The problem is that hiring someone who can't code doesn't help me. And a bad engineer is worse than no engineer at all.
It's not like a grocery store where a bad employee is just really slow and a really bad employee steals things costing me a percent of a percent on my margins. A single bad engineer could theoretically destroy a company if they were savvy enough.
Pretty much anyone who does technical hiring will tell you that it's way worse to hire a bad candidate than to not hire a good one and interviews are tuned for that. It's the opposite of the American criminal justice system. A no-hire doesn't mean you're a bad candidate but a hire recommendation is very likely to mean that you're a good one.
This sentence does not compute.
Anyways, I think a big portion of the "hiring problem" is that companies are too afraid of hiring a "bad" developer. They somehow have this notion that if they fail to hire a great developer then its no big deal, but if they accidentally hire a bad developer, their company will all go to shit. This results in crazy interview multi-hour processes.
In the real world, the worst employees rarely ever negate the work of more experienced workers. Thats like saying if you hire a bad teacher, students will forget information rather than learn.
I've worked in organizations where the hiring process the OP is advocating for (short one hour, possibly non-technical, interviews) demonstrably led to serious problems in that organization. Have you ever sat in on technical software engineer interviews? It's beyond horrifying how many developers I've seen that have a great resume, a decent portfolio, talk a big game, but literally can't solve a fizzbuzz-style problem on a whiteboard. I'd say the percentage in my area of developers who fail at this is at least 80%!
Think about that... if your hiring procedures are lax, possibly 4 out of every 5 developers on your team will be ludicrously incompetent. I don't care how tight the hiring market gets; dumbing down interviews is simply not an option with those kinds of odds.
If you haven't already seen it, you should read Atwood's article on the same subject: "Why Can't Programmers.. Program?" [1].
[1] http://www.codinghorror.com/blog/2007/02/why-cant-programmer...
I'll tell you a secret. The first time I was asked to do fizzbuzz in an interview I got it wrong. It was about a year ago. I have been programming sine I was in middle school (I'm 29 now). Walking out of that interview was probably the worst day in my entire life.
Ever since then, two things have happened. First, the lines of fizzbuzz are forever etched into my mind. I can recite fizzbuzz verbally now. I don't even need any writing utensil. For i in xrange 1 comma 101 colon if i mod fifteen equals zero colon print "fizzbuzz" elif... [and so on]". This has come in handy in interviews since. All you have to say is "fizzbuzz" and I go ahead and write away. I don't even need the interviewer to explain the actual wording of the problem.
And the second thing is that I now have absolutly zero interrest in hiring a company that has fizzbuzz (or any other live coding portion) as part of it's hiring process. If I'm at an interview and the interviewer tells me "we're going to write code now", I'm done. If the problem is interesting I'll do it, but if it's fizzbuzz or anything having to do with reinventing something, I'm done. Not gonna work there.
I've struggled to put into words how to describe why I think live coding is such a terrible thing to ask candidates during interviews. Joel Spolsky makes a great point thats hard to argue against. If you're hiring a guitar player, you want to hear him play. If you're hiring a programmer, you'd first want to see them write code. I think the problem is that programming is kind of more like song writing. You can't ask a song writer to go the white board and bang out a new song. Even if you tell him exactly what the song should be about. Song writing is a process, and so is programming.
People also seem to minimize or ignore the cost of a no-hire. If your company has an open position it is trying to fill, it should be feeling pain. Work isn't getting done. Projects or clients can't be bid on. Current engineers are working overtime to meet commitments. If you can look at this situation, shrug your shoulders, and say, "Eh, we're going along fine. We can wait several months for a perfect fit.", then I think you need to re-evaluate why you have an open position in the first place.
I would like to see some actual data to back up this conception. Something that gives the actual statistical risk. As someone mentioned on another story here recently, humans tend to over-estimate the risks of negative events. I think the risk of bad hires in SV has been seriously over-estimated.
I.e. the incompetent new guy just has to fake it long enough that he gets promoted. Now his incompetence has leverage. He can most likely shove responsibility onto others, and eventually gets promoted again. More leverage.
If I were to start an organization, I would make damned sure that every single new hire could do what they were hired to do - and then some. Otherwise, the office dynamics may lead to a stagnant organization over time.
Meaning. Someone who is bad at programming but savvy about interacting with their coworkers, bosses, etc. Someone who might be able to keep a job long enough to really fuck stuff up even though they're bad at what they're doing.
Six months in you realize that they've built a bunch of terrible stuff into the system and now you need multiple people to tear it out and rebuild.
How does one know that one is a top developer?
I'd imagine it's easy to have a false knowledge of being in the top 25% (what with Dunning-Krueger and all) but how does a true top 25 percenter know that he is such?
Questions to ask yourself:
- Do you consistently get higher ratings than your peers?
- Do other developers come to you first if they want advice on the correct way to do something?
- Have you produced features that cause applause in sales teams or customers?
- Have all of your products been successful (maybe allowing for slips in schedule or budget)?
- Do people understand your code? (i.e. do you get a lot of emails from people asking to explain things?)
- Do features that you have created have significantly fewer bugs than other features?
- Do tech leads or product managers seek you out to help them build products or features?
For example, the company for which I work bills me out at $250 / hr. I get something like 15-30 job requests from other companies a week. It would not make any economic sense for me to pursue any of these other companies unless the interview process was dead simple, transparent, and easy (and the total compensation was _at least_ 15% higher than my current compensation). But, 99% of them will only offer modest pay adjustment (%2-3?), and then they expect me to sit through several hours of interviews. And, if I pass that nontransparent process, they ask me to sign a ridiculous employment agreement that signs away all my personal IP. Seriously?
The problem with recruiting today is that most companies who have good talent understand how valuable their best talent is. They treat them very well. And, the companies that want to obtain that talent completely under-offer their side of the deal.
Another problem is that most developers suck at negotiation. I see lots of anger on HN about not getting the 6-figure salary. Well, that's their fault for not asking for it. If they're a relatively good developer (top 25%), they should be making at least 6-figure, easy. And, if they're not, that's their fault, not the fault of others that know their value and how to demand a cut of it.