3 ms·
Don't be insulted; it's just an argument on the internet. Besides, the correlation people like me are claiming is at most a statistical pattern with lots of roo
by gruseom 17y ago
Don't be insulted; it's just an argument on the internet. Besides, the correlation people like me are claiming is at most a statistical pattern with lots of room for variation. (You sound like a good programmer regardless. But what I find endearing in your comment is how you go and spoil your argument by hacking on an iPhone app in your spare time - and feel obliged by intellectual honesty to fess up to it!)
Of course, you might also mean that the debate is insulting because it's pointless. I can understand why it would seem that way. What does it matter whether someone programs outside of their job or not? It doesn't, intrinsically.
Here's why I think the subject is interesting anyway: there's no cheap way to tell who the really good programmers are. This is not just a hard problem, it's a hard problem that seems like it ought to be an easy one. Why can't you just devise a test? People have tried to figure this out every which way, spending many millions, maybe billions in the process. And no one has a good answer. That's genuinely surprising.
Given that situation, any criterion one can use as a (cheap) proxy for "good programmer" is of interest, even if it's imperfect.
What makes the "hacks in spare time" correlation with "good programmer" interesting is not that it captures all of the good programmers. There are certainly good ones who don't do that; i.e. the test does produce false negatives. What makes it valuable anyway is that it excludes the overwhelming majority of bad programmers, the ones for whom it's a job that they aren't very good at and aren't motivated to get good at, who form 90% if not 99% of the professional population. In short, the test is interesting because it excludes many more false positives than it admits false negatives.
A corollary is that all the people who protest "I don't code in my spare time, yet I am a good programmer" aren't really adding much data to the discussion. The ones we need to hear about are the bad programmers who do. :)
- d0mine 17y agoEven if all good programmers were programming at home the knowledge that a programmer is coding at home wouldn't give us much. G - good programmer C - coding at home P(G|C)*P(C) = P(C|G)*P(G) P(G) = 1% # percent of good programmers among all P(C) = 20% # percent of programmers that code at home P(C|G) = 100% # assume that all good programmers code at home Probability that a programmer is a good one given that it codes at home: P(G|C) = 1.0 * 0.01 / 0.2 = 5%
- gruseom 17y agoP(G) = 1% <-- Ok, that's in the range I suggested. P(C|G) = 100% <-- More than what I said, but I see your point. P(C) = 20% <-- Whoa, where did you get that? That's just an assumption that leads immediately to your conclusion.
- Retric 17y agoI think 20% is on the low side. But, you really need to define your terms. I would call 1/3 of all programmers useful, in that they can create more value then they cost. However, when you have a tiny budget and need to find useful programmers willing to work long hours for low pay then you really want young smart programmers with a lot of experience. And they only way to have a lot of experience at a young age is to do a lot of side projects. There are plenty of programmer that will blow your mind making 150+k/year, finding programmers that will blow you mind and are willing to take far less and a slice of equity is far harder. PS: There are talented older people that are also under payed, but that's normally because they are happy where they are. Finding them and getting them to work for you is hard and expensive.
- gruseom 17y agoGood points. It's worth remembering that when it comes to this stuff, we're all anecdotalizing in the dark. I say that because it sounds like we've had quite different experiences. (20% of programmers work on side projects sounds low to you, very high to me.) Frankly, I'm not even sure it's possible to properly define these terms, let alone measure them.
- d0mine 17y agoYes, my assumption is there are many more programmers that code at home (whatever it means) than there are good programmers. In this case as I've shown "coding at home" doesn't have sufficient predictive power to be used as a quick test of being good at programming.
- bootload 17y ago"... Don't be insulted; it's just an argument on the internet. .." Worth reading just for that :)