussed in most database classes. It certainly was in the one I was an assistant for.
We are talking about textbook CS fundamentals.
https://en.wikipedia.org/wiki/Semantics_(computer_science)
As someone that is "on the other side of the fence", and getting flamed for it, maybe I can shed some light since I have the other perspective as well. IMO, The reason for not seeing eye to eye is (for example) akin to saying "word problems are math". (Thinking of a grade school word problems for common reference). Yes, they are readily mapped to mathematical models that can solve the word problem & perhaps almost indistinguishably so. Though no - word problems are not math. Word problems are a series of phrases and words. That's where the talking past each other comes in... Different interpretations of "word problems are math", or "code is math". It's seemingly not clear whether we are talking about logical 'implies', 'element of', or 'equals'.
Which goes to "We can perfectly describe the CRUD operations with set theory.", we all agree there. That is not readily conveyed though when writing things like "code is math".
Yes, I agree we're misunderstanding each other because we're using different definitions but "you've" rejected "mine" without defining "yours" and expecting everyone to understand. Of course that'll lead to confusion. You can reject the mathematicians definition of math, but you sure gotta say more than "trust me" and it's a pretty wild thing to do, especially as non mathematicians.
The problem here is one side who's dabbled in cooking says "a chef makes desserts" and chefs are responding "we do a lot more than that". Maybe there's a few chefs that go "yeah all I do is dessert" and everyone points to that while ignoring the second part of their sentence is "but that's just my speciality." Wouldn't you think that conversation is insane? The only reason it's obviously so is because we all know what a chef is and agree on the definition. But who is better qualified to define the chef's job? The chef or consumer?
In another thread, you characterized my response as stating: " ¬(A ↦ B) ⟹ ¬(B ↦ A)" (and this is a great example of language not being math, but math being language!). That was not at all my claim.
My claim is "I believe you are saying 'A = B'. It appears that 'B != A', therefore 'A != B'." My only claims are
(1) I believe you are writing to convey that you mean Math IS Language in the sense they are equal, identical, interchangeable, and fully equivalent, and bi-directionally so
(2) that: B != A
The only results can either be:
- "yeah, because B != A, the statement A = B is not true"
- Your claim (1) is false, I'm not actually saying "A = B"
- Your claim (2) is false, "B = A" is in fact true. I would find that to be an interesting assertion and would have loved to explore more why you think that.
That is a good analogy, except programming languages are formal languages, not natural languages.
> they are readily mapped to mathematical models
With code we are not talking about something that is mapped to mathematical models. Code is not modelled by mathematics, it is defined by mathematics (denotational semantics and operational semantics).
"Code is math" is true in a theoretical sense. Practically, coding doesn't always feel like what is commonly thought of as "doing mathematics", but it is essentially mathematical.
https://en.wikipedia.org/wiki/Formal_language
https://en.wikipedia.org/wiki/Semantics_(computer_science)
Actually I'm starting to wonder whether the thing that made university math easy for me was I quickly developed a good internal "type checker"/"type inference engine" in my head, and that helped make the next steps of proofs seem straightforward.
[0] https://gist.github.com/ndriscoll/881c4f5f0398039a3a74543450...
Though middle-school or not, it's still math.
For anything CRUD, you put the data into a database. The underlying database is relational, or a key->store.
If it's relational, that's one branch of mathematics. If it's another kind of database, it's another branch of mathematics. Mathematics is extensive, and covers more things you can imagine at a glance.
The main difference between writing mathematics and programming, and this applies to any form of programming, is that in mathematics writing a formal proof is amazingly close to: you write the program, and you are also the compiler, and you are the CPU, performing all operations yourself. With programming you only have to write the program.
Source: On one hand I have studied pure mathematics (not the simplified applied mathematics that are taught in engineering, which is mostly equations), on the other hand I have been working as a software developer for over 15 years.
This is vague and doesn't mean anything. People can't even agree what 'objects' are and people did a lot of programming before the term 'object' was invented.
Programming is about is fundamentally about instructions and data. Yin and yang of two things that are completely different, even if people can occasionally mix them and conflate them.
I'm surprised I hit a nerve with so many people. I'm quoting someone who's considered one of the greatest mathematicians. Obviously I don't know the backgrounds of people but it seems like programmers have strong opinions on what math is that disagrees with what mathematicians say they do.
https://en.wikipedia.org/wiki/Category_theory
At what point does the abstract description just not offer any useful insight anymore?
I'm surprised I hit a nerve with so many people.
A lot of people had very good explanations for why you're pretty far off in trying to say two things are the same.
Programming is much more like building a machine than any sort of straight math. There is state, there is interactivity, there is performance, there is IO and there are real world implications to all of it.
Saying they are the same is like saying gardening is plumbing just because you sometimes use a hose.
It's not a wild claim since you misquoted me.
A lot of non-mathematicians disagreed with mathematicians.
Mathematicians can claim whatever they want, when it comes to programming, programmers understand it better and they're trying to explain to you why this is nonsense. Vanderbilt claims to be "the harvard of the south" but wouldn't you know it, harvard doesn't claim to be "the vanderbilt of the north".
Show me programming languages designed by mathematicians to be 'mathematically pure' and I'll show you a language that hasn't been used to ship software that people want to use.
In other words: It's math in a sense that for most of us with a computer science background is often not very relevant to how we work.
... In my experience, learning to write one component at a time (and try the code, and make sure it works before proceeding) is itself a skill that many struggle to develop. Similarly for avoiding unnecessary dependencies between components. Oh, and also being able to analyze the problem and identify separable components.
One of the most frustrating things about teaching programming, for me, is the constant insistence from other teachers that you have to maintain an "absolutely everyone can learn to program" attitude at all times. Many people who start to learn programming have misguided or confused reasons for doing so and - to say the least - could make much more effective use of their time developing other skills. (It's not a question of elitism; I'd surely flounder at some tasks that others find natural.)
I very much think many people could learn the more advanced Excel Formulas, Power Automate and even simple Bash/PowerShell scripting to make their work more effective. I've met quite a few folks who had been intimidated out of trying who could do it.
On the other hand, how many people on this site could bootstrap a linux kernel on either very new or very old hardware? I know there are some, but they are certainly not the majority. I certainly won't be the first person to get linux and doom to run on a quantum computer.
But that is similar to other professions. Everyone with a largely functioning body can learn to turn a few planks and some metal parts into a functional shed door with some basic tools or to put up a decent brick wall that won't topple over in a month.
That doesn't mean everyone is qualified to pour concrete for a dam or a bridge foundation, or to re-do some historical work in original style.
It's shocking how little physical and spatial ability some people have - that is definitely not true. Sometimes it might be a personal discount or lack of confidence, but this remains true regardless of the cause.
> That doesn't mean everyone is qualified to pour concrete for a dam or a bridge foundation, or to re-do some historical work in original style.
Exactly!
I think statements like that are more concerned with philosophy than reality. Any discussion surrounding topics like this typically ends up being a discussion around definitions.
I believe the vast majority of human beings are capable of learning how to program in the most extreme elementary sense of the word. As in, outside of severe disabilities or complete and utter inaccessibility to circumstances in which one could learn program, then I think the remaining population of people could learn to program to some degree. Obviously, not everyone will learn to program due to a near infinite number of reasons.
I would argue it's like music. Anyone can make 'music.' Just make a sound -- any sound. The difference between noise and music is subjective. I would not argue that everyone could be the next Lovelace, Turning, Ritchie, Thompson, Torvalds, etc..
Now, for my jaded opinion, I think a lot of the "everyone can learn to program" talk does not come from a place of desire to share the gift of knowledge and joy of programming. I think it's more of a subtle way to encourage people to go into programming so that they may be hired by mega corps. In order to keep the Capitalist machine running. It's like the National Hockey League's slogan, "Hockey is for everyone." That is just a fancy way of saying, "everyone's money can be spent on the NHL."
I might be capable of learning advanced accounting, but that sounds like torture to me and I'll be damned if I'll ever take that on! I'm sure programming feels like that to a wide variety of people, and I don't see any need for us to try to pretend otherwise - outside of a bizarre ideological desire for equivalent outcomes from disparate groups.
I'm sure everyone is capable of learning some basic level of programming, just as they are able to learn a basic (high school) level of any subject. However, not everyone is going to have the aptitude to take that to an advanced professional level, however hard they try. We're not all cut out to be artists, or writers, or doctors, or scientists, or developers, etc.
Personally I've always considered a solid grasp of algebra to be the minimum bar for being able to program, at least for anything that isn't utterly trivial. Being able to take a word problem and turn it into a system of equations and solve it is a pretty close analog to being able to take some sort of problem or business requirement and turn it into code.
And the sad truth is that a huge percentage of the population struggle with just arithmetic, let alone algebra.
After collecting data for a few semesters he concluded his students could be clearly divided into three categories: those who just "got" programming, those who understood it after working hard, and a small group that just didn't grasp regardless of effort.
Computation is following an algorithm. e.g. long division or computing a derivative.
Intuition, AKA brilliance, is finding a non-obvious solution. Think "solving an NP problem without brute force"*. e.g. solving an integral (in a form that hasn't already been memorized) or discovering an interesting proof.
Organization is recording information in a way that a) is easy for you to recall later on (and get insights from) and b) is digestible by others**. e.g. explaining how to compute a derivative, solve an integral, or anything else.
Math, programming, and writing each require all skills. The kind of math taught in school (e.g. long division) and your boring jobs are primarily computation. I believe advanced math (e.g. calculus) is primarily intuition; it requires some organization because big theories are broken into smaller steps, but seems to mostly involve smart people "banging their head against the wall" to solve problems that are still quite unclear***. Programming is primarily organization. It requires some intuition (I think this is why some people seemingly can't learn to code), but in contrast to math, most programs can be broken into many relatively-simple features. IMO implementing all the features and interactions between them without creating a buggy, verbose, and unmaintainable codebase is programming's real challenge. Writing is also primarily organization, but finding interesting ideas requires intuition, and worldbuilding requires computation (even in fiction, there must be some coherence or people won't like your work).
> Some people are really good at mucking around garbage code (they have no choice, they get paid to), but what part of programming did they get good at? Obviously, some part of it, but nothing to write home about.
I agree that work you find boring should be avoided, and I also try to avoid working with it. But some people really seem to like working on esoteric code, and I think there are some skills (beyond computation) developed from it, that even apply when working with good code. Building a mental model of a spaghetti codebase involves organization, and if the codebase uses "genius hacks", intuition. Moreover, the same techniques to discern that two code segments in completely different locations are tightly coupled, may also discern that two seemingly-separate ideas have some connection, leading to an "intuitive" discovery. There's an MIT lecture somewhere that describes how a smart student found interesting work in a factory, and I think ended up optimizing the factory; the lesson was that you can gain some amount of knowledge and growth from pretty much any experience, and sometimes there's a lot of opportunity where you'd least expect it.
* Or maybe it is just brute force but people with this skill ("geniuses") do it very fast.
** These are kind of two separate skills but they're similar. Moreover, b) is more important because it's necessary for problems too large for one person to solve, and it implies a).
*** And whatever method solves these problems doesn't seem to be simplification, because many theories and proofs were initially written down very obtuse, then simplified later.
Usually even deciding what the problem is is in part an art, requires an act of narrativization, to shape and form concepts of origin, movement, and destination.
A good problem solver has a very wide range of abstract ideas and concepts and concrete tools they can use to model and explain problem, solution, & destination. Sometimes raw computational intellect can arrive at stunningly good proposals, can see brilliant paths through. But more often, my gut tells me it's about having a breadth of exposure, to different techniques and tools, and being someone who can both see a vast number of ways to tackle a situation, and being able to see tradeoffs in approaches, being able to weight long and short term impacts.
It has been reimplemented in a wide variety of languages today. I look forward to the next iteration: WolfgangSort or FatimaSort or XiaomiSort or whatever.
I absolutely value & have huge respect for the deeply computational works that advance us all along!
But this is an exceedingly rare event. Most development is more glue work than advancing computional fundamentals. Very very very little of the industry is paid to work on honing data structures so generally.
"Good code" is very subjective. Even readability and modularity can be taken too far.
> concepts that are intrinsically complicated,
I'm not a mathematician, but I figure mathematicians aim for clean, composable abstractions the same way programmers do. Something complicated, not just complex in its interactions with other things, seems more useful as a bespoke tool (e.g. in a proof) than as a general purpose object?
> Whereas programming has more “small steps”: even someone who’s not smart (but has grit) can write an impressive program, if they write one component at a time and there aren’t too many components that rely on each other.
This is well put. I often wonder if a merely average working memory might be a benefit to (or at least may place a lower bound on the output quality of) a programmer tasked with writing maintainable code. You cannot possibly deliver working spaghetti if you can't recall what you wrote three minutes ago.
This is a baldly self-serving hypothesis.
Forth programmers make a similar point. Forth is stack based; you typically use stack operations rather than local variables. This is ok when your 'words' (analogous to functions/procedures) have short and simple definitions, but code can quickly become unreadable if they don't. In this way, the language strongly nudges the programmer toward developing composable words with simple definitions.
(Of course, Forth sees little use today, and it hasn't won over the masses with its approach, but the broader point stands.)
Strongly disagree.
There are plenty of cases where non-modular code is more readable and faster than modular code (littered, presumably, with invocations of modular logic).
There are also countless cases, particularly in low-level languages or languages with manual memory management, where the best solution -- or the correct solution -- is far from readable.
Readability is anyways in the eye of the beholder (My code is readable. Isn't yours?) and takes a back seat to formal requirements.
Just as a matter of personal style, I prefer long, well-organized functions that are minimally modular. In my experience, context changes -- file boundaries, new function contexts, etc -- are the single greatest source of complexity and bugs, once one has shipped a piece of code and needs to maintain it or build on it. Modular code, by multiplying those, tends to obscure complexity and augment organizational problems by making them harder to reason about and fix. Longer functions certainty feel less readable initially but I'd wager they produce better, clearer mental models, leading to better solutions.
I'd say that readability, which often boils down to consistency, and modularity are ways to do this, but they aren't the only ways. And as you say, sometimes there's a need for "unreadable" code, so not everything can be easy.
I've always found the car metaphor to work very good to understand this: A car is a machine that can transport itself to point A to B (some other rules apply). There are different types of cars, but you certainly haven't understood the definition of you say that something is not a car because is not a Volvo, or because it doesn't look like a Ford, when it's clearly able to transport itself.
Math is the study of ideal objects and the way they behave or are related to each others. We have many branches of mathematics because people have invented so many objects and rules to play with them. Programming is nothing if not this very definition. The fact that you don't have to "use math" when programming is not really addressing the point, it's like saying a car is not a car because it has no discernible brand.
"Time is not a thing in math" is not understanding what math is. Time is another ideal object following certain rules under a given domain. Programming is coming up with objects of different size, with different characteristics, with interact at different points in time, i.e. following certain rules.
The study itself claims:
- fluid reasoning and working-memory capacity explained 34%
- language aptitude (17%)
- resting-state EEG power in beta and low-gamma bands (10%)
- numeracy (2%)
They take math skills to equal numeracy. The study itself implies this too. I disagree on a fundamental level with that. Math skills align much more closely to fluid reasoning than to numeracy.
In small peer groups (“pods”) that debug and learn together, communication becomes a core skill—and that can actually change how math skills are applied and developed. Language doesn’t just support learning; it reshapes the process.
This has the effect of making programming easier, but don't confuse it.
English majors.
It'd be interesting to see correlations (language brain vs math brain) for how easy or hard it is for people to solve new problems with language after they already know the basics.