Programming Innocence | Hacker News Reader

Programming Innocence

projectb14ck.org

81 pointsb14ck24 comments

I have always been a "fearless" programmer, but never realized it until reading this post. Here's how:

Fear of not knowing the best way to do things (best practices).

The sooner you realize that there is never a best way of doing anything, the sooner you can release this silly fear. Some ways are better that others, but any way is better than no way. Just get the thing done. Later, when you refactor, you'll have the best of all worlds: code that did the job right away, a better way of doing things, a satisfied customer, and a great learning experience. Woo hoo!

Fear of not using the right tools and languages.

Give me an adjustable wrench, 2 screwdrivers, and a big hammer and I can fix just about anything. Same thing with programming. I'm too busy getting work done to learn every new tool or technique. As I've told many programmers over the years: "Whatever you can do, I can do in BASIC. Maybe not as pretty, but probably just as fast and just as effective."

Fear of errors (especially compiler errors).

You're in the wrong business. Errors are what point you in the right direction. The sooner you learn to embrace errors and use them to refine your work, the sooner you'll become fearless (and better).

Fear of schedules.

"I see only one move ahead, but it is always the correct one." - chess master Jose Raul Capablanca. That's what my schedule looks like. One item. One day. Project managers can't stand this, but then again, I get way more work done than they do.

Fear of publicity (what will other programmers think about this code?).

I never publish my code. Ever. Users get to give me feedback, but I don't give a shit what other programmers think. Sure, I learn from them, but never in the context of reviewing the code I wrote. I learn from the code of others and apply those lessons to my own work.

I don't give a shit what other programmers think. Sure, I learn from them, but never in the context of reviewing the code I wrote.

Can you elaborate on this a little? On the surface, it sounds like ego getting in the way of learning and closing your mind to good mentors. On the other hand, you might also have had a helping hand in inventing Lisp or something, and I am your young whippersnapper that you won't be listening to because you have already written my algorithms better before I learned to crawl. The context helps, since I am far from convinced this is a good attitude for programmers (especially my coworkers) to adopt.

Can you elaborate on this a little?

Let me put it this way...

If you gave me a page of your code and asked for feedback, I would return it with 50 red marks on it.

If you gave it to me a second time, there would still be 50 red marks, but most of them would have you changing it back to the way it was in the first place.

The point is that any programmer can give endless feedback to any other programmer based on a sample of code. In my experience, 90% of that feedback was flat out wrong, and the other 10% was a waste of time.

I prefer learning new things the way I have for years now. When I see something cool, either as a user or by looking at the code, I figure out what was done and how I'd like to do it (rarely keeping 100% of it), and then incorporate that into my own style.

Having others look at my code has never delivered much value. This may seem like ego, but it's not. It's just a matter of what works best.

Having others look at my code has never delivered much value.

Maybe not much value for you, but after enjoying your many insightful comments on this site, I suspect that I would learn a lot by looking at your code. :-)

EDIT: (Ignore this. It's wrong, and I should have been more polite anyway. I'm leaving it to teach myself a lesson about being hasty...)

If you understood his point, you wouldn't say that.

Not that I agree with Ed on this, 95% of my code is published. But still.

He himself says that he learns by looking at others' code. Perhaps I don't understand his point.

... or perhaps I read it too quickly.

Which going back over it, seems to be the case. My bad!

  var TwoBirdsWithOneStone="Thank you, tjr!";

Interesting. In my experience, the feedback

(1) introduces me to new concepts or utilities I had not considered using -- may or may not incorporate

(2) shows me what patterns other developers are accustomed to seeing -- if I chose a pattern for subjective reasons and I am outnumbered, incorporate and adapt

(3) sometimes uncovers bugs or places where I am not fitting into the architecture well enough (or someone else's new development) -- extra eyes, extra attention, all nice

(4) allows me to turn around and show my "reviewers" approaches in #1 and #2 -- educates and changes local culture

(5) introduces other developers to parts of the code they did not write

In short, the code review is not just about me. Although your two code reviews may give me red marks going back and forth, it is ultimately up to me to decide which way to go, and I may benefit from seeing both sides if I have not already seen those. There is a pretty small percentage that I would say is actually completely useless to me. I suppose it depends on how one conducts reviews, and it may also depend on whether your project source is shared with and maintained by others.

When coding fast, nobody knows your code like you do. Reviews are then either a) too late to do any good, or b) shallow. Both are nearly useless.

If you have a mentor and are new, then the mentor's review can help. But in that case, the mentor probably could have written your code faster than you did. The review may be useful to you, but the project would have in fact gone faster had you not been there.

If you code fast, outrun the others around you, are any good at all, then code review is at best annoying.

That's phase 2. Phase 3 is the middle way: be clever about balancing all those forces. Use the new library, but only if your gut feeling tells you that it's also the right choice for the future. Change the code without hesitation, but do hesitate if you think that it may cause more problems than it resolves. Etc.

Experience is a bitch. There are no blacks and whites, only a grey mess.

That's true. I'm assuming that you (the developer) will already be consciously making these decisions as you discover new things, and will only feel the 'fear' if you realize that you do indeed need it.

Thanks for your input :)

Hm. What about "There are neither blacks nor whites, only a lot of black and blue?"

I think if you read a lot of programming forums, it's easy to confuse the fact that there are lots of people with highly specialized knowledge with the idea that everybody else knows more about everything than you do.

Yes, and this is a great force in my betterment. I'll see 4 different people talk about 4 completely different advanced techniques they use, and each of them master them so much that they talk about their feats like it's not really hard at all.

So eventually I tend to forget the context of the whole thing and come to think: "Hey, I can master these 4 advanced techniques myself, after all they're not really hard at all". And I eventually succeed ;P

Actually, programming isn't the only thing with the innocence problem. I think he same thing happens with almost every creative work. From playing music to writing to designing sites, most have experienced similar feelings in other projects. The only solution is to Press on, keep working, and realize that your only way to create great stuff is to keep creating. That's what I keep having to tell myself lately.

Press on ... Don't give up! And try to keep the innocent excitement as long as possible ;)

At first, I though "Man, this sounds like a recipe for technical debt" but then I realized I'm guilty of exactly what you're talking about. In fact, I was browsing hacker news as a break from googling the best way to implement a feature that I could do in 10 minutes if I weren't concerned about having it perfect from the get-go.

I recently realized that I code slower than I used to. I assumed this was because I stopped hacking for a few years during law school, but now I think it has more to do with being immersed in a risk-averse perfectionist environment that leads to exactly the kind of timidity you're talking about. This may be one of the more valuable insights I've gotten from hacker news in a while. Thanks!

Be careful. Sometimes the reason you code slower is that you are thinking from an architectural level and are making sure your code is a reasonable fit. Sometimes you are taking the time to shape the code in a way that is maintainable.

I have felt this loss of what the author calls "programming innocence", but when I dig down deeply, I think it is more that I am doing MORE work than less. When I was young and stupid, I thought I was implementing features, and I whipped things out quickly. Lord do I remember also spending a lot of time debugging those things, and if I came back to the project a few months later, I would find it hard to read, figured I knew better, and would rewrite the whole thing with my newfangled enlightenment. Today, I spend a lot less time debugging and rewriting, and my projects are more feature-complete. It may feel like I am programming slower, but I am still whipping things out, and unlike my young days where I was my own customer, my requirements are constantly changing: I write code that can evolve.

Just yesterday I was thinking about this issue. I think the article takes the biscuit.In my opinion, this is the main reason for why people who start out early are that good.

I partly relearned this skill when I started out with Taekwondo. Desperately, I was trying to memorize the motion sequences. Whenever I asked a question concerning which angle my right leg should have, which direction I should turn to ... the master would always say: "Don't think or talk. Watch & do it!"

Surprisingly, you get a lot better in no time. Don't get me wrong I still suck but at least what I do now is worlds apart from my first humble attempts.

Now I think that I could become a decent programmer with a more child-like attitude (like the one I had when I discovered DOS & Win 3.11 at the age of 5 - no adult taught me anything, I couldn't even understand the English menus ;-))

But unfortunately we get older. Whilst it is never too late to learn something for the joy of it I'd like to have a halfway-decent paid job within the next 10 years. At the moment I am 23. I could quit university, devote myself to open source software & hack the hell out of the day. Hopefully, I'll someday know enough to have paid work. However, if I'd fail I wouldn't even have a degree & my perspectives would be quite bleak. With that in mind programming is only a (way too little) hobby for me.

Very similar blog post: http://cam.ly/blog/2010/12/code-fearlessly/ And associated HN chatter: http://news.ycombinator.com/item?id=1964060

But what if my favorite way of hacking, what makes me feel amazing is something more like Rich Hickey's Hammock Driven Development (clojure.blip.tv/file/4457042/)?

Of course I don't get a chance to do this as much as I'd like at my day job. Still, some of my most satisfying programming experience has been when I've spent days researching and thinking about a problem, had a moment of enlightenment, and wrote 200 lines of beautifully simple code that strikes at the heart of the problem. That makes me feel like a programming god.

And I'd argue the result is somewhat better than "innocently" hammering out reams of code without taking time to really consider it.

The more I read about our profession in general, the more it seems to suck as you build a career / get in your thirties. I really hope it doesn't for me, because I'm not good at much else, lets see.

Your fear is only a result of your use of flawed languages, probably of the Algol family, rather using a flawless language like Lisp.

Fear of errors (especially compiler errors)

That programmer should be really innocent. ;)

Reply on news.ycombinator.com