Hacker News Reader

It's fragile, it's highly specialized, and it's overwhelmingly human-constructed. The fiddly bits have very little wedding in anything else, not even human-constructed hardware. If you're fussing in bits which don't impose their own forms of inertia, e.g., CLI tools whose arguments are exposed to shell scripts and hence, if changed tend to break a great many things which is Generally Not Recommended, then you can find that years or decades of expertise can be wiped out in a very brief time.

Unlearning is generally as hard as or harder than learning.

And there's a strong tendency to view the first system you learned as "the proper way" to do things. Sometimes it is, sometimes it isn't.

The transition you're going through in Perl => Java is a pretty good example. You happened into a skillset early in your career which was highly capable and offered ample opportunities, but its flower has faded. That's something that the kids today who're building up competency in various currently-popular toolsets might do well to consider. They're starting to see hints of that as tools such as RoR are fading. What happens when we, say, bin the entire present set of Web dev, will be interesting.

And it's happened to whole industries before. Sucked to graduate in nuclear engineering in 1979, or petroleum engineering in 1990.

I don't agree that programming expertise is fragile. There is a kernel of knowledge that transcends language and syntax, and that is by far the bigger part of "programming knowledge", much more important than knowing specific syntax and how to use certain tools. For more, see [1].

This doesn't mean you aren't correct. People may still be reacting negatively to change, because it does require work to learn new libraries, frameworks and languages. But that is not the same as saying that programming expertise is fragile.

[1]. "Programmer Knowledge" http://henrikwarne.com/2014/12/15/programmer-knowledge/

I'm going to have to disagree with you, especially in relation to large frameworks.

I spend a lot of time working with medium to large ruby on rails applications. I know them very well and I am pretty good at doing things quickly and correctly. If I have to work on a php, python, or javascript project, it takes me one or two orders of magnitude longer to get things done than it would in a rails application.

The ability to start working with a new language after reading a blog is useful but it doesn't mean you'll be effective and useful in that language in a week.

Here is the limit of what I was saying.

A person with the ability to program, probably will have that raw ability in various languages. There is a knowledge core that is very transferable. An experienced programmer can come up to speed very quickly as concepts get mapped into things they already know. A polyglot will have been through this several times, and won't be afraid of the process.

But there is a great deal of knowledge that we use every day that is environment dependent. A programmer with 20 years of experience in various languages will not be noticeably better by day 5 in a new language than someone who has been programming for 2 months only, but in that language. A simple change can leave a programmer feeling incompetent.

Programmers who respond to this badly are not going to handle such transitions well. Furthermore if they get stuck in defensive behavior, then the initial discomfort will turn into permanent failure.

I never bothered memorizing stuff which is "fragile", e.g. browser quirks. Luckily, there's no shortage of programming knowledge that is the very opposite of fragile. Under a few common assumptions, sorting can't be done faster than O(n log n). All NFAs can be converted to DFAs. 2-SAT is polynomial but 3-SAT is NP-complete. And so on.

People who disdain math as "brain teasers" are deprived of the rush of knowing that some piece of your knowledge will stay true forever.

Sounds good, but this is a bad idea.

You can't get useful work done on real problems without learning a lot of that "fragile" stuff. How are things scheduled on your OS? Where do you find your logs? How do you run unit tests? What are your local coding standards? What is the library call for doing X? How do you find your documentation? What was that section of code you found foo in the other day?

All of this stuff is fragile. None of it is going to last. But it is context for your current life, and you're going to be more productive if you learn it. And when this stuff changes on you, you're going to feel the productivity drop. But there is no sense in hamstringing yourself so that you won't feel so bad later. You learn it now, and you learn the replacement later.

Back in the 1800s there was a belief that you could only learn so much, so you had to keep your brain clear for the important stuff. (Go read Sherlock Holmes for an example of this prejudice.) But these days we know better. People can practice learning. And those who do, constantly are learning. About everything. Including trivia. And it pays off. It really, really does.

Are you now saying Java isn't awful? That's news, coming from btilly.

He doesn't appear to be offering any opinions on Java, or any other framework, language, etc., at all.

That is true. But I have offered opinions in the past. For example see http://programmers.stackexchange.com/questions/66808/java-or... for an opinionated comparison I offered on Python vs Java. I still stand behind everything that I said there.

If you go farther back, you'll find more negative opinions of Java like the one at http://www.perlmonks.org/?node_id=41244 that I stated 15 years ago. I believe that the substance of my complaints were true then. They are true of some organizations today. But they are not true of all of Java, hence my willingness to use Java today.

I can spell it out if anyone needs it:

Java isn't awful. Neither is PHP btw.

(I won't defend everything written in those languages though.)