I rarely see advice with which I wholly agree. This is one of those rare times that I fully agree with everything g. Good seasoned advice from someone that has been doing this fo r a while.
Those are way too abstract advice when you start programming.
You can only understand them because you lived those situations, which implies experience you don't have.
I would say (specifically to my young self):
- There is no substitute for doing. Less tutorials, more coding.
- Stop being obsessed with quality: you are not at the level where you can provide it yet. Do dirty. Do badly. But ship. Some people will be mad at you, and they are
right. But do it anyway. Yes, reboot the servers with users on it. Yes, commit the spaghetti. You'll reach the level you want to avoid doing all this.
- The users don't care about the tech. They care about the result.
- Doing something for the hell of it is worth it. Just make it separate from the point above so they don't conflict.
- Programming is a battle against complexity. Again and again. Put that on a post-it. Make your decisions based on it. You will suck at it, but try.
- You have imposter syndrome because you are an imposter. You are really bad. It's ok, doctors hurt people for years while learning to save them. Don't sweat it. It will come.
- You need faith it will work out. On the long run, you'll get better at this. But in the short term also. You'll find the bug. You'll figure out the solution. It will happen if you keep at it even if it can be frustratingly long and unfair. Even if it doesn't feel like that right now.
- The right team is way more important than the right tech. If you are alone, get in touch with people who will lift you up from time to time.
- don’t read opinions from a list and take it as gospel. Adapt to the jobs you’re in, and you’ll develop your own opinions, but now with experience to explain why. Opinions are formed by getting repeatedly hit with the consequences of your (and other’s) decisions, and everyone just has to take enough hits till the pattern seeking area of your brain takes over.
Yes true. Mostly just opining on the “list of everything you gotta’ know” is a bit TOO concrete IMO. (As opposed to too abstract)
I was quoting out of similar things years ago at the start of my career, “it should be done this way because X said it” didn’t help me at all.
It feels like insulation against being called too junior, since, you can just wholesale adopt someone else’s list of ideas and you’re good. But making mistakes because you’re new to the career is precisely what forms the base of those opinions in the first place.
Maybe add to that: try to stay long enough in a role to really feel the consequences of your actions. Even better if you're on pager for a while too. I know it's not trendy to stay in a job for long these days, and conventional wisdom is it's not great for your salary either, but one thing it will do is allow you to understand whether decisions you made were actually good or not.
There are roles I've been in where it's only been years later that the true impact of decisions made was actually apparent. I'm glad I hung around long enough to experience that.
I wholeheartedly agree with this advice and have for over a decade now. I think there’s something in the fact that you know you caused that that helps the lesson stick. Potentially you will understand the assumptions you had then vs now that allows you to clearly see the underlying lesson to take.
He is better at creating actual products than most people I met because his livelihood is on the line: he makes money only from websites.
So he is not living in the abstract idea of best practices, he had to make it profitable for the last two decades, with little resources to go by on top of that.
And he does.
After 10 years, he is still asking me how to write git commands. And I'm still learning from him to shed a lot of BS from my practices.
I've been called back to code long after I left an org, and have had to review my own code 10-12-15 years after the fact. Seeing both the positive and negative aspects of decisions having played out in the real world was something it's hard to get from reading, and I'd argue somewhat harder to get even if you stay inside the same org for that length of time. Staying on the inside, you'll rationalize a lot of the changes to the code, the team and org over time, and it may be harder to be more objective about the impact the code has had.
I was quite proud of some decisions, but realized the negative long term impact of others. Trying to share that experience and whatever 'wisdom' or 'lessons learned' with others has been its own challenge in some situations, because you can easily come across as "the old person" who doesn't "get it" wrt to current trends. Some issues are evergreen and fundamental, but it's difficult for less experienced people to understand why some of these things are really core. I'm not sure there's much substitute for experience in many cases.
To the extent users do care about the tech, they care about performance not how "clean" the code is or whether you're using the newest framework. Users hate when software is slow or uses an exorbitant amount of memory.
Most users have no idea how much memory a piece of software uses. They care about user experience though which means they will care if the UI hangs or is unresponsive.
Bingo. I’d go further and say they don’t care about your application at all. They just want to do something, and your application’s quality is measured by how little it stands in the way of accomplishing that.
The sad fact is this encapsulates features (ease of development, a framework probably does help you ship faster), adaptability (clean abstractions that are easy to work with), and performance.
Finding that balance is always going to be hard but they’re all important!
Yeah, I totally agree. Those other factors are internal optimizations. What gets me is when a team wants to switch horses to new tech and do a forklift upgrade just to implement something using the new hotness.
This 100%. It's especially noticeable in the world of game development, where you see games that are ridiculously buggy and poorly made (on a coding basis) selling millions of copies and changing the industry. The original generation 1 Pokemon games are probably some of the best examples of this, though I'm pretty sure anyone who's reversed engineered any game from the Atari era onwards has probably been left wondering "what the hell were they thinking?"
But it doesn't matter. They were designed well, they were fun to play, and millions of people enjoyed them.
These seem just as abstract as mine, if not more so, plus at least I provided examples where I could. Feels weird to criticize my post for general advice + examples, then come up with your own general advice without examples.
Also this was just an analogy I know, but doctors definitely don’t hurt people for years while trying to save them, very different profession from ours, if anything doctors earlier in their career have been shown to have better results.
I think you are trying to address different audiences. While your tips are mostly targeted at people who are already working as programmers, the parent comment's tips are mostly targeted at complete beginners.
E.g. this tip:
- There is no substitute for doing. Less tutorials, more coding.
is directly addressing a common mistake for absolute beginners. Many beginners will read (or worse yet, watch) loads of coding tutorials while doing little themsves. It is an issue a complete beginner encounters and understands.
Your tip on the other hand:
> If you (or your team) are shooting yourselves in the foot constantly, fix the gun
is addressing people working on medium to large projects with internal tooling. That is not a situation a complete beginner finds themselves in; it's a situation someone who already works in programming for a while finds themselves in.
I wouldn't necessarily say your tips are too abstract; they are simply too high level for a complete beginner.
That is not necessarily a bad thing; perhaps the you of 15 years ago already had the basic understanding necessary to be able to comprehend and make use of your tips.
I really liked “You should know all the major shortcuts in your editor. You should be a confident and fast typist. You should know your OS well. You should be proficient in the shell. You should know how to use the browser dev tools effectively.”
Typing skills are severely underrated in order to professions and roles adject to our professions like PM.
I think you both provide some kind of generational advice, like parent serving kid with life advice. Unfortunatelly, or fortunatelly, they will have to learn it by experiencing own failures first.
I didn't read this as a criticism of your post, just as another point of view. Your post is more organized and probably thought through, this guy's is an opinion. No need to get defensive.
My experience with doctors, of which there has unfortunately been plenty, is that the older doctors are much better. If I need something delicate done I want a doctor who's 50-65ish. They know what they're doing, they don't screw around, and they've seen enough to know that yes, your case is the common thing they've seen 500 times before, and that this treatment works and that one doesn't.
Don't worry, you're not a real imposter. You've just inadvertently ended up in a position where you're expected to be one. Just fake it until you actually become a true imposter.
You probably are. But most of your colleagues as well :)
Most adults are kids in big meat suits, they fake it a lot.
I started to live like I was not completely worthless at 35.
Not saying that to be proud of it, just stating that if you think humanity should do better, the first person you'll judge is you.
It will be glaringly obvious you are not meeting your own standards.
The higher your standard, the longer it will take for you to reach them.
And the way to get there faster is to ignore the shame, and do it anyway. Because if you don't, your growth will be slower, and you will do more damage for longer.
Real life means real consequences.
It will make you more tolerant of others as well. Way more tolerant.
> There is no substitute for doing. Less tutorials, more coding.
This may be good advice for yourself in the past, and for many people at lots of times, but I'd hesitate to give it as general advice. Reading code others have written should not be understated as way to learn valuable things from fundamental patterns and algorithms to language features and idioms. If you have a job in a team, this may happen anyway, but it's possible to write lots of code without realizing there's a better way.
Same. I would fall into the same trap. One more article, another tutorial, another chapter of a book… juuust to make sure I understood the concepts; and what actually helped? Just coding, getting it wrong, fixing it, getting it wrong, fixing it, etc.
Although, I should point out that reading code is not the same as reading tutorials. Reading code occurs with a kind of intent and focus that is missing when you don’t know much, and thus you may end up falling into the trap of studying multiple tutorials without really trying out much yourself.
Tutorials include some code which is intended to be exemplary and simple enough for a new programmer to make sense of, so I wouldn't discount it as a part of reading code. Practical code does not always have those properties, although you'd certainly be missing a lot if you only read code from tutorials.
I completely agree with the claim "There is no substitute for doing", and I might even say that code you read without running and tweaking it doesn't count.
We might have different ideas about what constitutes a tutorial. Online tutorials vary massively in quality, but something like the official Python tutorial https://docs.python.org/3/tutorial/ is a fine resource. I would even say that the categories are not mutually exclusive - I would call "Structure and Interpretation
of Computer Programs" both a book and a tutorial. Different people probably find different resources most helpful.
Also, "To learn how to implement computer programs, read books about it." contradicts the original comment's "There is no substitute for doing" more than I'm happy with.
I wouldn't call it a tutorial. It's the reference manual, but with more words and some stuff about pip and so on to make it more palatable to someone who isn't already initiated.
You need to start by reading somewhere, or you'll probably not get anywhere, as evidenced by pretty much zero office grunts using web browsers every day having figured out the JS console and application development on their own without reading about it. I recommend books over tutorial texts.
But yeah, as I alluded to, you need to build stuff and learn from other builders.
I'd definitely agree that reading others' code can be a good way of learning, especially if you take the time to disassemble it line by line and look up things like what language features and functions they're using, how it's structured, etc.
I feel like this is bad advice, really, you need to be obsessed with quality and growth, but you can't let that stop you from shipping. Try for clean to the best of your ability in the time constraints you have, but accept that it will be dirty.
I think aiming for clean is good, but it’s really hard to pin down what clean means when you’re starting out.
I feel like just emulating what you see in your first few jobs is ideal. (As in, ask coworkers who know the thing you’re working on) It could be great code to be inspired from, or mediocre. Either way you get some input about what decisions result in what outcomes, and what the outcome “feels” like.
And if it comes time to change one or many of those decisions later on in this codebase, the person doing it gets a uniform codebase to work from! Unique abstractions and fixes in random places makes refactoring harder.
The medical industrial complex clearly thinks it is, otherwise there would drastically lower patient loads, allow more spots for medical school, and provide better working conditions that would allow for fewer mistakes. None of this is happening and is an implicit acceptance of medical error deaths
You think they're called practices because of the less common usage of the term to mean an amateur learning something, to imply they're inexperienced/without training?
Rather than it being the place where a practitioner works? ("someone whose regular work has involved a lot of training")
They obviously do. Just look at the way they run residency programs. They work those poor kids to the bone under minimal supervision. Only the most disingenuous can pretend it's an accident when something goes wrong. The system is clearly designed to harm people.
Show me a hospital, and I'll show you overworked, under slept young doctors that have to deal with paper work, and are swimming in a field with heavy lobbying.
But even without this, the thing is, when you don't know, you will make mistakes.
And when your job is to take care of people, mistakes hurt them.
It's the nature of things.
Last year I had 3 medical errors, one that almost got me killed, by very well-meaning professionals.
Closing your eyes and pretending it doesn't happen is naive at best.
But people have to start somewhere. They can't stay in theory forever. And no amount of preparation will save you from making terrible mistakes.
Since we can't put an experienced doctor behind every intern 24/7, there is no real solution to this for now.
Planes and bridges are problems where the quality control can be centralized and you can therefore put a lot of redundancy.
Even then, the Boeing scandal shows it's not bulletproof.
It's not the same for medicine. There are way more doctors than you can put safety nets. Also, when a plane crashes, it's on the news, it costs money and PR. Much less so with doctors.
The plane industry is not inherently more moral, just more liable. Responsibility in the health industry is way more diluted.
It's not that we WANT to pay the price of people learning. I wish I would not have had the wrong meds given to me last year.
It's just that the system is not currently set up to do it otherwise.
So you can feel guilty and stop moving, and you will not learn. You will not grow. And you will actually hurt people for a longer period because you will still make mistakes.
Or you can own it and accept the inevitable: doing things means having actual consequences on the world.
If you want, you can dedicate your life to change the whole system and make it better. But I don't think anybody in this thread is doing that right now.
I did see a lot of people on their high horses, but doing nothing though, paralyzed because they were looking for a way to never break anything.
However, normalizing the loss of human life as part of the learning algorithm is psychopathic. There's a time and a place to make mistakes... In mission critical situations that's in training and everywhere else we should aspire for safeguards that prevent the loss of life due to mistakes.
In medicine a lot of these deaths can be prevented through greater care. It's not different than engineering in that respect. The greater problem is decision makers putting profits over life. And this kind of mentality in this thread is just fuel for that kind of behavior. It's gross.
I think you're misreading a singular opinion as occurring between two disparate points here.
The initial phrase was
> doctors hurt people for years while learning to save them
It's then a separate reply from someone else about deaths from errors/malpractice.
So, nobody seems to be expressing the mentality you are, correctly, lambasting (at least so far as I've read in the comments). But, as it is relevant to the lessons we'd all want to pass back to ourselves (in my opinion, it's because we wish to hold ourselves accountable to said lessons for the future), let's address the elephant in the comment.
Normalising deaths, especially the preventable ones, is absolutely not something that anybody here is suggesting (so far as my read of the situation is).
Normalising responsibility, by recognising that we can cause harm and so do something about it, that seems far more in-line with the above comments.
As you say it yourself, there's a time and a place, which is undoubtedly what we hope to foster for those who are younger or at the start of learning any discipline, beyond programming, medicine, or engineering.
Nobody is saying that the death and loss is acceptable and normalised, but rather that in the real world we need a certain presence around the knowledge that they will occur, to a certain degree, regardless. So, we accept responsibility for what we can prevent, and strive to push the frontier of our capacity further with this in mind. For some, that comes from experience, unfortunately. For others, they can start to grapple with the notion in lieu of it through considering the consequences, and the scale of consequence; as the above comments would be evidence of, at least by implication.
These are not the only ways to develop a mindset of responsibility, fortunately, but that is what they can be, even if you find the literal wording to suggest otherwise. I cannot, of course, attest to the "true" feelings of others, but neither can anyone else... But in the spirit of the matter, consider: Your sense of responsibility, in turn, seems receptive to finding the areas by which such thinking can become justification for the very thing it would otherwise prevent, either as a shield purpose-built for the role or co-opted out of convenience. That too becomes integral, as we will always need to avoid complacency, and so must also promote this vigilance to become a healthy part of the whole for a responsible mindset -- lest we become that which we seek to prevent, and all that.
Exactly as you say, there's a greater problem, but this thinking is not necessarily justification for it, and can indeed become another tool to counter it. More responsibility, more mindfulness about our intents, actions, and consequences? That will prove indispensable for us to actually solve the greater problem, so we must appreciate that different paths will be needed to achieve it, after all, there are many different justifications for that problem which will all need to be robustly refuted and shown for what they are. Doing so won't solve the problem, but is rather one of many steps we will need to take.
Regardless, this mindfulness and vigilance about ourselves, as much as about each other, will be self-promoting through the mutual reinforcement of these qualities. If someone must attempt to visualise the staggering scale of consequence as part of developing this, then so be it. In turn, they will eventually grapple with this vigilance as well, as the responsibility behoves them to, else they may end up taking actions and having consequences that are equivalent to the exact mentality you fear, even if they do/did not actually "intend" to do so. The learning never ends, and the mistakes never will, so we must have awareness of the totality of this truth; even if only as best we can manage within our limited abilities.
> A recent Johns Hopkins study claims more than 250,000 people in the U.S. die every year from medical errors. It is ok to be an imposter.
Is what I was originally replying to, which, at least to me does seem to imply preventative deaths are some kind of training tax during ones 'imposter stage'. Perhaps not the intention of the poster.
Making a mistake or a reasonable but ultimately incorrect call is not malpractice. Doctors are just people, just like the rest of us, and certainly you. It’s scary if you think that the medical system is more of a safety net than it actually is, but your gripe is with the precarious nature of life, not medicine.
Figuring out how to be a doctor by faking it til you make it most definitely is malpractice though, and ‘don’t worry, you’re an imposter but you only learn by screwing up’ is very much not how doctors learn their job.
The Hippocratic oath is not ‘first, do some harm, as long as you’re learning’.
>Do dirty. Do badly. But ship. Some people will be mad at you, and they are right. But do it anyway. Yes, reboot the servers with users on it. Yes commit the spaghetti. You'll reach the level you want to avoid doing all this.
I think you're likely to reach that level a lot faster by trying and failing than by not trying at all.
A way to rephrase your point about complexity is this wonderful quote "Developers are drawn to complexity like moths to a flame, often with the same outcome" (Neal Ford)
I've experienced this too, but I never understood why this phenomenon happens.
Is it because we start adding abstractions before we understand the problems?
Like, there's a certain threshold where keep things too simple makes them too complex, so people start introducing abstractions to reduce complexity. But often it's the wrong abstractions and then we accidentally end up with worse complexity that's harder to unravel.
The commonly used term, also mentioned by Fred Brooks, is accidental complexity.
Accidental highlights the non intentional nature of devs introducing complexity.
Isn't accidental complexity just complexity that's not part of the problem domain? Say in the context of serving content/downloads, accidentally complexity would be caching, proxies, CDNs (etc). Basically stuff that we have to deal with to handle or optimise downloads, but isn't inherently part of the "just downloading files" problem?
It’s because when you’re in the middle of things with a lot of context in your head, adding another little wrinkle feels like a negligible complication (in particular if the new wrinkle is ostensibly to reduce some complexity, or to ship more quickly), but those complications accumulate up to the limits of any developer’s comprehension (and beyond) rather sooner than later. Hence developers tend to work close to the limit of what they can handle in terms of complexity, which for anyone who hasn’t all the same context in their head (like the developers themselves some time later) is really more than they can handle in an effective manner.
From another perspective, this is a form of entropy: There are many more ways to increase complexity than to reduce it. This is also the reason why biological life has been getting more complex over time, the only criterion being if it’s fit to survive and reproduce.
Your original statement completely ignores the massive amount of barriers put up during your training. From my own experience I can tell you that students are usually quite self aware that they are learning and on a short leash. The most egregious mistakes I've seen were almost always a result of understaffing and gaps stemming from leadership issues coming from the top down. The small mistakes and cut corners are from tired overworked people.
Nobody accepts "hurting people" as a component of training. It's an absurd statement.
I never really got the "it's ok to be an imposter", "it's ok to be bad" part... And the internet is full of people saying things like "I am a developer and I have no idea what I'm doing haha".
Seriously, you might be inexperienced or not know everything, but you should clearly not be an imposter and you should be confident in your ability to improve and understand what you don't understand yet.
Take responsabilities and ownership in what you do, don't get behind the easy excuse that it's too complicated. You are the professional and you are getting paid for this.
> - There is no substitute for doing. Less tutorials, more coding.
I'd rephrase that as "just write something!"
Many times I find myself being the classic example of 'perfect is the enemy of good' - I'll think about the perfect solution, rather than write something *now*, that works, and refactor towards perfect. TDD and all that.
Other things:
- Beware the beta and the boss. If it works, it will invariably get shipped if your manager sees it. Many managers cannot put a value on the future cost of maintaining something that's barely good enough.
- Classic Confucius: "I hear and I forget. I see and I remember. I do and I understand." If you're interested enough to read about some tech/language/framework, write something (again!).
- Learn at least one other language and ecosystem in addition to your main one. Even if it is syntactically similar (C++, Java, C#, for example).
> Doing something for the hell of it is worth it. Just make it separate from the point above so they don't conflict.
This is a great use for new, standalone open source modules: feel free to experiment with styles or techniques or goals that you wouldn't want to justify. (Or for experiments that you start and then abandon without ever showing anyone.)
For example, when I was making lots of packages to help build out the Racket ecosystem, I'd already made an HTML-writing package, but I felt a craving to do one that used syntax extension at compile time, rather than dynamic s-expressions. So I just did it, as a separate package: https://www.neilvandyke.org/racket/html-template/
I don't recall anyone saying they saw value in it, but I like it, and I scratched that itch, and added another skill to my programming mental bag of tricks.
Most of these lists do not advise on "programming" the task but rather "programming" the job/position. There's no problem with that, I'd just wish it was labelled "software engineering advice" or "advice for programmers". Few of the points are about programming itself and the list is overall pretty good.
Consider this just a rant from an older programmer who hasn't fully recognized that 'programming' is now largely a social endeavor with online info for everything, language and library ecosystems, etc. I wonder how much of this information I would have internalized if it were all available in my time. Seems like the kind of thing I might read and nod in agreement and forget to apply when relevant without some hard earned run-ins (which is how I'd picked them up).
As for actual programming advice, the thing I'd highlight is to look at the data first and foremost. It goes from initial conditions to post conditions. The differences are what your program/function does, but both the pre/post conditions should be able to be fully described as a valid static state. Once you understand that, the problem, the code is largely plumbing with a small part that applies the functional transformation. There's so much focus on the form and style of "the code" that seems to consider it the main thing rather than an artifact of getting the main thing done: think any time there seems to be fancy abstractions that don't provide value to offset its complexity. To relate it to a point in the post, if you can't connect the difficulty you're having with the logical transformation that needs to be done (e.g. from database state to state), it's likely self-inflicted (or it could be due to a poor choice of database schema). Similarly for poor choices of request/response formats--basically bad plumbing between systems (and not intrinsically hard because of the information being handled).
"Those are way too abstract advice when you start programming."
I have come to the conclusion that the use of these sorts of posts is not that the reader, young or otherwise, will instantly and correctly apply all the lessons to their lives.
It's more about sensitizing people to problems they may not currently see, and solutions they may not currently be aware of. It's about shortening the learning curve, rather than eliminating it.
A 1-year programmer is not going to read themselves into a 20-year programmer, no matter what they read. But at the 5 year level I think you'll see a lot of difference between someone who never considers their craft and never pushes themselves, just keeps their heads down and doing the next bug, and the person who has even just occasionally read this sort of post, pondered how it may apply to their current situation, and taken out of it what they can... which may be something completely different than what they take out if they read the exact same post two years later.
> - The users don't care about the tech. They care about the result.
Seasoned, grownup engineers and tech business leaders are forgetting this, even today. Users don't care that your product is made with AI, but techies just will not shut up about Generative AI, LLMs, Transformers and all this shit that should be implementation details. Users don't care about any of what goes into the sausage.
I don't understand why investors care, either. Product A is made with traditional algorithms, product B is made with AI and LLMs, product C is made with literal magic and wizardry. But they all do exactly the same thing. Why does an investor prefer to invest in product B?
I like this but I think all advice suffers from the problem of the people needing it being incapable of using it and the people who are capable of using it already know it.
You learn by doing x time. There's also a place for study, reading, etc. to supplement this.
This is true in software development. It's true in martial arts. It's true in chess. (the 3 things I've invested some effort in progressing in). I've taught martial arts but there's no magic advice I can give people that will instantly take them to the next level, they need to experience and work through things to progress. Having a teacher helps (a lot).
Fine advice (better than OP). This works till you realize you're burned out, plus didn't get that promo, instead going to that bikeshedding worthless co-worker who's clearly less deserving.
Motivation, productivity go out the window. You don't give a sh*t about the work, team, code quality, made up deadlines or real issues.
Real maturity comes now. You learn to appreciate your limits, see what really matters despite what's being said and observe who make the calls and what their agenda is.
And finally, when you've picked a path and it seems like it works, you'll find equilibrium, somewhere between an idealistic comment on a PR and your next interview.
Till something else like health or kids rock the boat.
Take it easy, you matter more to you than your work. 15 years on, only ones who will remember the weekends you spent at work are your family and kids. Get out more, find solace in the varied arts. Enjoy you painless body till it lasts. Make the weekdays show up between weekends not the other way. Laugh more. Eat well. Feel your surroundings. Make friends, keep friends. Make memories. Don't throw your life away chasing menial pursuits.
If a piece of work is really that substantial and meaningful, it will show itself as such. Don't go looking for it.
Most of this advice has already clicked with me, but this part:
> If you can't easily explain why something is difficult, then it's incidental complexity, which is probably worth addressing
That's a real eye-opener. Hope I remember this next time I implement something complex.
The thing is, I don't think this stuff would've helped me much when I was a junior dev. A lot of them are just too nuanced.
'Addressing' can be ambiguous. Writing comments to explain why it's right and needs to be this way? Or changing the code without considering if they have enough knowledge of the rest of the codebase.
There's also a chance it's not understood yet to easily explain it.
A key part of joining a team is remaining open to the fact that just because I don't understand something, doesn't mean there isn't understanding in it.
Sometimes there's understanding, sometimes the person who did understand has left the team. This is why it's worth writing good commit messages and going back to read them when there's a funny piece of code. A lot of my commit messages are to the effect of "I did it this way to be backwards compatible with how it always worked. It still looks funny to me though"
Then at least future person knows I don't have a good reason for it and they can dig deeper or get PMs to sign off on something better if they want.
Reading code is harder than writing it, and too many people don't have the patience for it.
Instead, in cases like these they'd rather invest the time re-learning the lessons of the past because they thought it should be a certain way, and if they can't instantly see a reason why it's done a certain way, then they behave (unintentionally or otherwise) as if understanding is not possible.
Especially where it's tricky it's worth including a larger pre-amble in the code itself, as well as source code control.
Often one or the other are suffering, and it's sometimes worth giving a description of how the components work together in detail, or are relied up on by more than one use case (therefore not a simple change or solution).
Nobody could easily explain why early (1950s - 1970s) fusion reactors were so difficult. It must have been that everyone was stupid and creating their own problems.
No one can explain why the Collatz Conjecture is so hard to prove. It must be because everyone is so stupid and is adding incidental complexity to the problem. Why is everyone so stupid?
Or, maybe, possibly, that advice is utter bullshit nonsense and some problems are just hard for reasons that we don't fully understand. Go write a reliable reactive library that works how people expect. Oh, you can't? Better explain exactly why it's so hard then, or it's obviously all your fault.
That advice is so utterly at odds with actual reality that it makes it clear this author should be completely ignored.
A sign of a good educator is to - while it may take a lot of time and a lot of building up of base concepts - enlighten people through plain language the complexities of their topic.
Oftentimes, research papers and similar are written for an audience that is already deeply ingrained in their representative topic, such that those first principles aren't discussed anymore, so perhaps that's where you're coming from when you speak about how "no one can explain ..."; but I would argue that many people can, and have. It may just take a lot of speaking, paper, drawings and similar.
And even if / when there are obscenely difficult things to prove, the chances that the thing the proverbial "you" is working on is one of the incredibly difficult things is very unlikely.
There's a saying that a sign of actually understanding something is being able to describe it to a child. It's also why Thing Explainer and ELI5 exists.
"Good educators" are not solving difficult problems, they're teaching about long-standing solutions to long-understood problems. Long-standing solutions to long-understood problems in software engineering are called "packages" or "libraries" and it usually takes a single command to include them. We're talking about building software, not teaching.
The "explain it to a child" comes after decades of humanity's understanding something. Feynman might have been able to explain quantum mechanics to a child, but quantum mechanics was roughly 90 years old at the time. Nobody would have said Michelson and Morley were introducing "incidental complexity" into their experiment because they couldn't explain why the diffraction pattern looked so weird. What are the odds that they're working on an actual difficult problem!? They must be stupid.
You ignored my comment about fusion power and didn't give any examples of someone actually explaining why the Collatz Conjecture is hard. Nobody knows. The first person who deeply understands why the Collatz Conjecture is so hard is going to be the one who solves it. That's what life is like when you're near the edge of humanity's understanding.
> the chances that the thing the proverbial "you" is working on is one of the incredibly difficult things is very unlikely.
On the contrary. Being on the cutting edge of software engineering is incredibly easy. The discipline is (arguably) about 100 years old. The space of problems is unfathomably vast. Software engineering is applicable to literally any other discipline: pick any one and you should immediately find throngs of unsolved (hard) problems around how to apply software approaches to that problem.
And when anyone in the industry solves a specific problem, that specific problem is largely solved forever, for everyone, for free. So if you're not working on a difficult problem, what are you doing? Copy/pasting what someone else did?
Fusion: Turbulence, plus Naiver-Stokes was known to be hard, plus adding in EM fields, plus the exothermic state change of the nuclei fusing introduces discontinuous jumps in the free energy.
Collatz: if you replace 1 and 3 with arbitrary parameters, Conway showed that this class of problems is undecidable, so any particular instance could be arbitrarily hard.
Genuinely great responses. I'm not sure "it can be generalized to an undecidable problem" really explains why it's hard though, as the 5n + 1 problem is very easy (to prove false) and generalizes just the same. And all good points about Fusion, but my point was that it took some time (months, at the very least) for them to realize that's what was spoiling their machines. Remember that this is all a counterexample to the statement that "If you can't easily explain why something is difficult, then it's incidental complexity" -- meaning all the difficulty you are encountering is your fault (and therefore the actual problem must be easy, and you must be stupid.) That's literally what TFA says.
> If you can’t easily explain why something is difficult, then it’s incidental complexity, which is probably worth addressing.
This one is good and has been following me since I've became a manager. Thanks to you, I know how to apply that objection to "press" people when I feel we're dealing with this kind of complexity.
The one thing I disagree is the thing about editors.
When I began coding I spent hours tweaking my vimrc file and learning all the essentially random shortcuts, and dealing with the absurdities of vimscript. (and debugging vim plugins that broke each other.) It felt like actual work while I was producing nothing.
Now I just open vscode with default settings and I am productive right away. Who cares about editors, vscode is good enough.
But maybe just vim sucks and I should have been playing with emacs all along, I don't know.
I just open vi with default settings and am productive right away. I tried VS Code and couldn't get a "hello world" example to run out of the box after a half hour of trying.
And it's fine. Do what works for you. I just don't think playing with editor endlessly is a well spent time. It makes you feel like you do actual work when you don't. In my opinion.
You only play with the editor endlessly if you’re the type of person to play with the editor endlessly.
My vimrc was set up over the years a bit at a time based on my needs. After several years, I did an overhaul based on what I truly need and don’t need from experience. I rarely change anything in it these days. Maybe a few lines if I start using a new language.
With that stated, most of my vim use is still understanding buffers and motions, which takes no configuration.
These days you can write your neovim config in a real language -- lua. I used a pre-defined config called Lazy that does everything vscode can. Then I started tweaking shortcuts, removing and adding plugins, learning more lua.
Then, after I knew what I wanted, I wrote my config from scratch with only the plugins and settings and keybinds I wanted.
It was a very fun process for me. I enjoy simply using my editor. It makes me want to code. I don't get that same level of joy or customization from vscode -- which is phenomenal software btw.
e: replied to the wrong comment in the chain, but the point stands! Customization is how I got into programming. Crafting a status bar on linux, tweaking colorschemes, etc. Don't hate on the tinkerers just because it's not how you enjoy to work. It can be immensely valuable
> I just open vi with default settings and am productive right away. I tried VS Code and couldn't get a "hello world" example to run out of the box after a half hour of trying.
I take it that you've learned vim (otherwise you wouldn't be "productive right away" - you wouldn't even know how to input text or save a file) whereas you had apparently never tried vscode before. How can that be a fair comparison?
The VS Code launch settings crap is a mess. But like as an editor, it's an editor. You can just use it like that. It works well and has good defaults, you can open a terminal with a hotkey, etc.
Same here. I don't spend most of my time typing but rather thinking. So getting good at vim or getting faster at typing will not make me any better. That's not to say that I'm not in favor of being good at typing. I know my keyboard well enough to touch type also I know some keyboard short cuts specific to vscode but they are intuitive and have a GUI alternative if I don't feel like using them.
Shipping the same code will happen sooner if you have tooling and typing setup to allow for quick iterations, and being quick in making those iterations. Selecting a language/framework that might take more work can slow you down too.
Typing faster does seem to look towards thinking faster, trouble shooting faster.
One place where it can get someone into trouble is typing too fast, faster than it might take to think through and not over look something which might invalidate what you're doing.
I have all hires do typeracer.com. Not for a high or a low score, but seeing the development of the score and understanding the importance of typing fast relating to thinking fast.
Typing code inherently is slower than writing sentences, so if you can type faster sentences in written communication, typing code will be faster, even if it's not at the full typing speed.
I think everyone's got a different threshold for where returns start diminishing sharply. While I'm squarely in the "don't waste time micro-tweaking your editor" camp, there are some little bits of shortcuts and tooling that made me much more fluent at code-editing with very little investment. One example that stands out is the multi-cursor support that Sublime Text popularized (and which I use all the time in vscode now). It eliminates a good 80% of repetitive typing, or symbol refactoring that would have involved clunking through menus in old IDEs, and makes experimentation that much quicker. Feels fundamental, like copy/paste shortcuts which everyone knows now.
Yes, I tend on the side of only tweaking things when I realise I'm repetitively using the same context menus. That's the only point I learn the keyboard shortcuts, or map one, these days. I really haven't had to learn too many shortcuts. The best ones are multi-cursor editing like you've mentioned.
If I have to hunt down a bug or have something with a lot of touch points, yeah, there's probably less typing and more thinking.
But if it's a new mostly self contained feature usually I think for a bit, then get to the point where I know what I need to build and the next step is about shitting out a mountain of code. Being able to go from nothing -> mountain in a fast time is useful.
VSCode can also be used more or less effectively though. There are quite a few shortcuts & tricks that can make you 10x more productive, even if maybe it's not as much as vim.
Multi-cursor on ten lines, sure. You get 10x for some seconds or maybe minutes. I'd expect practically zero people have gotten 3x more productive over a two week period for having advanced knowledge of tips and tricks in their editor. Saying 10x seems a wild exaggeration
I'm grateful that I started learning vim and C at the same time in undergrad. Sure vscode is good enough, but a lot of value is gained tweaking your own tools beyond the tool itself.
Producing something doesn't always need to be the goal. Exploration has value.
I don't think this point is about configuring your editor/environment as much as just knowing it. There's certain features that come up all the time: find file, find class, go to implementation, find references, rename var, etc.
It stuns me how many devs do this stuff manually, when virtually all editors/ides have ways of doing these things.
In some ecosystems, the tooling is just not that good. To use ruby on rails as an example: if you lean into the IDE's expectations for code layout it works ok, but there will always be generated methods and variables with nothing to show but the name—no documentation, no source, no googleable identifier, nothing. In these cases there's nothing to do but pull out the rails (or library) source to try and discern their intention.
In my experience, lisp also has this issue of being very difficult to tool in a general sense, as did aspects of writing c/c++ years ago (maybe recognizing stuff like macro-generated symbols has improved by now).
> lisp also has this issue of being very difficult to tool in a general sense
Most of the common lisp tooling is already present in the language itself. Things like inspect, trace, describe and apropos already gives you the equivalent of most IDEs. I agree with you for some dynamic language and magic methods. It can be hard to trace back the exact function that are being called. But you can always design some tooling for it as long as the code follows the ecosystem convention (Laravel plugin in PhpStorm).
The nice thing about Vim (and other configurable editors) is how easy to mesh existing tooling with the editor itself, without requiring for that extension to be a whole project unto itself.
This is a disadvantage of the more dynamic languages, I find. IDEs for languages like C++ and Java are relatively fast and accurate when jumping to definitions, showing references, performing automated refactorings and so on. I rarely see that same precision for any editor or IDE working with TypeScript, Python, and other very dynamic languages, even with the improvements recently thanks to LSP and generally better tools. When almost anything could theoretically have been patched at runtime — even if doing so would be a terrible idea and every style guide says don’t do it — it’s not safe for tools to make the same kind of assumptions with the more dynamic languages that they can with the more static ones.
100% agree about defaults. I do think it’s critical you actually learn the features of you IDE. Yes being able to type fast has nothing to with programming, but being able to jump to symbols, make refactors, use keyboard shortcuts etc. reduces the gap between hands and brain.
Possibly my all-time favourite XKCD, Is it worth the time?¹, demonstrates two important points. If you only do something very rarely anyway, spending time to automate it won’t have a great ROI. But for things you do moderately often that take a minute or even just a few seconds, you can afford to spend a surprisingly large amount of time optimising them and still get a big pay-off over a time frame measured in years.
I view time spent learning to use my tools efficiently and automating common tasks as a sound investment. Editors are a great example. Sure, I could fire up any of the usual suspects and write code somewhat productively. But in my fully customised editor of choice, I can insert and adapt common code patterns that would take me 10–30 seconds to type out fully in mere moments using templates and macros. I can jump to any position visible on-screen with around three keystrokes. I can see syntax highlighting for various file types that I use all the time, including some unusual ones that don’t have definitions readily available, and warnings for several different programming languages in real time as I work.
These save me a few seconds every time I use them, but how many times is that every day? The effort to set them up has probably repaid itself 100x over by now! And the lack of latency is also a qualitative improvement since it means once I’m ready to start writing, I can do so roughly as fast as I can think, instead of constantly being held back and interrupted for a moment.
> If you only do something very rarely anyway, spending time to automate it won’t have a great ROI
For code-editing, maybe. But in general software engineering, there are tasks that I have to do maybe once a year or less that are always way more painful than they need to be because I don't remember the details, and anytime I automate even part of them (or yes, just document a little better), it turns out to be well worth it. Stuff like bootstrapping new environments, some database-related work, etc.
I’m a big fan of writing things down for future reference. Keeping a journal with things like exactly which commands I used, or transferring that knowledge to something like a README or a wiki page for others to see, is a low effort but sometimes high reward habit. Even for something done only rarely, saving a lot of time next time can justify the documentation effort.
Similarly, if the automation effort is literally only copying those commands verbatim into a shell script or similar, that’s often time well spent. The trap, as the XKCD helpfully demonstrates, is when you then start spending significantly longer on making that script more “flexible”. Turning hard coded values into parameters. Looking context up from the environment. On multiple development platforms. In staging, UAT and production deployments too. With detailed help text to explain it all and say what the defaults for everything are if you don’t specify them explicitly. And then it turns out that you only actually run that script once every six months anyway and never use 95% of that extra flexibility anyway…
I have that XKCD printed out and posted on my cubicle wall. I refer to it quite frequently as justification as to why some requested feature/etc. should NOT be implemented.
I use it so much that many non-technical people at my company have started adopting it as well.
I've been using neovim pretty much since I began programming. I had some initial setup and have tweaked it once or twice, but it's by no means been a source of sunk time for me. I recently had to switch to VSCode and had a lot of issues learning all the out of the box key bindings and not having the telescope and fuzzy find windows I was used to was a huge productivity loss for me. When I able to jump back to neovim it was such a huge relief.
I'm reasonably familiar with both and use them for different cases than each other. vscode at least is my text editor in the gui... but consistently growing :)
The time getting up to speed on something new shouldn't be used against something else that is already familiar.
Now if the new tool (ie., vscode) could import your neovim setup and get you an equivalent, that would be neat.
Comparing the sunk cost time for neovim vs vscode might be a more interesting comparison. Neovim definitely is a little more polished than vim out of the box so it's entirely plausible it's pretty close.
If a vim user wanted to learn something like vscode, they could just install a vim extension in vscode to navigate most of vscode in vim keys. Lots on youtube.
It's hard to use a web browser without a vim extension installed.
It's true if you have sunk the time to super customize a setup for you in anything (currently I'm working through my Aerospace setup), the part that I get caught by sometimes is that something I had to customize in neovim, might already be built into vscode, or something else, or vice versa.
Repetitions is about creating the muscle memory, similar to building the same keys with vim.
I'm lucky I was able to get to a place of comfort with vim when I had lots of free time.
It's great when I don't have my setup with me, but my setup with neovim was a step forward, but vscode still sometimes just ends up where I am, and I figure out how to get it behaving a little better.
I guess my point was that, often times people say some tool (or language, or anything) is easier when what they really mean is that they are used to it.
The fact is, you use something because it resonates with you (or its the only resource you have) and then you get good at it, and then everything else becomes a handicap
I wish there existed an editor/IDE which would consist of a good editing model, good settings, programmability and extensive plugin system, but without arcane controls and language.
Basically take vim, remap keys so it becomes ms++, replace vimscript with ts, add optional regular vscode around it. This editor would kill them all, imo.
I’m on the fence. When I obsessed more about Flow State, I really did see dividends from spending spare cycles reading the docs (especially release notes) and keyboard shortcut mappings to find faster ways to do things I did often.
Maybe I hit the point of diminishing returns. Maybe it’s because I don’t Code Like Hell much anymore, but it’s a habit I got out of. I keep resolving to getting back to it, but the fact I don’t maybe telling me something. Like maybe I’ve filled that spare time with other things that are more valuable.
The author claims that a strong knowledge of development tools is a good indicator of general proficiency, but what if it's that learning how to use your editor makes you proficient? After all, nobody gets familiar with Emacs without inadvertently becoming a practitioner - if not an enthusiast - of Lisp!
That happened with me, and I'm now a big fan of functional programming, Clojure, Scheme etc., and that experience and exposure to a particular way of writing code has undoubtedly made me a better programmer. It sounds like you've had a similar experience with Vimscript :)
It’s a little bit like some productivity blogs obsessing over the notepad and pens and apps they use and spending a lot of time obsessing/procrastinating over things that don’t matter. I still learn the same shortcuts I’ve used for the last 15 years in whatever editor I use but most of my time is spent thinking and talking about problems rather than just typing things in the editor
As with most things there's a balance. I've found if I try to learn one or two new features or shortcuts every couple weeks with the tools I use extensively, eventually I do get a sharp axe.
Typing speed is nice but not an important metric I don't think. It's like someone speeding past you on the road only for you to see them at the next red light. Other things will slow you down like code review or needing to wait until a set time to deploy. Also things like copilot and advanced auto-completion are making it less relevant.
Now I just open vscode with default settings and I am productive right away.
I am right away distracted by a huge rectangle around cursor line, current word highlighting and general jumpiness of everything. Coding in vscode feels like writing a book in the middle of brazilian festival. And then it cannot do random simple things like proper indents or humane snippets. The amount of work required to unfuck vscode is really comparable to creating .vimrc from scratch.
I'm a novice. I have only used vscode and sublime. Both have a problem when copying code from elsewhere, where the indent will be wrong every time somehow. If that's what you're talking about, is there an editor that will figure it out correctly somehow? Thanks
In Sublime, there is a command to paste while matching the indentation as well. I think it’s Ctrl+Shift+V on most platforms by default. Maybe that’s what you’re looking for?
Vscode cannot indent it properly due to how indent rules (simple regexes) work. Formatters like “prettier” exist but they can only format whole-file. Maybe try that (downside being you’re at its mercy completely).
I guess you are experiencing either auto-indent like that or the partially selected lines problem (when selection doesn’t touch \n at both ends). Vscode and sublime should also have a way to “paste raw” without reindenting. Try copying proper non-partial lines and/or pasting raw at column 0, removing automatic indent and adding yours afterwards. I know it’s tedious but I’ve seen enough people editing in ms-style editors and they all had this issue. Most just accept it as a fact of life.
As a vim user I’m afraid I can’t help much here. My copy-pasting is almost always line-oriented (V-mode copied lines always pasted after current line, regardless where the cursor is) and indent rules are very robust so it rarely indents things wrong. I also have gv> and gv= to re/indent just pasted text. I’d ask on stackoverflow (stackexchange) or on HN which popular editors have whole-line modes or maybe vscode/sublime plugins which can help with that.
i am much junior but from my experience, working with the toolchain of your team is better in the long run than against it.
if you become proficient quickly with whatever your team works with, it is much better overall than making everyone accommodate you. i have worked in teams where getting neovim installed would confuse the IT as they don't get those requests (or may not even know it exists!).
Learning vim teaches you new ways to think about code editing and is useful if you ssh into machines. I think learning in itself has value, and I think even if you go back to vscode it might have taught you how to better navigate, maybe you realized for example that file trees aren’t so efficient and it’s best to navigate via the fuzzy search.
Then there is also the collective argument that letting Microsoft have editor monopoly can be really disastrous.
> But maybe just vim sucks and I should have been playing with emacs all along, I don't know.
Honestly, yes, I think this is it. Ok, vim doesn't suck as a text editor, but it sucks if what you really wanted was VS Code. Emacs, however, doesn't suck. I spend very little time tweaking my Emacs config these days, but when I do it's to fix things VS Code users just have to put up with. I've watched over their shoulders. They put up with a lot.
> Now I just open vscode with default settings and I am productive right away.
Is always so painful to watch how slow average developers are with the tool they use the most. Using mouse/touchpad to scroll/select word is unbelievably slow. Learning and configuring text editor that is customizable enough to not get in a way pays dividends really well.
Programming is mostly reading and navigating through the code with "go to definition" and similar code actions. Just being able to do this really fast would give a huge productivity boost right away.
> Assess the trade-off you’re making between quality and pace
Very important! But also important to distinguish between kinds of quality. Using a sub-optimal algorithm or bad identifiers or copy-pasted code can always be fixed later. But some problems like a bad database schema is much harder to fix later.
I’d say be very skeptical of this advice if you’re a beginner. You almost certainly aren’t qualified to identify quality so don’t intentionally leave any on the table
This is interesting. If I were to complement the list, these are some items I’d add that helped me:
- learn functional programming. Doing that was how I finally “grokked” programming and could solve problems more easily. Before I was ready to give up.
- learn CS history. I studied UX and what I learnt was mostly one side of how to think about computing where you spoon feed users and remove things. There are other ways to conceptualize software and design, which would have left me less disillusioned.
- learn fundamentals: data structures, networking, performance, operating systems, security, unix, math. These are so neglected in the industry, and we’re left with super complex systems we don’t actually understand as a result.
"learn fundamentals: data structures, networking, performance, operating systems, security, unix, math. These are so neglected in the industry, and we’re left with super complex systems we don’t actually understand as a result."
I sometimes hold a "command line fundamentals" course for my teams. Just being able to understand the basics puts them above any team that doesn't.
You have to know the ground you're building on. Otherwise even your foundation will be faulty.
Sure thing! This assumes everyone has macOS, but isn't very mac-specific.
* Introduction
Quick history of Unix
* When you log in
Difference between login and interactive shells
System shell files vs user shell files
.zshenv for environment variables like PATH, EDITOR, and PAGER
.zprofile for login shells, we don't use it
.zshrc for interactive shells
Your login files are scripts, and can have anything in them
* Moving Around
** Where am I?
pwd = "print working directory"
stored in variable $PWD
Confusingly, also called current working directory, so you may see CWD or cwd mentioned
** What is here?
ls
ls -al
ls -alt
. prefix to filenames makes them hidden
. is also the current directory!
.. means the parent directory
** Finding my way around
cd
cd -
dirs | sed -e $'s/ /\\\n/g'
** Getting Help From The Man
man 1 zshbuiltins
manpage sections
** PATH
echo $PATH | sed -e $'s/:/\\\n/g'
zshenv PATH setting
** Environment Variables
env | sort
EDITOR variable
** History
ctrl-r vs up arrow
** Permissions
Making something executable
** Prompts
zsh promptinit
zsh prompt -l
** Pipes
Iterate to show how pipes work
cat ~/.zshrc | grep PATH
** Commands
*** BSD vs GNU commands
BSD are supplied by Apple, and Apple often uses old versions
GNU are installed via homebrew, and match those commands available in Linux
> You have to know the ground you're building on. Otherwise even your foundation will be faulty
Exactly, well put. Knowing the fundamentals leads to a more solid, safer foundation as you understand better what you can do with what you have and avoid layers of abstraction.
Not sure what 15 years means, but if that’s where I started:
(Blasphemies warning)
- Skip low level and go as high as you can. Ditch C, assembly, hardware. Take python, ruby, js. Never touched C++ cause it’s awful? Good.
- All the money is in the hands of a client. If you’re taking it from someone else, best case you’re taking less than a quarter for essentially the same job. Useless leeches everywhere who perceive you as a magic money generator. Go around them once you’re confident.
- Read sicp, htdp, taoup, but don’t take them to heart. Seriously, extensively test any idea you learn against reality, cause ideas may sound excellent out of context.
- Pragmatic is the only way. Don’t listen to industry peasants jumping through imaginary hoops, they are crazy. Religion in programming is worse than religion irl. There’s insane amount of it in programming. Don’t argue, let them be. It’s actually easy to test approaches by yourself and learn your own ways. Most of these are just habits making no difference once learned.
- Doing something together only turns out faster if you’re very good communicators. Good communicators are rare and you are not the one.
> All the money is in the hands of a client. If you’re taking it from someone else, best case you’re taking less than a quarter for essentially the same job. Useless leeches everywhere who perceive you as a magic money generator. Go around them once you’re confident.
My better advice would be to avoid work that involves outside clients unless you own your own shop or freelance for yourself. They're always going to be price and deadline sensitive in a way that makes for shittier work conditions and lower pay. Working on well-funded internal projects that generate revenue is the way to go. Your work conditions are also going to be much nicer because a client being an asshole can't be used as an excuse for constant overtime or other BS.
>Skip low level and go as high as you can. Ditch C, assembly, hardware. Take python, ruby, js. Never touched C++ cause it’s awful? Good.
In my personal experience, I think that having to work on different high AND more low level programming languages over time is what sets me apart from people/coworkers with low basic knowledge of computer foundations (the order in my case being Basic, Pascal, C, Perl, C/C++, PHP, JS, etc). Also it depends on the projects, but just my two cents.
I think you’re right, I should have separated it in two parts. First is learning low-level “how it works, exactly” — that is useful. Second is writing actual low-level code — that I’d advise myself against.
So, I guess:
- Learn how sockets, filesystems and processes work, read apue, learn how jits/compilers work. But don’t write serious C, that’s a waste of time even if you’re smart enough.
I’m glad you’ve found a happy place in your work, but maybe take some of your own advice, drop this bit of your religion and change this to “use the right tool for the job”.
If you’re doing business logic in a micro service, C is probably the wrong choice. Python, Ruby, whatever probably makes more sense.
But if we are trying to build a (very) high-performance trading engine, or a device driver for a new piece of hardware we are building, and you are on my team and you show up with something in TypeScript because “it’s better”, we are going to need a conversation about whether you’re in the right place. I love high level languages but I also know the problem sets when not to use it.
I also know that if I am going to be writing something very low level, it’s a conscious choice and I’m going to invest in the tooling and the time to make sure my code is safe and secure.
The more generic advice might be that if you love a particular language or approach you might want to work on problems where that language is a good fit. Likewise, if you like particular problems, you should learn the language others think are a good fit in that problem space.
This is the first time I’ve ever seen taoup ever recommended besides when I try to bring it up. It’s full of collective knowledge that’s foundational, along with Steven’s books, (if you want to work in Linux land), yet it’s never mentioned :(
this is not on the content but on the website color choices for text and background: wonderfully well done, easy to read in dark mode. with this kind of sensible coloring i feel i don’t need eink devices for reading.
This was great. I particularly enjoyed the "Bad code gives you feedback, perfect code doesn’t. Err on the side of writing bad code" section, I'd never thought about how spending time writing "perfect" code means you don't get to figure out which aspects of that code actually matter.
I like this a lot. While those commenters who say it is too advanced for novices have a point, I feel these are still issues worth thinking about - and coming back to - as they learn.
The one exception is “Bad code gives you feedback, perfect code doesn’t. Err on the side of writing bad code.” My experience with bad code is that it does not tell me much; instead, it presents inscrutable and baffling mysteries. From the caveats and examples, I think the author is trying to say something rather different; something like “don’t obsess over completeness” or other concepts of ideal software - especially, I might add, dogmatic concepts of this nature.
I think you got the wrong idea from that sentence; the author is talking more about the feedback you get from your end-users and/or customers, rather than development feedback a la debug info.
No I actually do mean the feedback you get as a developer when you realize which “bad” decisions actually end up being tech debt and which were totally fine and just saved time
The Bad Code advice is for people who have the tendency for perfectionism (apparently like the author 15y ago).
My advice to me: write simple, dumb, repetitive code, but make an effort to be consistent.
Consistent, dumb code is really easy to visually parse. The repetitions slide into the background and the differences stick out. The more code there is, the easier to see what goes together (data aggregates/structures), what the overall flow should be (control, conditionals, order etc.). Then it's easier to factor things out or how to express more clearly what the code does.
But again, consistency is key. The code can be "bad", but it's very beneficial to do simple edits like renaming and reordering of things, grouping them under succinct comments so the structure stays clear.
Younger me would do the opposite: inconsistent code that is complex. I think its impatience that gets in the way, rushing to build abstractions and factoring out things way too early or trying out competing ways to express the same kinds of things (playing around with code).
Every programmer has their own tendencies and their own journey, so they need to hear different advice, in order to ignore it for now, but to have an AHA moment at some point when they realize why the advice was given.
I think this could have been phrased as, err on the side of shipping faster.
i.e. in most cases the cost of bugs is so low that the benefits of a faster feedback loop greatly outweigh the benefits of a more rigorous development process.
(On top of that, the number of defects asymptotically approaches zero with greater effort invested, i.e. diminishing returns.)
Agreed. "Perfect" code doesn't exist, but good code is written with the understanding that bugs happen, and exposes information at the right level for people to get the feedback they need, when they need it.
> Spending time sharpening the axe is almost always worth it
I hear this advice constantly. But I feel like it really need to be qualified. I have been in projects that never saw any real value produced because people constantly focused on DX, projects management, and other tools.
Sometimes one just needs to do with tools at hand, and not worry if there are better tools.
Learn to become comfortable implementing and working on state machines.
It might be one of the most useful abstractions out there that if implemented well leads to a great extensible design without compromising performance.
Mostly pretty sound advice here, but oh, this rankles! --
"In situations where bugs aren’t mission critical (ex. 99% of web apps), you’re going to get further with shipping fast and fixing bugs fast, than taking the time to make sure you’re shipping pristine features on your first try."
In 99% of web apps, your end users have no possible way of telling you that you shipped a bug, and your bug will remain there forever, frustrating users and losing your client money as they abandon your site. Telemetry won't help you either becuase you'll misunderstand the observations it provides.
> If you (or your team) are shooting yourselves in the foot constantly, fix the gun
This is a reasonable step to take if you're working in a reasonable company with reasonable "management". If you're unlucky enough to be working under a manager that refuses to recognize there is a problem, even with compelling data, prepare for pain. In that case nothing is allowed to be done because "well, that's how we've always done it and doing something about it might upset someone". It's really time to quit and find something new at that point because there is generally no hope of turning that around.
> If you can’t easily explain why something is difficult, then it’s incidental complexity, which is probably worth addressing
See above. "That's how we've always done it, and someone is used to that work flow, so we can't change it." Sometimes this will lead to a legitimate Chesterton's Fence situation, in which case you'd want to regroup and rethink it. Otherwise, the batshit insane is the default. Target the latter.
> Try to solve bugs one layer deeper
"There is no time to do that! We have so many other projects!" While being blind to the fact (or better yet, burying their head in the sand and trying to ignore) that it's all self-induced.
> Don’t underestimate the value of digging into history to investigate some bugs
"We can't change that because one of our people close to management made that decision and it might drive them away." Good, get them out, that's the only way to really fix it. Ideally if the management that hired them is pushed out too.
> When working on a team, you should usually ask the question
"Who are you to question our wisdom? Who are you to suggest something else, it's working in production."
These are great things to start conversations, but you have to start them in an organization that is willing to change and be willing to take your advice to heart. Don't waste your precious life trying to change what can't be change, no matter how much you want to see it change.
> "It’s really easy to write terrible code. But it’s also really easy to write code that follows absolutely every best practice, with 100% test coverage, and has been fuzz-tested and mutation-tested for good measure – your startup will just run out of money before you finish. So a lot of programming is figuring out the balance."
The situation appears to be changing rapidly. 15 years ago there were no advanced code-generating LLMs, and while relying on them for the core logic without having a good understanding of the language and system is still a bit iffy, it does seem that they're pretty good for generating tests and spotting deviations from whatever your group has decided are best practices.
What sort of tests does it seem good at generating? I was hoping it would give me a decent start writing tests with Jest and React Testing Library for an already existing but untested codebase. It seemed ok at small components that I can also quickly write tests for, but the larger, messier things would receive useless, failing tests.
The point that resonated most with me, and that I repeat every time someone early in their career asks for advice (or one thing I wish I had been told when I first started out) is "When working on a team, you should usually ask the question".
Early in my career, I spent a lot of time reading unclear or obsolete documentation, poring over code, etc, when I could have asked the person sitting next to me and gotten an answer in 5 minutes. Sometimes, I didn't even know who the right person to ask was, but if I had just asked my tech lead, he would have been able to point me in the right direction.
Claiming that it reduces your overall time from several hours to < 5 minutes is maybe a bit exaggerated, since it also takes time to figure out what question you want to ask in the first place. But it's still worthwhile even if you end up investigating for 30 minutes instead of 3 hours.
On the other hand, though, I really wish people would take at least ten minutes to do that digging through the code or reading through the documentation before asking.
One, reading code in general is a good thing to do when you're new, and it's a skill that should be built up in general.
Two, you might actually find what you're looking for, and answer your own question! It's a nice little boost to realize you're getting self-sufficient.
Three, you'll have a lot more context for me, which will make it easier for me to give you the answer. "Hey, how come this function doesn't return what I expected" is a much harder question to answer than "This function says it returns a list of items a user has accessed in a given workspace, but it's returning an empty list; I've checked my local database and it has a good set of seed data I think, and from looking at the function there's nothing obviously filtering out users, so what's going on?"
At my first job, at a consulting company, they had a good rule, which was basically "show initiative when asking questions". What this boils down to is if you ask how to do X, you should first research and think about it, and then you can say "I think Y is how I do X, am I on the right track?"
A lot of the time, you'll answer your own question getting to Y. If not, it will save the person you're asking a bit of legwork, and show you are not trying to have them do your job for them and have respect for their time.
Agreed, now that I'm the person getting these questions, it's hard to reconcile "it would have saved me a ton of time when I was new to this codebase" with "why can't you do even a basic amount of due diligence to even figure out what you're trying to do". (As I mentioned -- it's still important to figure out what the right question is!)
It's hard to find the right balance, and I agree with the author's framing of it as a spectrum, but I also agree that I at least (and most junior engineers, in my experience) leaned way too far on the side of being, um, self-motivated.
I certainly do notice myself getting frustrated by people asking questions when IMO they should know better, and I'm never sure whether I should be more lenient there or if I should tell them off for not putting in that tiny bit of effort.
It can go both ways. The most common one I've seen is definitely taking too long to ask about something. The other is asking too many trivial questions about things you can look up in minutes or ought to know given your job.
I think the problem is knowing when to ask is like many things a skill that gets better with experience. It isn't obvious especially to newer people. I also think if you're a senior you should be much more forgiving of juniors when it comes to this stuff. We were all there once.
I think asking too slowly tends to be caught more easily in things like standup before it snowballs. Asking too quickly tends to be habitual and seems to do more damage since it's harder to notice when managing projects, and often under the vague auspices of "pairing."
I spent a ridiculous amount of hours tweaking my vim config over the last two decades. I've used vscode the last few years and rarely jump to vim to do some text manipulation (that I usually could do with a raw install)
326 comments
[ 5.3 ms ] story [ 163 ms ] threadYou can only understand them because you lived those situations, which implies experience you don't have.
I would say (specifically to my young self):
- There is no substitute for doing. Less tutorials, more coding.
- Stop being obsessed with quality: you are not at the level where you can provide it yet. Do dirty. Do badly. But ship. Some people will be mad at you, and they are right. But do it anyway. Yes, reboot the servers with users on it. Yes, commit the spaghetti. You'll reach the level you want to avoid doing all this.
- The users don't care about the tech. They care about the result.
- Doing something for the hell of it is worth it. Just make it separate from the point above so they don't conflict.
- Programming is a battle against complexity. Again and again. Put that on a post-it. Make your decisions based on it. You will suck at it, but try.
- You have imposter syndrome because you are an imposter. You are really bad. It's ok, doctors hurt people for years while learning to save them. Don't sweat it. It will come.
- You need faith it will work out. On the long run, you'll get better at this. But in the short term also. You'll find the bug. You'll figure out the solution. It will happen if you keep at it even if it can be frustratingly long and unfair. Even if it doesn't feel like that right now.
- The right team is way more important than the right tech. If you are alone, get in touch with people who will lift you up from time to time.
- don’t read opinions from a list and take it as gospel. Adapt to the jobs you’re in, and you’ll develop your own opinions, but now with experience to explain why. Opinions are formed by getting repeatedly hit with the consequences of your (and other’s) decisions, and everyone just has to take enough hits till the pattern seeking area of your brain takes over.
> - There is no substitute for doing. Less tutorials, more coding.
I was quoting out of similar things years ago at the start of my career, “it should be done this way because X said it” didn’t help me at all.
It feels like insulation against being called too junior, since, you can just wholesale adopt someone else’s list of ideas and you’re good. But making mistakes because you’re new to the career is precisely what forms the base of those opinions in the first place.
For example, if you can’t decide between two data structures or tech, pick one and add a comment:
// I’m not sure
There are roles I've been in where it's only been years later that the true impact of decisions made was actually apparent. I'm glad I hung around long enough to experience that.
I have a terrible programmer friend.
He is better at creating actual products than most people I met because his livelihood is on the line: he makes money only from websites.
So he is not living in the abstract idea of best practices, he had to make it profitable for the last two decades, with little resources to go by on top of that.
And he does.
After 10 years, he is still asking me how to write git commands. And I'm still learning from him to shed a lot of BS from my practices.
I was quite proud of some decisions, but realized the negative long term impact of others. Trying to share that experience and whatever 'wisdom' or 'lessons learned' with others has been its own challenge in some situations, because you can easily come across as "the old person" who doesn't "get it" wrt to current trends. Some issues are evergreen and fundamental, but it's difficult for less experienced people to understand why some of these things are really core. I'm not sure there's much substitute for experience in many cases.
The sad fact is this encapsulates features (ease of development, a framework probably does help you ship faster), adaptability (clean abstractions that are easy to work with), and performance.
Finding that balance is always going to be hard but they’re all important!
But it doesn't matter. They were designed well, they were fun to play, and millions of people enjoyed them.
Also this was just an analogy I know, but doctors definitely don’t hurt people for years while trying to save them, very different profession from ours, if anything doctors earlier in their career have been shown to have better results.
E.g. this tip:
- There is no substitute for doing. Less tutorials, more coding.
is directly addressing a common mistake for absolute beginners. Many beginners will read (or worse yet, watch) loads of coding tutorials while doing little themsves. It is an issue a complete beginner encounters and understands.
Your tip on the other hand:
> If you (or your team) are shooting yourselves in the foot constantly, fix the gun
is addressing people working on medium to large projects with internal tooling. That is not a situation a complete beginner finds themselves in; it's a situation someone who already works in programming for a while finds themselves in.
I wouldn't necessarily say your tips are too abstract; they are simply too high level for a complete beginner.
That is not necessarily a bad thing; perhaps the you of 15 years ago already had the basic understanding necessary to be able to comprehend and make use of your tips.
Typing skills are severely underrated in order to professions and roles adject to our professions like PM.
My experience with doctors, of which there has unfortunately been plenty, is that the older doctors are much better. If I need something delicate done I want a doctor who's 50-65ish. They know what they're doing, they don't screw around, and they've seen enough to know that yes, your case is the common thing they've seen 500 times before, and that this treatment works and that one doesn't.
Most adults are kids in big meat suits, they fake it a lot.
I started to live like I was not completely worthless at 35.
Not saying that to be proud of it, just stating that if you think humanity should do better, the first person you'll judge is you.
It will be glaringly obvious you are not meeting your own standards.
The higher your standard, the longer it will take for you to reach them.
And the way to get there faster is to ignore the shame, and do it anyway. Because if you don't, your growth will be slower, and you will do more damage for longer.
Real life means real consequences.
It will make you more tolerant of others as well. Way more tolerant.
This may be good advice for yourself in the past, and for many people at lots of times, but I'd hesitate to give it as general advice. Reading code others have written should not be understated as way to learn valuable things from fundamental patterns and algorithms to language features and idioms. If you have a job in a team, this may happen anyway, but it's possible to write lots of code without realizing there's a better way.
I read about them and tried to apply them.
That's backward and didn't produce anything good.
I started to understand them by doing it the other way around:
- Coding, solving problems.
- Reading other people's sources.
- Then reinventing half a terrible design pattern.
- Later on, looking at a book: "ohhh, that's what I tried to do" or "ohhh, hence the snippet in that source".
- Now I can discuss with people about the pattern and name my code entities according to that.
Design patterns are a communication tool.
I completely agree with the claim "There is no substitute for doing", and I might even say that code you read without running and tweaking it doesn't count.
To learn how to implement computer programs, read books about it.
Also, "To learn how to implement computer programs, read books about it." contradicts the original comment's "There is no substitute for doing" more than I'm happy with.
You need to start by reading somewhere, or you'll probably not get anywhere, as evidenced by pretty much zero office grunts using web browsers every day having figured out the JS console and application development on their own without reading about it. I recommend books over tutorial texts.
But yeah, as I alluded to, you need to build stuff and learn from other builders.
You'll reach the level you want to avoid doing all this.
What if future you has reached that level and people on your team are shipping spaghetti?
I feel like just emulating what you see in your first few jobs is ideal. (As in, ask coworkers who know the thing you’re working on) It could be great code to be inspired from, or mediocre. Either way you get some input about what decisions result in what outcomes, and what the outcome “feels” like.
And if it comes time to change one or many of those decisions later on in this codebase, the person doing it gets a uniform codebase to work from! Unique abstractions and fixes in random places makes refactoring harder.
Or it's not an obsession.
It's caring.
Very few people can pull off a Steve Job level of nitpicking and actually finish a project.
I certainly couldn't, and that advice is for young me.
What in the medical malpractice?
It is ok to be an imposter.
Rather than it being the place where a practitioner works? ("someone whose regular work has involved a lot of training")
But even without this, the thing is, when you don't know, you will make mistakes.
And when your job is to take care of people, mistakes hurt them.
It's the nature of things.
Last year I had 3 medical errors, one that almost got me killed, by very well-meaning professionals.
Closing your eyes and pretending it doesn't happen is naive at best.
But people have to start somewhere. They can't stay in theory forever. And no amount of preparation will save you from making terrible mistakes.
Since we can't put an experienced doctor behind every intern 24/7, there is no real solution to this for now.
Same for programming.
Individuals can be new to a practice and feel like imposters but we shouldn't be pointing to statistics like this as an example of why it's ok.
I can't believe people think like this.
Even then, the Boeing scandal shows it's not bulletproof.
It's not the same for medicine. There are way more doctors than you can put safety nets. Also, when a plane crashes, it's on the news, it costs money and PR. Much less so with doctors.
The plane industry is not inherently more moral, just more liable. Responsibility in the health industry is way more diluted.
It's not that we WANT to pay the price of people learning. I wish I would not have had the wrong meds given to me last year.
It's just that the system is not currently set up to do it otherwise.
So you can feel guilty and stop moving, and you will not learn. You will not grow. And you will actually hurt people for a longer period because you will still make mistakes.
Or you can own it and accept the inevitable: doing things means having actual consequences on the world.
If you want, you can dedicate your life to change the whole system and make it better. But I don't think anybody in this thread is doing that right now.
I did see a lot of people on their high horses, but doing nothing though, paralyzed because they were looking for a way to never break anything.
Me, for example. For years.
However, normalizing the loss of human life as part of the learning algorithm is psychopathic. There's a time and a place to make mistakes... In mission critical situations that's in training and everywhere else we should aspire for safeguards that prevent the loss of life due to mistakes.
In medicine a lot of these deaths can be prevented through greater care. It's not different than engineering in that respect. The greater problem is decision makers putting profits over life. And this kind of mentality in this thread is just fuel for that kind of behavior. It's gross.
The initial phrase was > doctors hurt people for years while learning to save them
It's then a separate reply from someone else about deaths from errors/malpractice.
So, nobody seems to be expressing the mentality you are, correctly, lambasting (at least so far as I've read in the comments). But, as it is relevant to the lessons we'd all want to pass back to ourselves (in my opinion, it's because we wish to hold ourselves accountable to said lessons for the future), let's address the elephant in the comment.
Normalising deaths, especially the preventable ones, is absolutely not something that anybody here is suggesting (so far as my read of the situation is).
Normalising responsibility, by recognising that we can cause harm and so do something about it, that seems far more in-line with the above comments.
As you say it yourself, there's a time and a place, which is undoubtedly what we hope to foster for those who are younger or at the start of learning any discipline, beyond programming, medicine, or engineering.
Nobody is saying that the death and loss is acceptable and normalised, but rather that in the real world we need a certain presence around the knowledge that they will occur, to a certain degree, regardless. So, we accept responsibility for what we can prevent, and strive to push the frontier of our capacity further with this in mind. For some, that comes from experience, unfortunately. For others, they can start to grapple with the notion in lieu of it through considering the consequences, and the scale of consequence; as the above comments would be evidence of, at least by implication.
These are not the only ways to develop a mindset of responsibility, fortunately, but that is what they can be, even if you find the literal wording to suggest otherwise. I cannot, of course, attest to the "true" feelings of others, but neither can anyone else... But in the spirit of the matter, consider: Your sense of responsibility, in turn, seems receptive to finding the areas by which such thinking can become justification for the very thing it would otherwise prevent, either as a shield purpose-built for the role or co-opted out of convenience. That too becomes integral, as we will always need to avoid complacency, and so must also promote this vigilance to become a healthy part of the whole for a responsible mindset -- lest we become that which we seek to prevent, and all that.
Exactly as you say, there's a greater problem, but this thinking is not necessarily justification for it, and can indeed become another tool to counter it. More responsibility, more mindfulness about our intents, actions, and consequences? That will prove indispensable for us to actually solve the greater problem, so we must appreciate that different paths will be needed to achieve it, after all, there are many different justifications for that problem which will all need to be robustly refuted and shown for what they are. Doing so won't solve the problem, but is rather one of many steps we will need to take.
Regardless, this mindfulness and vigilance about ourselves, as much as about each other, will be self-promoting through the mutual reinforcement of these qualities. If someone must attempt to visualise the staggering scale of consequence as part of developing this, then so be it. In turn, they will eventually grapple with this vigilance as well, as the responsibility behoves them to, else they may end up taking actions and having consequences that are equivalent to the exact mentality you fear, even if they do/did not actually "intend" to do so. The learning never ends, and the mistakes never will, so we must have awareness of the totality of this truth; even if only as best we can manage within our limited abilities.
Is what I was originally replying to, which, at least to me does seem to imply preventative deaths are some kind of training tax during ones 'imposter stage'. Perhaps not the intention of the poster.
Thanks for your thoughtful reply.
The Hippocratic oath is not ‘first, do some harm, as long as you’re learning’.
I think you're likely to reach that level a lot faster by trying and failing than by not trying at all.
Is it because we start adding abstractions before we understand the problems?
Like, there's a certain threshold where keep things too simple makes them too complex, so people start introducing abstractions to reduce complexity. But often it's the wrong abstractions and then we accidentally end up with worse complexity that's harder to unravel.
There must be a term for this waves hand thing?
Edit: gosh typical on phones is hard
From another perspective, this is a form of entropy: There are many more ways to increase complexity than to reduce it. This is also the reason why biological life has been getting more complex over time, the only criterion being if it’s fit to survive and reproduce.
Modern medical education doesn't work this way.
It totally does.
They make grave mistakes all the time. And they hide them. They lie about them.
They have their ego and career on the line.
And they don't have enough resources at their disposal, not enough hours, too many patients, and they are exhausted.
In short, they are humans in a human system.
In fact you are. You're speaking to one.
Your original statement completely ignores the massive amount of barriers put up during your training. From my own experience I can tell you that students are usually quite self aware that they are learning and on a short leash. The most egregious mistakes I've seen were almost always a result of understaffing and gaps stemming from leadership issues coming from the top down. The small mistakes and cut corners are from tired overworked people.
Nobody accepts "hurting people" as a component of training. It's an absurd statement.
Seriously, you might be inexperienced or not know everything, but you should clearly not be an imposter and you should be confident in your ability to improve and understand what you don't understand yet.
Take responsabilities and ownership in what you do, don't get behind the easy excuse that it's too complicated. You are the professional and you are getting paid for this.
> - There is no substitute for doing. Less tutorials, more coding.
I'd rephrase that as "just write something!"
Many times I find myself being the classic example of 'perfect is the enemy of good' - I'll think about the perfect solution, rather than write something *now*, that works, and refactor towards perfect. TDD and all that.
Other things:
- Beware the beta and the boss. If it works, it will invariably get shipped if your manager sees it. Many managers cannot put a value on the future cost of maintaining something that's barely good enough.
- Classic Confucius: "I hear and I forget. I see and I remember. I do and I understand." If you're interested enough to read about some tech/language/framework, write something (again!).
- Learn at least one other language and ecosystem in addition to your main one. Even if it is syntactically similar (C++, Java, C#, for example).
This is a great use for new, standalone open source modules: feel free to experiment with styles or techniques or goals that you wouldn't want to justify. (Or for experiments that you start and then abandon without ever showing anyone.)
For example, when I was making lots of packages to help build out the Racket ecosystem, I'd already made an HTML-writing package, but I felt a craving to do one that used syntax extension at compile time, rather than dynamic s-expressions. So I just did it, as a separate package: https://www.neilvandyke.org/racket/html-template/
I don't recall anyone saying they saw value in it, but I like it, and I scratched that itch, and added another skill to my programming mental bag of tricks.
Consider this just a rant from an older programmer who hasn't fully recognized that 'programming' is now largely a social endeavor with online info for everything, language and library ecosystems, etc. I wonder how much of this information I would have internalized if it were all available in my time. Seems like the kind of thing I might read and nod in agreement and forget to apply when relevant without some hard earned run-ins (which is how I'd picked them up).
As for actual programming advice, the thing I'd highlight is to look at the data first and foremost. It goes from initial conditions to post conditions. The differences are what your program/function does, but both the pre/post conditions should be able to be fully described as a valid static state. Once you understand that, the problem, the code is largely plumbing with a small part that applies the functional transformation. There's so much focus on the form and style of "the code" that seems to consider it the main thing rather than an artifact of getting the main thing done: think any time there seems to be fancy abstractions that don't provide value to offset its complexity. To relate it to a point in the post, if you can't connect the difficulty you're having with the logical transformation that needs to be done (e.g. from database state to state), it's likely self-inflicted (or it could be due to a poor choice of database schema). Similarly for poor choices of request/response formats--basically bad plumbing between systems (and not intrinsically hard because of the information being handled).
I have come to the conclusion that the use of these sorts of posts is not that the reader, young or otherwise, will instantly and correctly apply all the lessons to their lives.
It's more about sensitizing people to problems they may not currently see, and solutions they may not currently be aware of. It's about shortening the learning curve, rather than eliminating it.
A 1-year programmer is not going to read themselves into a 20-year programmer, no matter what they read. But at the 5 year level I think you'll see a lot of difference between someone who never considers their craft and never pushes themselves, just keeps their heads down and doing the next bug, and the person who has even just occasionally read this sort of post, pondered how it may apply to their current situation, and taken out of it what they can... which may be something completely different than what they take out if they read the exact same post two years later.
Seasoned, grownup engineers and tech business leaders are forgetting this, even today. Users don't care that your product is made with AI, but techies just will not shut up about Generative AI, LLMs, Transformers and all this shit that should be implementation details. Users don't care about any of what goes into the sausage.
You learn by doing x time. There's also a place for study, reading, etc. to supplement this.
This is true in software development. It's true in martial arts. It's true in chess. (the 3 things I've invested some effort in progressing in). I've taught martial arts but there's no magic advice I can give people that will instantly take them to the next level, they need to experience and work through things to progress. Having a teacher helps (a lot).
Motivation, productivity go out the window. You don't give a sh*t about the work, team, code quality, made up deadlines or real issues.
Real maturity comes now. You learn to appreciate your limits, see what really matters despite what's being said and observe who make the calls and what their agenda is.
And finally, when you've picked a path and it seems like it works, you'll find equilibrium, somewhere between an idealistic comment on a PR and your next interview.
Till something else like health or kids rock the boat.
Take it easy, you matter more to you than your work. 15 years on, only ones who will remember the weekends you spent at work are your family and kids. Get out more, find solace in the varied arts. Enjoy you painless body till it lasts. Make the weekdays show up between weekends not the other way. Laugh more. Eat well. Feel your surroundings. Make friends, keep friends. Make memories. Don't throw your life away chasing menial pursuits. If a piece of work is really that substantial and meaningful, it will show itself as such. Don't go looking for it.
> If you can't easily explain why something is difficult, then it's incidental complexity, which is probably worth addressing
That's a real eye-opener. Hope I remember this next time I implement something complex. The thing is, I don't think this stuff would've helped me much when I was a junior dev. A lot of them are just too nuanced.
There's also a chance it's not understood yet to easily explain it.
A key part of joining a team is remaining open to the fact that just because I don't understand something, doesn't mean there isn't understanding in it.
Reading code is harder than writing it, and too many people don't have the patience for it.
Instead, in cases like these they'd rather invest the time re-learning the lessons of the past because they thought it should be a certain way, and if they can't instantly see a reason why it's done a certain way, then they behave (unintentionally or otherwise) as if understanding is not possible.
Especially where it's tricky it's worth including a larger pre-amble in the code itself, as well as source code control.
Often one or the other are suffering, and it's sometimes worth giving a description of how the components work together in detail, or are relied up on by more than one use case (therefore not a simple change or solution).
No one can explain why the Collatz Conjecture is so hard to prove. It must be because everyone is so stupid and is adding incidental complexity to the problem. Why is everyone so stupid?
Or, maybe, possibly, that advice is utter bullshit nonsense and some problems are just hard for reasons that we don't fully understand. Go write a reliable reactive library that works how people expect. Oh, you can't? Better explain exactly why it's so hard then, or it's obviously all your fault.
That advice is so utterly at odds with actual reality that it makes it clear this author should be completely ignored.
Oftentimes, research papers and similar are written for an audience that is already deeply ingrained in their representative topic, such that those first principles aren't discussed anymore, so perhaps that's where you're coming from when you speak about how "no one can explain ..."; but I would argue that many people can, and have. It may just take a lot of speaking, paper, drawings and similar.
And even if / when there are obscenely difficult things to prove, the chances that the thing the proverbial "you" is working on is one of the incredibly difficult things is very unlikely.
There's a saying that a sign of actually understanding something is being able to describe it to a child. It's also why Thing Explainer and ELI5 exists.
The "explain it to a child" comes after decades of humanity's understanding something. Feynman might have been able to explain quantum mechanics to a child, but quantum mechanics was roughly 90 years old at the time. Nobody would have said Michelson and Morley were introducing "incidental complexity" into their experiment because they couldn't explain why the diffraction pattern looked so weird. What are the odds that they're working on an actual difficult problem!? They must be stupid.
You ignored my comment about fusion power and didn't give any examples of someone actually explaining why the Collatz Conjecture is hard. Nobody knows. The first person who deeply understands why the Collatz Conjecture is so hard is going to be the one who solves it. That's what life is like when you're near the edge of humanity's understanding.
> the chances that the thing the proverbial "you" is working on is one of the incredibly difficult things is very unlikely.
On the contrary. Being on the cutting edge of software engineering is incredibly easy. The discipline is (arguably) about 100 years old. The space of problems is unfathomably vast. Software engineering is applicable to literally any other discipline: pick any one and you should immediately find throngs of unsolved (hard) problems around how to apply software approaches to that problem.
And when anyone in the industry solves a specific problem, that specific problem is largely solved forever, for everyone, for free. So if you're not working on a difficult problem, what are you doing? Copy/pasting what someone else did?
Collatz: if you replace 1 and 3 with arbitrary parameters, Conway showed that this class of problems is undecidable, so any particular instance could be arbitrarily hard.
Hah! I agree with this. :)
This one is good and has been following me since I've became a manager. Thanks to you, I know how to apply that objection to "press" people when I feel we're dealing with this kind of complexity.
When I began coding I spent hours tweaking my vimrc file and learning all the essentially random shortcuts, and dealing with the absurdities of vimscript. (and debugging vim plugins that broke each other.) It felt like actual work while I was producing nothing.
Now I just open vscode with default settings and I am productive right away. Who cares about editors, vscode is good enough.
But maybe just vim sucks and I should have been playing with emacs all along, I don't know.
Who cares about other editors, vi is good enough.
My vimrc was set up over the years a bit at a time based on my needs. After several years, I did an overhaul based on what I truly need and don’t need from experience. I rarely change anything in it these days. Maybe a few lines if I start using a new language.
With that stated, most of my vim use is still understanding buffers and motions, which takes no configuration.
Then, after I knew what I wanted, I wrote my config from scratch with only the plugins and settings and keybinds I wanted.
It was a very fun process for me. I enjoy simply using my editor. It makes me want to code. I don't get that same level of joy or customization from vscode -- which is phenomenal software btw.
e: replied to the wrong comment in the chain, but the point stands! Customization is how I got into programming. Crafting a status bar on linux, tweaking colorschemes, etc. Don't hate on the tinkerers just because it's not how you enjoy to work. It can be immensely valuable
I take it that you've learned vim (otherwise you wouldn't be "productive right away" - you wouldn't even know how to input text or save a file) whereas you had apparently never tried vscode before. How can that be a fair comparison?
i never thought typing speed is all that important when you code. But he DID write crazy fast.
Shipping the same code will happen sooner if you have tooling and typing setup to allow for quick iterations, and being quick in making those iterations. Selecting a language/framework that might take more work can slow you down too.
Typing faster does seem to look towards thinking faster, trouble shooting faster.
One place where it can get someone into trouble is typing too fast, faster than it might take to think through and not over look something which might invalidate what you're doing.
I have all hires do typeracer.com. Not for a high or a low score, but seeing the development of the score and understanding the importance of typing fast relating to thinking fast.
Typing code inherently is slower than writing sentences, so if you can type faster sentences in written communication, typing code will be faster, even if it's not at the full typing speed.
If I have to hunt down a bug or have something with a lot of touch points, yeah, there's probably less typing and more thinking.
But if it's a new mostly self contained feature usually I think for a bit, then get to the point where I know what I need to build and the next step is about shitting out a mountain of code. Being able to go from nothing -> mountain in a fast time is useful.
Producing something doesn't always need to be the goal. Exploration has value.
It stuns me how many devs do this stuff manually, when virtually all editors/ides have ways of doing these things.
In my experience, lisp also has this issue of being very difficult to tool in a general sense, as did aspects of writing c/c++ years ago (maybe recognizing stuff like macro-generated symbols has improved by now).
Most of the common lisp tooling is already present in the language itself. Things like inspect, trace, describe and apropos already gives you the equivalent of most IDEs. I agree with you for some dynamic language and magic methods. It can be hard to trace back the exact function that are being called. But you can always design some tooling for it as long as the code follows the ecosystem convention (Laravel plugin in PhpStorm).
The nice thing about Vim (and other configurable editors) is how easy to mesh existing tooling with the editor itself, without requiring for that extension to be a whole project unto itself.
I view time spent learning to use my tools efficiently and automating common tasks as a sound investment. Editors are a great example. Sure, I could fire up any of the usual suspects and write code somewhat productively. But in my fully customised editor of choice, I can insert and adapt common code patterns that would take me 10–30 seconds to type out fully in mere moments using templates and macros. I can jump to any position visible on-screen with around three keystrokes. I can see syntax highlighting for various file types that I use all the time, including some unusual ones that don’t have definitions readily available, and warnings for several different programming languages in real time as I work.
These save me a few seconds every time I use them, but how many times is that every day? The effort to set them up has probably repaid itself 100x over by now! And the lack of latency is also a qualitative improvement since it means once I’m ready to start writing, I can do so roughly as fast as I can think, instead of constantly being held back and interrupted for a moment.
¹ https://xkcd.com/1205/
For code-editing, maybe. But in general software engineering, there are tasks that I have to do maybe once a year or less that are always way more painful than they need to be because I don't remember the details, and anytime I automate even part of them (or yes, just document a little better), it turns out to be well worth it. Stuff like bootstrapping new environments, some database-related work, etc.
I do this almost always for reproducibility and documentation purposes first.
It's not typing the commands that takes time, it's remembering what commands to type.
Similarly, if the automation effort is literally only copying those commands verbatim into a shell script or similar, that’s often time well spent. The trap, as the XKCD helpfully demonstrates, is when you then start spending significantly longer on making that script more “flexible”. Turning hard coded values into parameters. Looking context up from the environment. On multiple development platforms. In staging, UAT and production deployments too. With detailed help text to explain it all and say what the defaults for everything are if you don’t specify them explicitly. And then it turns out that you only actually run that script once every six months anyway and never use 95% of that extra flexibility anyway…
I use it so much that many non-technical people at my company have started adopting it as well.
The time getting up to speed on something new shouldn't be used against something else that is already familiar.
Now if the new tool (ie., vscode) could import your neovim setup and get you an equivalent, that would be neat.
Comparing the sunk cost time for neovim vs vscode might be a more interesting comparison. Neovim definitely is a little more polished than vim out of the box so it's entirely plausible it's pretty close.
If a vim user wanted to learn something like vscode, they could just install a vim extension in vscode to navigate most of vscode in vim keys. Lots on youtube.
It's hard to use a web browser without a vim extension installed.
It's true if you have sunk the time to super customize a setup for you in anything (currently I'm working through my Aerospace setup), the part that I get caught by sometimes is that something I had to customize in neovim, might already be built into vscode, or something else, or vice versa.
Repetitions is about creating the muscle memory, similar to building the same keys with vim.
I'm lucky I was able to get to a place of comfort with vim when I had lots of free time.
It's great when I don't have my setup with me, but my setup with neovim was a step forward, but vscode still sometimes just ends up where I am, and I figure out how to get it behaving a little better.
The fact is, you use something because it resonates with you (or its the only resource you have) and then you get good at it, and then everything else becomes a handicap
Vim has: a good editing model, good settings, programmability, arcane controls and language.
Ms-style usually has: extensive plugin systems, poor settings, poor editing model.
I wish there existed an editor/IDE which would consist of a good editing model, good settings, programmability and extensive plugin system, but without arcane controls and language.
Basically take vim, remap keys so it becomes ms++, replace vimscript with ts, add optional regular vscode around it. This editor would kill them all, imo.
Maybe I hit the point of diminishing returns. Maybe it’s because I don’t Code Like Hell much anymore, but it’s a habit I got out of. I keep resolving to getting back to it, but the fact I don’t maybe telling me something. Like maybe I’ve filled that spare time with other things that are more valuable.
That happened with me, and I'm now a big fan of functional programming, Clojure, Scheme etc., and that experience and exposure to a particular way of writing code has undoubtedly made me a better programmer. It sounds like you've had a similar experience with Vimscript :)
So you might think, until you meet a fervent Emacs advocate that doesn't know anything about "the underlying Lisp stuff".
Typing speed is nice but not an important metric I don't think. It's like someone speeding past you on the road only for you to see them at the next red light. Other things will slow you down like code review or needing to wait until a set time to deploy. Also things like copilot and advanced auto-completion are making it less relevant.
I am right away distracted by a huge rectangle around cursor line, current word highlighting and general jumpiness of everything. Coding in vscode feels like writing a book in the middle of brazilian festival. And then it cannot do random simple things like proper indents or humane snippets. The amount of work required to unfuck vscode is really comparable to creating .vimrc from scratch.
I guess you are experiencing either auto-indent like that or the partially selected lines problem (when selection doesn’t touch \n at both ends). Vscode and sublime should also have a way to “paste raw” without reindenting. Try copying proper non-partial lines and/or pasting raw at column 0, removing automatic indent and adding yours afterwards. I know it’s tedious but I’ve seen enough people editing in ms-style editors and they all had this issue. Most just accept it as a fact of life.
As a vim user I’m afraid I can’t help much here. My copy-pasting is almost always line-oriented (V-mode copied lines always pasted after current line, regardless where the cursor is) and indent rules are very robust so it rarely indents things wrong. I also have gv> and gv= to re/indent just pasted text. I’d ask on stackoverflow (stackexchange) or on HN which popular editors have whole-line modes or maybe vscode/sublime plugins which can help with that.
i am much junior but from my experience, working with the toolchain of your team is better in the long run than against it.
if you become proficient quickly with whatever your team works with, it is much better overall than making everyone accommodate you. i have worked in teams where getting neovim installed would confuse the IT as they don't get those requests (or may not even know it exists!).
Then there is also the collective argument that letting Microsoft have editor monopoly can be really disastrous.
Honestly, yes, I think this is it. Ok, vim doesn't suck as a text editor, but it sucks if what you really wanted was VS Code. Emacs, however, doesn't suck. I spend very little time tweaking my Emacs config these days, but when I do it's to fix things VS Code users just have to put up with. I've watched over their shoulders. They put up with a lot.
Is always so painful to watch how slow average developers are with the tool they use the most. Using mouse/touchpad to scroll/select word is unbelievably slow. Learning and configuring text editor that is customizable enough to not get in a way pays dividends really well.
Programming is mostly reading and navigating through the code with "go to definition" and similar code actions. Just being able to do this really fast would give a huge productivity boost right away.
To avoid the mess, design with the fail-fast principle in mind, which brings you closer to the spot where an error occurred.
Very important! But also important to distinguish between kinds of quality. Using a sub-optimal algorithm or bad identifiers or copy-pasted code can always be fixed later. But some problems like a bad database schema is much harder to fix later.
- learn functional programming. Doing that was how I finally “grokked” programming and could solve problems more easily. Before I was ready to give up.
- learn CS history. I studied UX and what I learnt was mostly one side of how to think about computing where you spoon feed users and remove things. There are other ways to conceptualize software and design, which would have left me less disillusioned.
- learn fundamentals: data structures, networking, performance, operating systems, security, unix, math. These are so neglected in the industry, and we’re left with super complex systems we don’t actually understand as a result.
I sometimes hold a "command line fundamentals" course for my teams. Just being able to understand the basics puts them above any team that doesn't.
You have to know the ground you're building on. Otherwise even your foundation will be faulty.
Exactly, well put. Knowing the fundamentals leads to a more solid, safer foundation as you understand better what you can do with what you have and avoid layers of abstraction.
(Blasphemies warning)
- Skip low level and go as high as you can. Ditch C, assembly, hardware. Take python, ruby, js. Never touched C++ cause it’s awful? Good.
- All the money is in the hands of a client. If you’re taking it from someone else, best case you’re taking less than a quarter for essentially the same job. Useless leeches everywhere who perceive you as a magic money generator. Go around them once you’re confident.
- Read sicp, htdp, taoup, but don’t take them to heart. Seriously, extensively test any idea you learn against reality, cause ideas may sound excellent out of context.
- Pragmatic is the only way. Don’t listen to industry peasants jumping through imaginary hoops, they are crazy. Religion in programming is worse than religion irl. There’s insane amount of it in programming. Don’t argue, let them be. It’s actually easy to test approaches by yourself and learn your own ways. Most of these are just habits making no difference once learned.
- Doing something together only turns out faster if you’re very good communicators. Good communicators are rare and you are not the one.
My better advice would be to avoid work that involves outside clients unless you own your own shop or freelance for yourself. They're always going to be price and deadline sensitive in a way that makes for shittier work conditions and lower pay. Working on well-funded internal projects that generate revenue is the way to go. Your work conditions are also going to be much nicer because a client being an asshole can't be used as an excuse for constant overtime or other BS.
In my personal experience, I think that having to work on different high AND more low level programming languages over time is what sets me apart from people/coworkers with low basic knowledge of computer foundations (the order in my case being Basic, Pascal, C, Perl, C/C++, PHP, JS, etc). Also it depends on the projects, but just my two cents.
Knowing the low level will prevent you from shooting yourself in the foot with the high level.
So, I guess:
- Learn how sockets, filesystems and processes work, read apue, learn how jits/compilers work. But don’t write serious C, that’s a waste of time even if you’re smart enough.
If you’re doing business logic in a micro service, C is probably the wrong choice. Python, Ruby, whatever probably makes more sense.
But if we are trying to build a (very) high-performance trading engine, or a device driver for a new piece of hardware we are building, and you are on my team and you show up with something in TypeScript because “it’s better”, we are going to need a conversation about whether you’re in the right place. I love high level languages but I also know the problem sets when not to use it.
I also know that if I am going to be writing something very low level, it’s a conscious choice and I’m going to invest in the tooling and the time to make sure my code is safe and secure.
The more generic advice might be that if you love a particular language or approach you might want to work on problems where that language is a good fit. Likewise, if you like particular problems, you should learn the language others think are a good fit in that problem space.
The one exception is “Bad code gives you feedback, perfect code doesn’t. Err on the side of writing bad code.” My experience with bad code is that it does not tell me much; instead, it presents inscrutable and baffling mysteries. From the caveats and examples, I think the author is trying to say something rather different; something like “don’t obsess over completeness” or other concepts of ideal software - especially, I might add, dogmatic concepts of this nature.
My advice to me: write simple, dumb, repetitive code, but make an effort to be consistent.
Consistent, dumb code is really easy to visually parse. The repetitions slide into the background and the differences stick out. The more code there is, the easier to see what goes together (data aggregates/structures), what the overall flow should be (control, conditionals, order etc.). Then it's easier to factor things out or how to express more clearly what the code does.
But again, consistency is key. The code can be "bad", but it's very beneficial to do simple edits like renaming and reordering of things, grouping them under succinct comments so the structure stays clear.
Younger me would do the opposite: inconsistent code that is complex. I think its impatience that gets in the way, rushing to build abstractions and factoring out things way too early or trying out competing ways to express the same kinds of things (playing around with code).
Every programmer has their own tendencies and their own journey, so they need to hear different advice, in order to ignore it for now, but to have an AHA moment at some point when they realize why the advice was given.
i.e. in most cases the cost of bugs is so low that the benefits of a faster feedback loop greatly outweigh the benefits of a more rigorous development process.
(On top of that, the number of defects asymptotically approaches zero with greater effort invested, i.e. diminishing returns.)
I hear this advice constantly. But I feel like it really need to be qualified. I have been in projects that never saw any real value produced because people constantly focused on DX, projects management, and other tools.
Sometimes one just needs to do with tools at hand, and not worry if there are better tools.
Learn to become comfortable implementing and working on state machines.
It might be one of the most useful abstractions out there that if implemented well leads to a great extensible design without compromising performance.
"In situations where bugs aren’t mission critical (ex. 99% of web apps), you’re going to get further with shipping fast and fixing bugs fast, than taking the time to make sure you’re shipping pristine features on your first try."
In 99% of web apps, your end users have no possible way of telling you that you shipped a bug, and your bug will remain there forever, frustrating users and losing your client money as they abandon your site. Telemetry won't help you either becuase you'll misunderstand the observations it provides.
This is a reasonable step to take if you're working in a reasonable company with reasonable "management". If you're unlucky enough to be working under a manager that refuses to recognize there is a problem, even with compelling data, prepare for pain. In that case nothing is allowed to be done because "well, that's how we've always done it and doing something about it might upset someone". It's really time to quit and find something new at that point because there is generally no hope of turning that around.
> If you can’t easily explain why something is difficult, then it’s incidental complexity, which is probably worth addressing
See above. "That's how we've always done it, and someone is used to that work flow, so we can't change it." Sometimes this will lead to a legitimate Chesterton's Fence situation, in which case you'd want to regroup and rethink it. Otherwise, the batshit insane is the default. Target the latter.
> Try to solve bugs one layer deeper
"There is no time to do that! We have so many other projects!" While being blind to the fact (or better yet, burying their head in the sand and trying to ignore) that it's all self-induced.
> Don’t underestimate the value of digging into history to investigate some bugs
"We can't change that because one of our people close to management made that decision and it might drive them away." Good, get them out, that's the only way to really fix it. Ideally if the management that hired them is pushed out too.
> When working on a team, you should usually ask the question
"Who are you to question our wisdom? Who are you to suggest something else, it's working in production."
These are great things to start conversations, but you have to start them in an organization that is willing to change and be willing to take your advice to heart. Don't waste your precious life trying to change what can't be change, no matter how much you want to see it change.
The situation appears to be changing rapidly. 15 years ago there were no advanced code-generating LLMs, and while relying on them for the core logic without having a good understanding of the language and system is still a bit iffy, it does seem that they're pretty good for generating tests and spotting deviations from whatever your group has decided are best practices.
Early in my career, I spent a lot of time reading unclear or obsolete documentation, poring over code, etc, when I could have asked the person sitting next to me and gotten an answer in 5 minutes. Sometimes, I didn't even know who the right person to ask was, but if I had just asked my tech lead, he would have been able to point me in the right direction.
Claiming that it reduces your overall time from several hours to < 5 minutes is maybe a bit exaggerated, since it also takes time to figure out what question you want to ask in the first place. But it's still worthwhile even if you end up investigating for 30 minutes instead of 3 hours.
One, reading code in general is a good thing to do when you're new, and it's a skill that should be built up in general.
Two, you might actually find what you're looking for, and answer your own question! It's a nice little boost to realize you're getting self-sufficient.
Three, you'll have a lot more context for me, which will make it easier for me to give you the answer. "Hey, how come this function doesn't return what I expected" is a much harder question to answer than "This function says it returns a list of items a user has accessed in a given workspace, but it's returning an empty list; I've checked my local database and it has a good set of seed data I think, and from looking at the function there's nothing obviously filtering out users, so what's going on?"
A lot of the time, you'll answer your own question getting to Y. If not, it will save the person you're asking a bit of legwork, and show you are not trying to have them do your job for them and have respect for their time.
It's hard to find the right balance, and I agree with the author's framing of it as a spectrum, but I also agree that I at least (and most junior engineers, in my experience) leaned way too far on the side of being, um, self-motivated.
I certainly do notice myself getting frustrated by people asking questions when IMO they should know better, and I'm never sure whether I should be more lenient there or if I should tell them off for not putting in that tiny bit of effort.
I think the problem is knowing when to ask is like many things a skill that gets better with experience. It isn't obvious especially to newer people. I also think if you're a senior you should be much more forgiving of juniors when it comes to this stuff. We were all there once.
I think asking too slowly tends to be caught more easily in things like standup before it snowballs. Asking too quickly tends to be habitual and seems to do more damage since it's harder to notice when managing projects, and often under the vague auspices of "pairing."