It's not about actually having to implement these algorithms in your practical, day-to-day work. It's a challenge to test your reasoning and problem-solving ability in abstract, that you can administer in 15 minutes. You can't really test a candidate with real-world workloads, can you.
You absolutely can test a candidate with real-world workloads. It takes longer than 15 minutes, though.
I have no idea why anybody cares about the 15-minute thing. Each person you hire adds thousands of hours to your available labor. So even if I spend 100 hours finding the right candidate, I'm still way ahead. And the better my working environment is, the lower my turnover, in which case I can spend even more.
While I fully agree with you in that interviewing is broken, and it's broken because we throw away a lot of great people, you'll be surprised how much time a company spends finding the right candidate.
Imagine we spend an hour per candidate on screening, and we pass 10% of people, and we spend 6 hours on each on site,with 25% of them passing.Finally 50% of candidates accept the offer: This is very typical math, and it ends with over 100 hours per developer hired. For each hire, we also had a declined offer, 6 candidates rejected on site, and 72 failed screenings!!
Given those round, but not really all that far from reality numbers, any increase in screening time will spiral out of control unless some other multipliers change. The one that is most likely is that said company is giving offers to less than half of the people that would have been successful.
Those numbers also show why it's so important for a company to make competitive offers and woo candidates, and why they'd not like it when people interview at 6 places at once: Run the calculations just changing the acceptance rate down to 30% and up to 80%. If you are trying to recruit from a big name university, chances are that your candidate really is talking to a dozen serious SV companies and gets 4 serious offers.
Therefore, while I agree with your sentiment, you can probably see why it's so hard to convince someone running the traditional SV pipeline to make changes, as you are either raising costs or telling them that their current outcomes are a gigantic dumpster fire.
Anybody spending 6 hours per candidate on site can afford to test them on some real work. I'll typically do 2 hours for pair programming, and/or 1 hour for reviewing some existing code. 1-2 hours is also a good amount of time for a joint design session on some real problem.
You're definitely right that we're missing good people. I keep coming across people who didn't get the right job until they invested a bunch of time into learning to beat the Mensa-puzzle interview. So one way to make the numbers better is to find more good people.
I think if we're going to do the math, I also think we need to account for the cost of getting people who are not so good. If we test some thing that we hope correlates with doing the work, we're asking for large downstream costs. The more our interview tests what people actually need for long-term success, the better off we are.
How do you keep interviews consistent for candidate comparison with pair programming? I'm assuming you mean that y'all are pairing on actual work. That can vary. Yesterday was simple CRUD, "candidate nailed it." Today, there is an obscure concurrency bug, and the candidate would need more than 1-2 hours to understand the landscape of the complex code base we are asking them to pair in; "the candidate asked some ok questions I guess."
I fully agree with "the more our interview tests what people actually need for long-term success, the better off we are."
^^ This. As tokenadult always points out, a work sample test is the way to go.
I do it by pairing on a standard problem, one I'll use over and over. That can be real work in the domain or a toy problem. Both seem to work pretty well to me.
I'm pretty sure having candidates do unpaid work is illegal in California, which is another reason not to have them pair on actual work that you plan to ship.
You are often weeding through a large number of applicants, so spending 100 hours on each candidate is just not feasible, and, generally, not necessary. Plus, high quality, experienced developers are not going to want to spend 100+ hours doing real work for companies just to see if they can get a job.
I'm more than happy to spend an hour here or there for a phone screen or coding test to show that I understand the basics of data structures and algorithms and that I can use that information and my reasoning ability to solve problems I haven't seen before.
And how many candidates do you have to consider to find the right one? And what proportion of the interviewer's time is spent on the live coding interview vs setting it up and arranging it? I've been in this boat recently. It doesn't really take much to get to double or even triple digit hours if you don't automate something.
No, it's probably more like 4 1-hour interviews where they do whiteboard coding 2 - 4 times for 30 minutes and the rest of it discussing other previous experience, what they are looking, what the company's doing now and looking to do in the future, etc.
"reasoning and problem-solving ability in abstract"
Implementing a syntactically correct tic tac toe program on a whiteboard doesn't fit it. Plus many are comfortable in working behind the screen than to work in front of white board (real world)
