If one wished to pursue this project with systematic diligence, the ontological questions would necessarily begin with, "What is a programming language?" and "How do we distinguish programming languages from natural languages?" -- particularly natural languages such as Wittgenstein describes as consisting only of "slab" and similar terms.
I fear that such an investigation would not get around to closures for a really long time if at all, and that OOP and PHP as objects of study would be so far in the future as to be archaic.
The ontological (and taxonomic) problem can be illustrated with a simple hello world written in Java running in the JVM running on a 2011 Macbook Air under OSX. What language is used to write the characters to the screen? A case can be made for JVM bytecode. A case can be made for C used to write the hardware's microcode. There's only a better case for one rather than the other when we have closed over a lot of assumed values.
The general problem of programming language is semiotic, not epistemic. They consist of symbols and meanings, and while these my glob together here and there - FORTRAN in any language is still possible.
Yes, that's the standard compusci response, and that's why debates in this area are presently fruitless. I dont really care to know to what extent PHP can be written in first order logic, nor does that help anyone decide between PHP and Javascript (via Node), for example.
With any philosophical project there are assumptions that you hold fixed and work that you assume "has been done" (or could be). Either way,
I think there is substantive and helpful contributions to be made by forgetting where programming languges come from, by ignoring their syntax and by treating their standard implementations as worlds-to-be-explored.
And if you like, the above statement is my starting point, and implicit rejection of the compusci approach which is to ask all the questions im ignoring.
That's a philosophical response, for that is my background. On the other hand, my general reaction is that the essay lacks rigor. Kant's Prolegama to Any Future Metaphysics may not set an obtainable standard for this sort of thing, but it may prove to be a useful guide.
Perhaps we are always required to make assumptions in order to act philosophically - even if I am inclined toward the belief that though we start somewhere, starting to do philosophy comes earlier in the day than any decision to assume. In other words, we don't assume our starting point or arrive at it after packing our bag - once we start packing our bag, we are well into the act.
I think the study of computer languages can tell us interesting things of the sort which philosophers - and particularly natural language philosophers - have found interesting over the more recent past. But I find a great-chain-of-being approach dubious in regard to generating epistemic epiphany.
Oh it certainly lacks rigour. I was just outlining "an intention to be rigorous" to see what the response would be. There seems to be, so far, quite a lot of interest in this kind of project.
But alas the taxonomy of programming languages is a solved problem. There is Lisp. And there is Blub.
Or less t-rolled: McCarthy and Von Neuman.
Or more poetic: Dionysian and Saturnial.
Rigor comes in many flavors. The starting point for all of them is rigorous thinking. One of the signs that my thinking is not rigorous is that I wind up proving the point I set out to prove. Then again, setting out to prove a point ought to be a warning flag.
Even if it is inevitable that I do so. The key to rigor is recognizing that I am doing it. Lisp and Blub was where I wanted to go. My dissatisfaction led me to McCarthy and Von Neuman and that to another metaphorical connection between programming languages and literary criticism. And that fits with my larger interests; those being the ones which dispose me toward writing about software.
As I said towards the end of the piece the intention is to come up with an abductive process an, "inference to the best representation", a formalized method for deciding between languages. I am aware that there are various taxonomies, and perhaps I shouldnt have made the aristotleian comment because it was rather misleading.
My intention was to take an honest (you may say phenomenological) look at programming languages: to have a system by which the idioms of the language, its communal metalogics, etc. are all part of how we assess a language. Not the reductionism of the "procedural-functional", pre-existing landscape.
I guess my point in referencing this landscape was to point out how inadequate it is, and how there might be a more complex and useful way of reasoning between languages.
My next article on the subject, which I'd imagine will follow soon, will certainly be slightly refocused and clearer. This is very much a just-finished-typing first draft that I put out to see what the response would be, and what the appetite for this kind of analysis in the programming community is.
> We are currently in a state of paralysis with respect to educating decision-makers within applied software development because the arguments made for one language or another are small incomparable islands that admit only buzzwords.
In my experience, decision makers don't want to be educated about the benfits of your favorite language. They want to use a 'full stack' supported solution that has corporate backup and industry acceptance. They want to be able to hire someone to take over your code when you leave.
In the end, you can compute anything compuatble with any Turning complete programming language, although some are more elegant than others at certain problems. Human psychology will lead you to desire what you don't have, which in this case is your ideal niche language for solving your current problem at your shitty job. In the end languages are just tools and decision makers are providing you with the tools you are to use to do the job which minimizes their risk and maximizes their return.
And if your language is so great and will solve all your problems in an elegant and productive way, then make something with it and start a successful company. Then, if you make it for 20 years, wait for the articles saying how 'uneducated' you are having used it and you should be using some university ressearch language based on n-th level hyper-metalogics because you can write fib in 21 symbols.
>In the end, you can compute anything compuatble with any Turning complete programming language
Well it's this defeatist position i'm arguing against.
> you should be using some university ressearch language
I'm largely a PHP developer / "architect" by trade; I'm not in the business of pushing Haskell onto CMS developers.
I am interested in the programming language debate, and I'm not convinced "the market" is the best judge of what the most useful language is. That it happened to be, for example, the internet was coming of age during PHP's invention which gave us a generation of script-kiddes (98 - 06) seems to me an accident of history. This accident is being corrected by Zend Framework and corresponding language changes (06 - 13), so by today PHP is a language which resembles a pythonized java, rather than a javascripty perl.
The market in this instance is doing precisely what has been recomended by the leading theoretical voices in this area: coming from Zend, Sympfony, etc.
My intended contribution is to provide a conceptual framework in which to make sense of these changes, to advocate for these changes, and to compare potential changes. Mailing lists are abound with favorite features, I've yet to see a feature suggestion with a complete argument of the kind I would advocate.
Though some of the very best do actually get 95% of the way there (ontology: showing the new feature in other languages to "get a sense of what it is"; epistemology: showing how to reason about it in different areas; abduction: showing how it represents problems better than current technologies): if you look at a modern high quality PHP-RFC today you're getting very close to what i'm proposing.
The open source market of merit is putting these individuals in positions of power over the evolution of languages, it would be useful - i suggest - if we could have a standard which provides the means for establishing good arguments from bad arguments in this area.
The market will never be the best judge of what the most useful language is, but it is a correct judge. 'Best', being a relative term, takes an idividual's ideals into account, 'correct' being what is actual in practice.
The Best is chosen by the market that which is used to create the most successful product. By that measure, PHP is the best because Facebook is written in it. Best to the individual is much more complicated. The two will rarely coincide due to the uniqueness of the individual and the typical rejection of authority by intelligencia.
I would never say that PHP is the 'best' language for doing high abstraction, but it can be done. I also would not use it for such things, but you said you have (if you are the author of the article).
The 'Best' language is what you do your best work in, whether paid or not. Just don't expect to use it at work, and if you could, you would think something else would is best and pine for that.
facebook is not written in PHP, it's written in a PHP-like language developed by them which is then transcompiled into C++ and then to binary.
> It's the job, not the tools.
Again, this is what i disagree with. There really is something objective, independent, and "true" to be said about the utility of various languages in various problem domains. There really is a language which represents the problem space most accurately, for example.
It's just defeatism to suppose that "it's all the same, whatever works". Innovation is, of course, the opposite drive. It is at the innovative egdes of programming languages that these debates are had. If you arent part of this innovative edge then it may seem to you that the things youre using are just "market decisions", they are far from it. Every programming language you are using has not been created by the market, it has been shaped by the very debates i am talking about.
The market cannot create languages, it does not improve them, the market merely plays out a baggage'd history which language-theorists have to cope with. This is why PHP has changed so much: it is the force of the people who engage in this discussion on the market, to drive it towards "better".
It's true there are awesome specific language tools for your specific problems, that when tied together to create a total solution creates an unmaintainable mess that even the creator doesn't understand.
Look at the web.
I saw Kerrnigan talk about 'little languages' and I fully agree that a specific language designed to solve a problem is the way to go for small problems, like the ones when Unix was designed. Tying all these text based interfaces together is a maintenace nightmare.
Maintenenace is a much bigger problem than the actual problems being solved, at least for businesses.
Then there's Lisp, which is a mini-language designer's tool kit and the super glue to tie them together all in one integrated package using a well designed, common binding agent. Unfortunately the market has choosen not in it's favour.
The market will improve languages and in end game is that they'll all have the same features, and then it comes down to syntax and performance. They'll all be 'better' by today's standards. The problem is at that point is that the better bar will have moved.
And the guy that truly innovated during that time won't be found until the next century.
It seems to me a philosophy of code that begins at the "low" level of binary and machine instructions, would ultimately reduce to a contemporary British/American Analytic style of philosophy to the exclusion of all else, seeing as how the core level of communication is binary -- logic, to the exclusion of rhetoric, in the classical sense of both of those disciplines.
If you take a logic class at a basic level in both a Philosophy and CS context (as I did), you'll be struck by the massive overlap -- DeMorgan's theorem, modus ponens, et. al.
Setting aside for the moment the possibility of alternate CS architectures (neural nets, quantum computing and so forth), this would seem at first blush to limit the relationship of all computer languages to a 1:1 relationship with logic.
However this "reductionist" (foundationalist?) approach to finding a true semiotics of CS may not tell the whole story. As we move from the "low" level to the "high", say from the domain of machine code all the way up to, for instance, a rich, expressive Ruby DSL, the question arises, as it does for all language: have we acquired a surplus of content that cannot be simply reduced to core rules? At this point things get less Wittgensteinian and a little more late-Heideggerian, i.e., less analytic and more eidetic/phenomenological.
To remove the fuzziness from this notion, think of how, for instance, a first-person shooter creates the illusion of a smooth, analog reality from a rigid, frame-by-frame progression (as cinema does as well -- and this question has been explored at length in that world). As humans interacting with the result of the code, we infer content between the frames -- we make up the difference and experience the virtual world presented to our senses as continuous analogical rather than chopped-up and merely logical. Code never exists in a vacuum -- it is experienced by humans and humans bring meaning to it, as much as it brings meaning to them.
There's a similar phenomena in OOP, where the meaning of the code is not merely the sum of its final binary product ("true" or "false") in the world of machine instructions, but in the metaphor it presents as a model of an object or event in the world. This code is constantly changing (literally) as it intends towards richer analogies with the world, and we as coders (and users) make up the difference by bringing real-world experience to it when we "read" the code. In other words, code is never "true" or "false" as a function of its binary reduction, but is a constantly-changing artifact of some human being's attempt to analogize the world, read by other human beings attempting to understand and refine that analogy.
In this sense the general problem is, as another poster phrased it so well, semiotic rather than epistemic or even logical. But I'm not sure a "closing over" or reducing towards core CS mathematics helps us understand the ontological status of code, any more than Wittgenstein or the Vienna School has given us a salutary reform of philosophical discourse with pure analytical techniques in search of a perfect, ambiguity-free symbolic language.
Language of any kind acquires a surplus of meaning as human beings interact with it and bring their own analogies to the experience, and I think this applies to code as well. So it may be that it's OK to move towards a philosophy of code that takes a phenomenological approach to the final product -- the code itself -- rather than (or probably in addition to) looking for meaning in the binary rules that compose machine instructions.
OP writes:
>> With this in place we can make certain classifications (in a quais-aristotelian manner): functio...
I agree 100%. I think however that there are two projects: the "phenomenalist" and the "instrumentalist". In the former part it's a group of polygot philosopher-developers waxing philosophic about the nature of programs and in the latter it's "briefs for lead developers".
Whilst my inclination in being a Husserlian first-and-foremost is to engage in the former, I think the latter is definitely of interest.
The third choice, and this is i think what you mistake my current project to be, is this "binary reductionism" or "symbolic reductionism" where DSLs are either 0/1s or predictate-object statements. I disagree with this entirely, and i think it leads to the current "chose what you like" defeatism.
I would like to explore the middle ground for awhile, an "early heidegger" if you like. I think that's going to get a lot more traction in the implementation community. There is definitely a lot of space at this moment in time for some real substantive work in all non-reductionist areas.
NB. i think my use of "metalogic" has confused at least a couple of people into thinking im talking about something CSy. Logic in the sense a continental philosopher would use it: a mechanism of reasoning. As I say in my "notes on terminology" i mean to include things like the domain/business logic. I think that, for example, that "a car shouldnt start with no fuel" is part of the program (the metalogic of the program, not the logic itself: it is part of what we need to consider to fully understand the progran). Perhaps i shouldve used "meta-rationale" or something.
Excellent! Agreed completely. Steve mentioned state machines off-hand on the list a while back, and that stayed with me, as someone with a disposition towards Husserl and Bergson and Whitehead and a general Process approach -- that state machines, in the broadest sense, might provide a practical, code-level starting point to focus attention on the "flow" of objects, interfaces that adapt to context and communicate that context to one another. This, combined with more focus on a functional approach and reducing state to manage concurrent issues (where the most practical "ontological" concerns seem to erupt!), might move a "philosophy of code" towards something practical -- if nothing else with a deeper awareness of the general "incompleteness" of all languages and their necessary analogical nature. This at least might mitigate "language wars" and the search for the "one best way" of coding.
EDIT: A minor quibble: maybe "phenomenalist" isn't the best term to use in close proximity to Husserl. He used the idea of "phenomenalism" in contradistinction to "phenomenology" -- the former being a kind of reductionist approach to seeing essences as merely the "contents of consciousness", and the latter -- which is to say the general Husserlian approach -- sees essences, or "eidetic objects" as phenomena that intend towards the external world, at least external to subjectivity.
> As we move from the "low" level to the "high", say from the domain of machine code all the way up to, for instance, a rich, expressive Ruby DSL, the question arises, as it does for all language: have we acquired a surplus of content that cannot be simply reduced to core rules?
No, not if it's still executing on a physical computer.
The surplus he's talking about is in the structure of the program: why is it that way rather than any other? Because of some meaning attached to the program, some "metalogic" (my terminology): a logic about the program which is not formally in the program but which nevertheless has significantly affected it.
I would still say no, because such metalogic will still be reducable to a type elevated logic and thus be reducable to 'core rules' operating on a lesser logic. Computability dictates what a computer can be instructed to do, leaving the 'meaning' to also have to conform to what can be computed.
If it runs on a computer it can be reduced to a simple set of core rules that is a logic, no matter how complex, by virtue that it is running on a computer.
If the "it" in question is considered merely as the program and sum of CPU operations that the bytecode or binary-whatever reduces to, of course what you say is true, and also completely uninteresting. All that we've stated in this case is that software is software, electrons are electrons and bytecode is bytecode. Computers compute, end of story.
But of course that's not what we do as coders. The code is never "complete" -- it's always striving towards some "meaning" that is for a human being. You cannot exclude the meaning of the code from the subject that it is intended for. A human being perceiving the code is always redacting the meaning of the code, not as a function of ones and zeros, but as a function of the analogy that code creates, pointing not towards electrons but towards: a product in a warehouse; a sum of sales at the end of a quarter; a flaming ball of death hurling towards my first-person-virtual head. And so on.
In this sense the electrons/ones and zeroes are no more significant than the paper that The Iliad is written on + the ink it is written with.
Then what does this specifically have to do with programs?
All works of substance share such an 'incompleteness' and meaning. We just look to non-programming works as complete in a sense when they are recognized. But such works are constantly being rewritten in different forms in order to produce the effect they are known to be able to produce.
In the end, everything is a program in some sense whether to be executed by a brain or computer.
The point of what I was saying was to refute the idea that somehow our programs are so complex and have such amazing behaviour that they will not be reducable to a core set of rules. If the programs run on a computer they must be computable and thus reducable to a core set of rules. That's it.
Whether we can be reduced to that core set of rules is not evident yet. Using logic, a statement that denies it's own provability in the language it was writen in was crewated. Logic was not thought to be able to lie, until it did. Who knows what the right brain can make it do?
reduction implies elimination of something. the argument was never that a program could become incomputable, but that it is much more than merely what happens when it is run
> As we move from the "low" level to the "high", say from the domain of machine code all the way up to, for instance, a rich, expressive Ruby DSL, the question arises, as it does for all language: have we acquired a surplus of content that cannot be simply reduced to core rules?
This was the statement that I started all this with. I am taking "surplus of content" to mean "a bunch of programs" in this context due to the quote speaking of a Ruby DSL. Programs can be reduced to a sequence of logical operations if they are computable, which I would take as 'core rules'.
Therefore, only non-computable programs can not be reduced to a set of core rules by the definition of computability. QED
Maybe you can explain 'it is much more than merely what happens when it runs'? Do you mean that what it took to create the program was greater than the resulting program and it's effects caused by running it? That the meta-logic used to create the program is far greater than the logic in the program itself?
I am 'much more' than the results of actions when I act. I have a past history that has created me, a physical body being used to act with, the world I am acting in, and the resulting effects of my actions, plus everything else I have missed. I that what you mean by 'much more'?
>> I am 'much more' than the results of actions when I act. I have a past history that has created me, a physical body being used to act with, the world I am acting in, and the resulting effects of my actions, plus everything else I have missed. I that what you mean by 'much more'?
tl;dr -- Yes!
The non-tl;dr:
By "surplus of content" I meant to say that the rich, english-like syntax of something as high level as a Ruby DSL acquires rhetorical, intuitive meaning that cannot be reduced to a 1:1 symbolic relationship to the compiled instructions. But this "surplus" is supplied by the perceiving subject, in the same way that natural language itself is interpreted and redacted by the human being reading a particular work.
I didn't mean to say that mathematical magic had happened as a result of complexity.
I'm coming at this from a philosophical as well as a CS background (I studied both, but ended up in software). In philosophy, there is an ancient tension between rhetoric and logic. Logic is a brute force process -- a set of rules where the same input always produces the same output, just as (and, it turns out, according to the same rules) the same electrical series of pulses supplied to logic gates and pushed lockstep through them by the CPU clock will always produce the same output. Many analytic philosophical papers don't look all that different from a Karnaugh map describing the structure of a multiplexer, and the results of both are equally deterministic. This school of thought was galvanized by and built heavily upon the Principia we spoke of earlier.
But there's another school of thought in philosophy that uses rhetoric and poetic devices as well as logic to achieve its ends. Language in this view (From Aristotle's Rhetoric to Heidegger's Building, Dwelling, Thinking) is decidedly not deterministic, but provocative, emotive and intuitive. Human beings communicating with each other with language use analogy and a commonly shared narrative history to provoke meaning to each other, rather than reduce statements to mere facts that are reduced to true or false. This school is commonly referred to as the "Continental" school, due to its prominence in France and Germany. Analytical approaches dominate British and American universities.
So, my argument here is a function of a larger question about language itself. I'm saying that the English-like analogical techniques even at the lower levels are provocative and provide more meaning than can be reduced to the underlying logic. This is not to say that any software is less than deterministic -- of course it cannot be non-deterministic because the core functions are purely mechanistic and deterministic. So, an address changed in a DB is always changed in the same way, and a joystick moved in such a way to provide a stream of x,y coordinates to a controller will always move a character in the same way, etc.
But the meaning on the other side of those events is supplied by a subject. At an extreme level, think of a high-resolution bitmap of Rembrandt's Philosopher in Meditation. The presentation of the pixels is deterministic, but the meaning is supplied by the subject. I'm saying that to a greater or lesser degree this redaction, the interpretive gesture that supplies a surplus of meaning is taking place at basically every level of computation. To a lesser degree at the low levels ("push", and so forth), but to higher degrees the closer to English-like (3G languages, etc.) we get. Images and literature presented in UTF-8 are just radical examples on one side of the spectrum, of how deterministic CPU instructions can result in this "surplus".
I actually don't disagree with any of this, as far as I understand what you're saying.
>> The point of what I was saying was to refute the idea that somehow our programs are so complex and have such amazing behaviour that they will not be reducable to a core set of rules. If the programs run on a computer they must be computable and thus reducable to a core set of rules. That's it.
Considering a piece of software as a mathematical construction -- let's say, for instance, a C++ program compiled by the LLVM, and the LLVM bytecode expressed as machine instructions for the given chip, reduced to pure binary math, I totally agree with this.
But the computer and its physical state has no meaning unless the output is perceived (paging Berkeley). That meaning to the user or coder is more than a function of the position of electrons in silicon, but also, and I would argue primarily, in the meaning conveyed to the subject/user/coder. Certainly there is no meaning without physicality, but physicality without subjectivity is literally meaningless, and that subjectivity brings with it analogies and more content than can possibly be found in a pure reduction to electrons.
I think if we look hard at the structure of CPU instruction sets, GPUs, math coprocessors and the like, we'll see even at the lowest level attempts to present analogies and manage the limits of pure logic as presented to a human consciousness. So even at the lowest level of CPUs, the language of analogy is present, and that implies that a human subject is going to supply the content "between the electrons" as it were.
You are taking the system of computer and user as a single system, and are no longer talking about just the computer. This will lead to more than electronic reduction in silicon because there is more than electrons and silicon.
There are no attempts to manage the limits of pure logic at the silicon level. The entire analogy of the CPU is to implement pure boolean logic through strict controls and limits of voltage levels in analog circuits.
The logic available to humans is the same available to CPU. Godel used pure logic to create an unprovable statement that was both true and false. What he did to create such a statement is what computers do not (yet) have, but it most certainly is not logic, and I think that would be what you say is 'between the electrons'.
But whether we can program and execute this 'insight' on a finite logic system is another question that is yet to be answered. What we can program we can execute and thus can be reduced to basic forms of symbolic logic, whether powered by electrons, pen and paper or the brain.
>> The entire analogy of the CPU is to implement pure boolean logic through strict controls and limits of voltage levels in analog circuits.
But this isn't the whole story either, right? What is "pure boolean logic" anyway? It doesn't exist even at the lowest levels. Logic is always applied to a statement of some kind, it doesn't exist as a pure abstraction. It's interesting you bring up Godel. Russell and Whitehead attempted to map logic to math in the Principia, and were refuted by Godel's incompleteness. I don't want to make too much of incompleteness (as many do), but let's not make too little of it either. At some point logic can only do so much for you, and you have to resort to analogy. I think this happens much earlier in CPU architecture than you're allowing for.
What you have at a low level on the CPU are a set of instructions arbitrarily (to some degree) defined to efficiently model analogical actions like "push" and "move" and "shift" and the fundamental operations that map to assembly. And here we've already moved out of the realm of pure logic -> math envisioned by Russell and we've created logic -> language, by way of analogy. This is, I would argue, a rhetorical move rather than a logical one. We've already prepared/redacted the ones and zeroes for human consumption, and assumed a specific semantic to provide for a human imagination that will accept and adapt to the semantic.
The instructions in the CPU are no different from the statements in the Principia, just finite and physically realized. The instructions are combinations of logic operations, although the logic is now sequential and having state, which I would think sets it apart from the Principia since I haven't yet gotten my copy to study. The sequentiality gives rise to monadic world state, but all is still reducable to logic in the end.
There's no magic in the computer and no analogy other than designing analog circuits that clamp voltage levels and a very large finite state machine. It's logic all the way down, limits and all, though instead of incompleteness there's incomputability. 2 sides of the same coin.
Pure boolean logic can be reduced to symbol manipulation using rules of substituion and inferrence, but, of course, what is that composed of? Godel showed that metalogic is made of logic, you just elevate the types.
What is logic made of?
BTW I know computer architecture very well. There's no analogies, just abstraction.
Just as a preface -- this is a great conversation, very provocative to me. If I make a misstep with regards to to CPU arch, feel free to correct (naturally).
"The instructions in the CPU are no different from the statements in the Principia"
As the goal of the PM was to show that math was a function of logic, we can use 'statements in the Principia' and 'the classical rules of logic' interchangeably in this context.
There's two questions at play in this conversation. One is whether in principle the question of the meaning, as such, of software, can be considered apart from the whole act of perceiving the software, either as code or as the final UX.
As the idea of "meaning" always involves some kind of communication, a sign or set of signs shared between two subjects, and as the idea of software pointing back to its own internal state as the content of this communication would render the communication pointless and the signs merely tautological, I think the question of meaning must bring the subject, and the act of perception and interpretation, into consideration. Scientific statements of fact, regarding the physical position of all of the machine's atoms and electrons, have content, but not "meaning" in the common sense of that term intended in the current context.
The second question, which I'm pretty much winging here (although I have long-held opinions about the PM and its relationship to the same kinds of considerations about "meaning"), is whether CPU instruction sets, as subsets of the rules of logic, dictated by the fundamentals of CS (Babbage, Turing, et al), are "pure" expressions of logic, or whether in the act of abstracting logic into the restricted set of instructions, we've performed an analogy.
I think the same question comes to bear -- it matters which perspective you're taking, whether "meaning" is merely a construction that hides the more essential question of physicality, or whether ideas like "push" applied to things like registers and addresses and binary values are linguistic in nature and convey more than just tautological facts about the physical state of CPU internals.
I guess this pushes the question out -- why does a computer exist? To represent to the world the bare facts of its own internal state? Of course not. A 64-bit address that points to a sequence of binary values is never just that, it's always a sign that points to something else -- a person's address; a representation of an English character; a shipment of goods; a lat/long location of some specific place in the world.
So, a "move" or "shift", to you, may reduce backwards to a collection of logical operations. But even there, what has happened? Nothing's "moved" or "pushed", that's just an linguistic trick to help us think of the electrical currents being turned "off" and "on" (and even those are analogies) in mentally-manageable concepts that map to the rules of logic. To me the question of "meaning" has already been broached at this point, if we wish to see computations as significant beyond, again, tautological statements about CPU internals. If we take the latter point of view, we have mere mathematical statements that have the same point as pure math -- the elegance (if that) of its own proof, not any relationship to something external to itself. But of course, that was never the real point of computers from the very beginning.
The Godel/Russell/Whitehead question is apt here, because this is a very similar question, if not the same question. Is math a function of logic -- that is to say, can math be "true"? The point of the Principia was to bring about the dream of the Vienna Circle, a purely mathematical and systematic technical language to resolve all philosophical ambiguity once and for all. While there remain a few holdouts for that dream, the general consensus is that this did not work out -- Godel was only one of many problems with Positivism specifically, and I would argue, Analytical Philosophy in general, although in recent years the Analytical/Continental divide has b...
1. There are symbols and nonsymbols.
2. Symbols can be equality-compared (eq'd).
3. Nonsymbols can be observed for behavior (in symbols returned) but not eq'd.
The set of symbols is countable and includes:
1. Natural numbers.
2. Unique symbols for arbitrary "dictionary words", e.g. _error.
Nonsymbols represent mappings that take nonsymbols or symbols and input and return nonsymbols or symbols. That's not well-defined (and it's not even a set) but we are only concerned with those that return symbols (i.e. can be observed) after certain input patterns. For example, A : {_ => A} and B : {_ => B} are both nonterminating but behave identically based on what we can actually observe; it's meaningless to argue about whether they are "the same" and since nonsymbols can't be eq'd, also unanswerable.
Code (as in source) is essentially a process that takes a natural number (viewed as an array of bounded natural numbers) and converts it into something that behaves as a nonsymbol, at least over inputs we care about. Of course, there are many nonsymbols that we can't represent in code, such as one corresponding to the Halting Problem,
{$sym => 0, $nsym => 0 if exists k s.t. apply^k($nsym, 0) is a symbol,
1 otherwise.}
Maps, vectors, and sets give us tools for describing extremely well-behaved (finite support; returning _error on the not-found case) nonsymbols in code; functions give us a language to describe well-behaved infinite-support nonsymbols.
The neat thing about traditional functional programming is that it carries an implicit will to focus on well-behaved nonsymbols. Object-oriented programming hasn't stepped away from the full space of nonsymbolness-- and that's a big part of why it's so damn hard to reason about it.
I like this as a more systematic starting point, with more direct application than the high-level conceptual approach I gravitate towards.
One quibble -- I would prefer non-negative(!) terminology, i.e., what is a "nonsymbol"? A collection? A monad? A duration? An event? A process? A closure? Some category that encompasses all of these? If so, can we find a positive semantic rather than negative?
>> Object-oriented programming hasn't stepped away from the full space of nonsymbolness-- and that's a big part of why it's so damn hard to reason about it.
Well, this gets back to the ancient problem of "The Third Man" and categories in general, right? One must categorize and "contain" meaning, trap it in language a moment in order for it to remain meaningful, but paradoxically the external thing that it symbolizes is always changing, so it's always escaping any semantic trap we set for it. This of course has practical (concurrent) implications in software as well as the high-falutin metaphysical stuff Plato was talking about.
These days I'm starting to wonder more about how we may not be using state-machines quite as effectively, or as often, as we might -- changing functionality more aggressively as the data changes, in order to move towards a better analogy with the ever-changing "real world" that software is trying to model. Might this be more self-aware starting place to integrate functional and object concerns -- both of which are necessary?
One quibble -- I would prefer non-negative(!) terminology, i.e., what is a "nonsymbol"? A collection? A monad? A duration? An event? A process? A closure? Some category that encompasses all of these? If so, can we find a positive semantic rather than negative?
Good question. I tend to think of it as a stateless (I haven't added state to the model yet, and there are myriad ways of doing it so I'm debating how to do it) object. To apply it to something (symbol or nonsymbol) is like sending a message, and the result of the apply is what it returns.
Nonsymbol is deliberately vague, meaning "we don't know what this is". We need a context to make sense of it (e.g. observe _size, then if that's a natural k, observe 0, ..., k-1; it it's not, it's not in the Vector context) and if it doesn't play by the rules of a known context, we're screwed. We continue to need context when we "array-ize" nonsymbols into natural numbers.
Nonsymbols are like functions; however, a function has a well-defined domain and non-symbols don't. However, the only useful nonsymbols are "function-like" in that some sequence of symbols and known nonsymbols will produce an observable (symbol).
They're very similar. I would say the only difference is that I intend on using function for a specific subset of nonsymbols, and also that I intend for it to be a code construct (nonsymbols are not code; in fact, it's impossible to capture most nonsymbols in code). Pathological nonsymbols do not really feel like functions. However, one could construe nonsymbols as being HOFs (the worst of which would be ill-typed in any sane type system, such as Haskell's).
35 comments
[ 9.4 ms ] story [ 90.9 ms ] threadI fear that such an investigation would not get around to closures for a really long time if at all, and that OOP and PHP as objects of study would be so far in the future as to be archaic.
The ontological (and taxonomic) problem can be illustrated with a simple hello world written in Java running in the JVM running on a 2011 Macbook Air under OSX. What language is used to write the characters to the screen? A case can be made for JVM bytecode. A case can be made for C used to write the hardware's microcode. There's only a better case for one rather than the other when we have closed over a lot of assumed values.
The general problem of programming language is semiotic, not epistemic. They consist of symbols and meanings, and while these my glob together here and there - FORTRAN in any language is still possible.
With any philosophical project there are assumptions that you hold fixed and work that you assume "has been done" (or could be). Either way,
I think there is substantive and helpful contributions to be made by forgetting where programming languges come from, by ignoring their syntax and by treating their standard implementations as worlds-to-be-explored.
And if you like, the above statement is my starting point, and implicit rejection of the compusci approach which is to ask all the questions im ignoring.
Perhaps we are always required to make assumptions in order to act philosophically - even if I am inclined toward the belief that though we start somewhere, starting to do philosophy comes earlier in the day than any decision to assume. In other words, we don't assume our starting point or arrive at it after packing our bag - once we start packing our bag, we are well into the act.
I think the study of computer languages can tell us interesting things of the sort which philosophers - and particularly natural language philosophers - have found interesting over the more recent past. But I find a great-chain-of-being approach dubious in regard to generating epistemic epiphany.
Or less t-rolled: McCarthy and Von Neuman.
Or more poetic: Dionysian and Saturnial.
Rigor comes in many flavors. The starting point for all of them is rigorous thinking. One of the signs that my thinking is not rigorous is that I wind up proving the point I set out to prove. Then again, setting out to prove a point ought to be a warning flag.
Even if it is inevitable that I do so. The key to rigor is recognizing that I am doing it. Lisp and Blub was where I wanted to go. My dissatisfaction led me to McCarthy and Von Neuman and that to another metaphorical connection between programming languages and literary criticism. And that fits with my larger interests; those being the ones which dispose me toward writing about software.
My intention was to take an honest (you may say phenomenological) look at programming languages: to have a system by which the idioms of the language, its communal metalogics, etc. are all part of how we assess a language. Not the reductionism of the "procedural-functional", pre-existing landscape.
I guess my point in referencing this landscape was to point out how inadequate it is, and how there might be a more complex and useful way of reasoning between languages.
My next article on the subject, which I'd imagine will follow soon, will certainly be slightly refocused and clearer. This is very much a just-finished-typing first draft that I put out to see what the response would be, and what the appetite for this kind of analysis in the programming community is.
In my experience, decision makers don't want to be educated about the benfits of your favorite language. They want to use a 'full stack' supported solution that has corporate backup and industry acceptance. They want to be able to hire someone to take over your code when you leave.
In the end, you can compute anything compuatble with any Turning complete programming language, although some are more elegant than others at certain problems. Human psychology will lead you to desire what you don't have, which in this case is your ideal niche language for solving your current problem at your shitty job. In the end languages are just tools and decision makers are providing you with the tools you are to use to do the job which minimizes their risk and maximizes their return.
And if your language is so great and will solve all your problems in an elegant and productive way, then make something with it and start a successful company. Then, if you make it for 20 years, wait for the articles saying how 'uneducated' you are having used it and you should be using some university ressearch language based on n-th level hyper-metalogics because you can write fib in 21 symbols.
Well it's this defeatist position i'm arguing against.
> you should be using some university ressearch language
I'm largely a PHP developer / "architect" by trade; I'm not in the business of pushing Haskell onto CMS developers.
I am interested in the programming language debate, and I'm not convinced "the market" is the best judge of what the most useful language is. That it happened to be, for example, the internet was coming of age during PHP's invention which gave us a generation of script-kiddes (98 - 06) seems to me an accident of history. This accident is being corrected by Zend Framework and corresponding language changes (06 - 13), so by today PHP is a language which resembles a pythonized java, rather than a javascripty perl.
The market in this instance is doing precisely what has been recomended by the leading theoretical voices in this area: coming from Zend, Sympfony, etc.
My intended contribution is to provide a conceptual framework in which to make sense of these changes, to advocate for these changes, and to compare potential changes. Mailing lists are abound with favorite features, I've yet to see a feature suggestion with a complete argument of the kind I would advocate.
Though some of the very best do actually get 95% of the way there (ontology: showing the new feature in other languages to "get a sense of what it is"; epistemology: showing how to reason about it in different areas; abduction: showing how it represents problems better than current technologies): if you look at a modern high quality PHP-RFC today you're getting very close to what i'm proposing.
The open source market of merit is putting these individuals in positions of power over the evolution of languages, it would be useful - i suggest - if we could have a standard which provides the means for establishing good arguments from bad arguments in this area.
The Best is chosen by the market that which is used to create the most successful product. By that measure, PHP is the best because Facebook is written in it. Best to the individual is much more complicated. The two will rarely coincide due to the uniqueness of the individual and the typical rejection of authority by intelligencia.
I would never say that PHP is the 'best' language for doing high abstraction, but it can be done. I also would not use it for such things, but you said you have (if you are the author of the article).
The 'Best' language is what you do your best work in, whether paid or not. Just don't expect to use it at work, and if you could, you would think something else would is best and pine for that.
It's the job, not the tools.
> It's the job, not the tools.
Again, this is what i disagree with. There really is something objective, independent, and "true" to be said about the utility of various languages in various problem domains. There really is a language which represents the problem space most accurately, for example.
It's just defeatism to suppose that "it's all the same, whatever works". Innovation is, of course, the opposite drive. It is at the innovative egdes of programming languages that these debates are had. If you arent part of this innovative edge then it may seem to you that the things youre using are just "market decisions", they are far from it. Every programming language you are using has not been created by the market, it has been shaped by the very debates i am talking about.
The market cannot create languages, it does not improve them, the market merely plays out a baggage'd history which language-theorists have to cope with. This is why PHP has changed so much: it is the force of the people who engage in this discussion on the market, to drive it towards "better".
It's true there are awesome specific language tools for your specific problems, that when tied together to create a total solution creates an unmaintainable mess that even the creator doesn't understand.
Look at the web.
I saw Kerrnigan talk about 'little languages' and I fully agree that a specific language designed to solve a problem is the way to go for small problems, like the ones when Unix was designed. Tying all these text based interfaces together is a maintenace nightmare.
Maintenenace is a much bigger problem than the actual problems being solved, at least for businesses.
Then there's Lisp, which is a mini-language designer's tool kit and the super glue to tie them together all in one integrated package using a well designed, common binding agent. Unfortunately the market has choosen not in it's favour.
The market will improve languages and in end game is that they'll all have the same features, and then it comes down to syntax and performance. They'll all be 'better' by today's standards. The problem is at that point is that the better bar will have moved.
And the guy that truly innovated during that time won't be found until the next century.
It seems to me a philosophy of code that begins at the "low" level of binary and machine instructions, would ultimately reduce to a contemporary British/American Analytic style of philosophy to the exclusion of all else, seeing as how the core level of communication is binary -- logic, to the exclusion of rhetoric, in the classical sense of both of those disciplines.
If you take a logic class at a basic level in both a Philosophy and CS context (as I did), you'll be struck by the massive overlap -- DeMorgan's theorem, modus ponens, et. al.
Setting aside for the moment the possibility of alternate CS architectures (neural nets, quantum computing and so forth), this would seem at first blush to limit the relationship of all computer languages to a 1:1 relationship with logic.
However this "reductionist" (foundationalist?) approach to finding a true semiotics of CS may not tell the whole story. As we move from the "low" level to the "high", say from the domain of machine code all the way up to, for instance, a rich, expressive Ruby DSL, the question arises, as it does for all language: have we acquired a surplus of content that cannot be simply reduced to core rules? At this point things get less Wittgensteinian and a little more late-Heideggerian, i.e., less analytic and more eidetic/phenomenological.
To remove the fuzziness from this notion, think of how, for instance, a first-person shooter creates the illusion of a smooth, analog reality from a rigid, frame-by-frame progression (as cinema does as well -- and this question has been explored at length in that world). As humans interacting with the result of the code, we infer content between the frames -- we make up the difference and experience the virtual world presented to our senses as continuous analogical rather than chopped-up and merely logical. Code never exists in a vacuum -- it is experienced by humans and humans bring meaning to it, as much as it brings meaning to them.
There's a similar phenomena in OOP, where the meaning of the code is not merely the sum of its final binary product ("true" or "false") in the world of machine instructions, but in the metaphor it presents as a model of an object or event in the world. This code is constantly changing (literally) as it intends towards richer analogies with the world, and we as coders (and users) make up the difference by bringing real-world experience to it when we "read" the code. In other words, code is never "true" or "false" as a function of its binary reduction, but is a constantly-changing artifact of some human being's attempt to analogize the world, read by other human beings attempting to understand and refine that analogy.
In this sense the general problem is, as another poster phrased it so well, semiotic rather than epistemic or even logical. But I'm not sure a "closing over" or reducing towards core CS mathematics helps us understand the ontological status of code, any more than Wittgenstein or the Vienna School has given us a salutary reform of philosophical discourse with pure analytical techniques in search of a perfect, ambiguity-free symbolic language.
Language of any kind acquires a surplus of meaning as human beings interact with it and bring their own analogies to the experience, and I think this applies to code as well. So it may be that it's OK to move towards a philosophy of code that takes a phenomenological approach to the final product -- the code itself -- rather than (or probably in addition to) looking for meaning in the binary rules that compose machine instructions.
OP writes:
>> With this in place we can make certain classifications (in a quais-aristotelian manner): functio...
Whilst my inclination in being a Husserlian first-and-foremost is to engage in the former, I think the latter is definitely of interest.
The third choice, and this is i think what you mistake my current project to be, is this "binary reductionism" or "symbolic reductionism" where DSLs are either 0/1s or predictate-object statements. I disagree with this entirely, and i think it leads to the current "chose what you like" defeatism.
I would like to explore the middle ground for awhile, an "early heidegger" if you like. I think that's going to get a lot more traction in the implementation community. There is definitely a lot of space at this moment in time for some real substantive work in all non-reductionist areas.
NB. i think my use of "metalogic" has confused at least a couple of people into thinking im talking about something CSy. Logic in the sense a continental philosopher would use it: a mechanism of reasoning. As I say in my "notes on terminology" i mean to include things like the domain/business logic. I think that, for example, that "a car shouldnt start with no fuel" is part of the program (the metalogic of the program, not the logic itself: it is part of what we need to consider to fully understand the progran). Perhaps i shouldve used "meta-rationale" or something.
EDIT: A minor quibble: maybe "phenomenalist" isn't the best term to use in close proximity to Husserl. He used the idea of "phenomenalism" in contradistinction to "phenomenology" -- the former being a kind of reductionist approach to seeing essences as merely the "contents of consciousness", and the latter -- which is to say the general Husserlian approach -- sees essences, or "eidetic objects" as phenomena that intend towards the external world, at least external to subjectivity.
No, not if it's still executing on a physical computer.
If it runs on a computer it can be reduced to a simple set of core rules that is a logic, no matter how complex, by virtue that it is running on a computer.
Real math doesn't have to follow such rules.
But of course that's not what we do as coders. The code is never "complete" -- it's always striving towards some "meaning" that is for a human being. You cannot exclude the meaning of the code from the subject that it is intended for. A human being perceiving the code is always redacting the meaning of the code, not as a function of ones and zeros, but as a function of the analogy that code creates, pointing not towards electrons but towards: a product in a warehouse; a sum of sales at the end of a quarter; a flaming ball of death hurling towards my first-person-virtual head. And so on.
In this sense the electrons/ones and zeroes are no more significant than the paper that The Iliad is written on + the ink it is written with.
All works of substance share such an 'incompleteness' and meaning. We just look to non-programming works as complete in a sense when they are recognized. But such works are constantly being rewritten in different forms in order to produce the effect they are known to be able to produce.
In the end, everything is a program in some sense whether to be executed by a brain or computer.
The point of what I was saying was to refute the idea that somehow our programs are so complex and have such amazing behaviour that they will not be reducable to a core set of rules. If the programs run on a computer they must be computable and thus reducable to a core set of rules. That's it.
Whether we can be reduced to that core set of rules is not evident yet. Using logic, a statement that denies it's own provability in the language it was writen in was crewated. Logic was not thought to be able to lie, until it did. Who knows what the right brain can make it do?
This was the statement that I started all this with. I am taking "surplus of content" to mean "a bunch of programs" in this context due to the quote speaking of a Ruby DSL. Programs can be reduced to a sequence of logical operations if they are computable, which I would take as 'core rules'.
Therefore, only non-computable programs can not be reduced to a set of core rules by the definition of computability. QED
Maybe you can explain 'it is much more than merely what happens when it runs'? Do you mean that what it took to create the program was greater than the resulting program and it's effects caused by running it? That the meta-logic used to create the program is far greater than the logic in the program itself?
I am 'much more' than the results of actions when I act. I have a past history that has created me, a physical body being used to act with, the world I am acting in, and the resulting effects of my actions, plus everything else I have missed. I that what you mean by 'much more'?
tl;dr -- Yes!
The non-tl;dr:
By "surplus of content" I meant to say that the rich, english-like syntax of something as high level as a Ruby DSL acquires rhetorical, intuitive meaning that cannot be reduced to a 1:1 symbolic relationship to the compiled instructions. But this "surplus" is supplied by the perceiving subject, in the same way that natural language itself is interpreted and redacted by the human being reading a particular work.
I didn't mean to say that mathematical magic had happened as a result of complexity.
I'm coming at this from a philosophical as well as a CS background (I studied both, but ended up in software). In philosophy, there is an ancient tension between rhetoric and logic. Logic is a brute force process -- a set of rules where the same input always produces the same output, just as (and, it turns out, according to the same rules) the same electrical series of pulses supplied to logic gates and pushed lockstep through them by the CPU clock will always produce the same output. Many analytic philosophical papers don't look all that different from a Karnaugh map describing the structure of a multiplexer, and the results of both are equally deterministic. This school of thought was galvanized by and built heavily upon the Principia we spoke of earlier.
But there's another school of thought in philosophy that uses rhetoric and poetic devices as well as logic to achieve its ends. Language in this view (From Aristotle's Rhetoric to Heidegger's Building, Dwelling, Thinking) is decidedly not deterministic, but provocative, emotive and intuitive. Human beings communicating with each other with language use analogy and a commonly shared narrative history to provoke meaning to each other, rather than reduce statements to mere facts that are reduced to true or false. This school is commonly referred to as the "Continental" school, due to its prominence in France and Germany. Analytical approaches dominate British and American universities.
So, my argument here is a function of a larger question about language itself. I'm saying that the English-like analogical techniques even at the lower levels are provocative and provide more meaning than can be reduced to the underlying logic. This is not to say that any software is less than deterministic -- of course it cannot be non-deterministic because the core functions are purely mechanistic and deterministic. So, an address changed in a DB is always changed in the same way, and a joystick moved in such a way to provide a stream of x,y coordinates to a controller will always move a character in the same way, etc.
But the meaning on the other side of those events is supplied by a subject. At an extreme level, think of a high-resolution bitmap of Rembrandt's Philosopher in Meditation. The presentation of the pixels is deterministic, but the meaning is supplied by the subject. I'm saying that to a greater or lesser degree this redaction, the interpretive gesture that supplies a surplus of meaning is taking place at basically every level of computation. To a lesser degree at the low levels ("push", and so forth), but to higher degrees the closer to English-like (3G languages, etc.) we get. Images and literature presented in UTF-8 are just radical examples on one side of the spectrum, of how deterministic CPU instructions can result in this "surplus".
>> The point of what I was saying was to refute the idea that somehow our programs are so complex and have such amazing behaviour that they will not be reducable to a core set of rules. If the programs run on a computer they must be computable and thus reducable to a core set of rules. That's it.
Considering a piece of software as a mathematical construction -- let's say, for instance, a C++ program compiled by the LLVM, and the LLVM bytecode expressed as machine instructions for the given chip, reduced to pure binary math, I totally agree with this.
I think if we look hard at the structure of CPU instruction sets, GPUs, math coprocessors and the like, we'll see even at the lowest level attempts to present analogies and manage the limits of pure logic as presented to a human consciousness. So even at the lowest level of CPUs, the language of analogy is present, and that implies that a human subject is going to supply the content "between the electrons" as it were.
There are no attempts to manage the limits of pure logic at the silicon level. The entire analogy of the CPU is to implement pure boolean logic through strict controls and limits of voltage levels in analog circuits.
The logic available to humans is the same available to CPU. Godel used pure logic to create an unprovable statement that was both true and false. What he did to create such a statement is what computers do not (yet) have, but it most certainly is not logic, and I think that would be what you say is 'between the electrons'.
But whether we can program and execute this 'insight' on a finite logic system is another question that is yet to be answered. What we can program we can execute and thus can be reduced to basic forms of symbolic logic, whether powered by electrons, pen and paper or the brain.
But this isn't the whole story either, right? What is "pure boolean logic" anyway? It doesn't exist even at the lowest levels. Logic is always applied to a statement of some kind, it doesn't exist as a pure abstraction. It's interesting you bring up Godel. Russell and Whitehead attempted to map logic to math in the Principia, and were refuted by Godel's incompleteness. I don't want to make too much of incompleteness (as many do), but let's not make too little of it either. At some point logic can only do so much for you, and you have to resort to analogy. I think this happens much earlier in CPU architecture than you're allowing for.
What you have at a low level on the CPU are a set of instructions arbitrarily (to some degree) defined to efficiently model analogical actions like "push" and "move" and "shift" and the fundamental operations that map to assembly. And here we've already moved out of the realm of pure logic -> math envisioned by Russell and we've created logic -> language, by way of analogy. This is, I would argue, a rhetorical move rather than a logical one. We've already prepared/redacted the ones and zeroes for human consumption, and assumed a specific semantic to provide for a human imagination that will accept and adapt to the semantic.
There's no magic in the computer and no analogy other than designing analog circuits that clamp voltage levels and a very large finite state machine. It's logic all the way down, limits and all, though instead of incompleteness there's incomputability. 2 sides of the same coin.
Pure boolean logic can be reduced to symbol manipulation using rules of substituion and inferrence, but, of course, what is that composed of? Godel showed that metalogic is made of logic, you just elevate the types.
What is logic made of?
BTW I know computer architecture very well. There's no analogies, just abstraction.
"The instructions in the CPU are no different from the statements in the Principia"
As the goal of the PM was to show that math was a function of logic, we can use 'statements in the Principia' and 'the classical rules of logic' interchangeably in this context.
There's two questions at play in this conversation. One is whether in principle the question of the meaning, as such, of software, can be considered apart from the whole act of perceiving the software, either as code or as the final UX.
As the idea of "meaning" always involves some kind of communication, a sign or set of signs shared between two subjects, and as the idea of software pointing back to its own internal state as the content of this communication would render the communication pointless and the signs merely tautological, I think the question of meaning must bring the subject, and the act of perception and interpretation, into consideration. Scientific statements of fact, regarding the physical position of all of the machine's atoms and electrons, have content, but not "meaning" in the common sense of that term intended in the current context.
The second question, which I'm pretty much winging here (although I have long-held opinions about the PM and its relationship to the same kinds of considerations about "meaning"), is whether CPU instruction sets, as subsets of the rules of logic, dictated by the fundamentals of CS (Babbage, Turing, et al), are "pure" expressions of logic, or whether in the act of abstracting logic into the restricted set of instructions, we've performed an analogy.
I think the same question comes to bear -- it matters which perspective you're taking, whether "meaning" is merely a construction that hides the more essential question of physicality, or whether ideas like "push" applied to things like registers and addresses and binary values are linguistic in nature and convey more than just tautological facts about the physical state of CPU internals.
I guess this pushes the question out -- why does a computer exist? To represent to the world the bare facts of its own internal state? Of course not. A 64-bit address that points to a sequence of binary values is never just that, it's always a sign that points to something else -- a person's address; a representation of an English character; a shipment of goods; a lat/long location of some specific place in the world.
So, a "move" or "shift", to you, may reduce backwards to a collection of logical operations. But even there, what has happened? Nothing's "moved" or "pushed", that's just an linguistic trick to help us think of the electrical currents being turned "off" and "on" (and even those are analogies) in mentally-manageable concepts that map to the rules of logic. To me the question of "meaning" has already been broached at this point, if we wish to see computations as significant beyond, again, tautological statements about CPU internals. If we take the latter point of view, we have mere mathematical statements that have the same point as pure math -- the elegance (if that) of its own proof, not any relationship to something external to itself. But of course, that was never the real point of computers from the very beginning.
The Godel/Russell/Whitehead question is apt here, because this is a very similar question, if not the same question. Is math a function of logic -- that is to say, can math be "true"? The point of the Principia was to bring about the dream of the Vienna Circle, a purely mathematical and systematic technical language to resolve all philosophical ambiguity once and for all. While there remain a few holdouts for that dream, the general consensus is that this did not work out -- Godel was only one of many problems with Positivism specifically, and I would argue, Analytical Philosophy in general, although in recent years the Analytical/Continental divide has b...
The starting point of this computation model is:
The set of symbols is countable and includes: Nonsymbols represent mappings that take nonsymbols or symbols and input and return nonsymbols or symbols. That's not well-defined (and it's not even a set) but we are only concerned with those that return symbols (i.e. can be observed) after certain input patterns. For example, A : {_ => A} and B : {_ => B} are both nonterminating but behave identically based on what we can actually observe; it's meaningless to argue about whether they are "the same" and since nonsymbols can't be eq'd, also unanswerable.Code (as in source) is essentially a process that takes a natural number (viewed as an array of bounded natural numbers) and converts it into something that behaves as a nonsymbol, at least over inputs we care about. Of course, there are many nonsymbols that we can't represent in code, such as one corresponding to the Halting Problem,
Maps, vectors, and sets give us tools for describing extremely well-behaved (finite support; returning _error on the not-found case) nonsymbols in code; functions give us a language to describe well-behaved infinite-support nonsymbols.The neat thing about traditional functional programming is that it carries an implicit will to focus on well-behaved nonsymbols. Object-oriented programming hasn't stepped away from the full space of nonsymbolness-- and that's a big part of why it's so damn hard to reason about it.
One quibble -- I would prefer non-negative(!) terminology, i.e., what is a "nonsymbol"? A collection? A monad? A duration? An event? A process? A closure? Some category that encompasses all of these? If so, can we find a positive semantic rather than negative?
>> Object-oriented programming hasn't stepped away from the full space of nonsymbolness-- and that's a big part of why it's so damn hard to reason about it.
Well, this gets back to the ancient problem of "The Third Man" and categories in general, right? One must categorize and "contain" meaning, trap it in language a moment in order for it to remain meaningful, but paradoxically the external thing that it symbolizes is always changing, so it's always escaping any semantic trap we set for it. This of course has practical (concurrent) implications in software as well as the high-falutin metaphysical stuff Plato was talking about.
These days I'm starting to wonder more about how we may not be using state-machines quite as effectively, or as often, as we might -- changing functionality more aggressively as the data changes, in order to move towards a better analogy with the ever-changing "real world" that software is trying to model. Might this be more self-aware starting place to integrate functional and object concerns -- both of which are necessary?
Good question. I tend to think of it as a stateless (I haven't added state to the model yet, and there are myriad ways of doing it so I'm debating how to do it) object. To apply it to something (symbol or nonsymbol) is like sending a message, and the result of the apply is what it returns.
Nonsymbol is deliberately vague, meaning "we don't know what this is". We need a context to make sense of it (e.g. observe _size, then if that's a natural k, observe 0, ..., k-1; it it's not, it's not in the Vector context) and if it doesn't play by the rules of a known context, we're screwed. We continue to need context when we "array-ize" nonsymbols into natural numbers.
Nonsymbols are like functions; however, a function has a well-defined domain and non-symbols don't. However, the only useful nonsymbols are "function-like" in that some sequence of symbols and known nonsymbols will produce an observable (symbol).
A Haskell program consists of declarations of functions and constants. Constants belong to the class Eq
http://www.haskell.org/ghc/docs/latest/html/libraries/base/D...
Functions do not.
How is your notion of nonsymbol/symbol different from function/constant?
http://blog.mjburgess.co.uk/2013/05/notes-on-systematic-argu...