> One of the least effective predictors was unstructured interviews or 'chats’

Yeah those are the worst, one time I had an “interview” with a company that I really liked, the founder is also an awesome guy and we chatted few times and all is well. Then I got invited to their facility, great place and team, some of them were structured on how they evaluate, but most of them were an absolute mess, and some of them were hostile as if I would get hired it will get them fired the day after (the passive aggressive of trying to belittle your projects or work and not trying to understand your approach it but to attack it instead) and when I would ask them in a good faith about something they did, you would get a fake halo effect with “oh I can’t tell it’s secret! NDA bla bla” as if they did a patented work.. it was horrible method to hire people despite the great founder I knew.

In my opinion, the best way is what I usually do, after initial screening, I give them an assignment that they can do in few days and then return the work, the quality of the output will determine that, and it’s exactly how you will do in real work anyway, and you get to measure their critical thinking and problem solving rather than how would they sell or articulate something on the spot (maybe they are overwhelmed and their head went blank), as I am looking for an engineer not a sales dude, and they would tale some time to build and solve it.

I've been talking to a few companies lately, and one just keeps stringing it along, having me talk to manager after manager. It's been 6 weeks, and still no end in sight