The companies that are trying to push this agenda are companies that are in business of selling interview/collaboration tools.<p>I accept that the interview process can be improved but not so much as broken. I still feel that the ability to solve data structures and algorithms problems is a good measure of swe skill. Though one can argue that swe is much more than just ds and algo problems. But it still is a good measure.<p>Please give me an alternate perspective here?
Look.<p>After a certain point, software engineering becomes less about handling raw data structures, or coming up with some whizbang new algo and devolves to a matter of A)plumbing and B)bridging the gap between what everyone thinks the machine is doing vs. what it really is.<p>I keep around my old CS textbooks. Why? Because I know they have valuable details on algorithms and such in them. Those have been sorted out, are good things to refresh every now and again, and are important in terms of knowing in which circumstances they should be applied; but don't ask someone to play the part of a compiler. Also, all that pales in comparison to the importance of actually being able to cobble together, understand, and reason through systems, and to explain the "why"s of different implementations.<p>I also don't believe anyone should come out of an interview empty handed, which happens more often than not with places that fall back on "lets see how they do on brain teasers!" I do my best to shift interviewers towards actually talking problems they're having or things they think they'll have problems with to get a feel for whether I might be able to immediately help them or not.<p>I don't find that the traditional whiteboarding exercise is helpful, especially when I tend not to do my best work until I have time to reflect on a problem without it being the spotlight of a social interaction.<p>All that being said, I'm not sure there is much that will change about it in the long run. So, I just try to get practice when I can.
I have a CS degree from MIT. I built numerous systems at edX that are still in place today. At Stripe I’ve worked on software used by many of the folks here, and far more outside of this forum.<p>I haven’t focused on data structures or algorithms since early interviews when I graduated. My most complex data structures are usually hashmaps. The most complex algorithm I’ve written mostly deals with heuristic string matching to account for human error. I last used binary trees directly about 3-4 years ago.<p>I can probably do many of the software jobs in industry today, but I wouldn’t get them because most of the interviews are focused on measuring skills that have little to do with the actual job. All I need is Apple and Netflix to be able to say I’ve bombed interviews at all the FAANG companies.
It's broken because the interview has little to no correlation of on-the-job success. The current process does a poor job answering: "Can the candidate succeed in the role we need filled?"