Usually interviewers have already made up their mind before you even walk in the door. It's mainly a formality and confirmation of their decision of candidates.
Then why would they waste time and money on the interviews that are followed by rejection.
Wouldn't say so, we can conclude a person's coding abilities by looking at their code at Github or task they've received (this can be tricky tho) but you can't decide about hiring/not hiring upfront until you really met that person and see what's he/she like.
There is a bunch of other things that need to click in, besides obvious requirement of knowing how to program and how to think, stuff like a cultural fit, teamwork abilities and (the important one) is he/she Star Wars or Star Trek fan :)
I don't think, that you can conclude a person's coding abilities by looking at their code at Github, because it is easy to copy other peoples' code.
My best coding interview was live coding.
I had to build a React todo list from scratch, including setting up create-react-app, looking at my TDD workflow, git workflow, how I solve upcoming problems etc.
Yeah I agree, we can conclude that to some level, we do some testing on site too. Also, if the candidate has done some sort of interview task, we're going to go through code together and talk about it. Those who just copy-pasted code from the net, without understanding it, are easy to spot.
Live coding is good in a reasonable amount, but if someone is expecting from me to code an app from scratch while whole interview team is lookin' at me, I'll just bail out, it's just too weird.
I think it's very insightful.
How does the person react to this stressful situation?
How does he help himself when errors come up?
Does he actually do real TDD?
Does he really think about the architecture or does he start writing some useless stuff?
Can he talk about what he is doing?
How does he react to another person saying something?
Does he write clean code?
We're a place where coders share, stay up-to-date and grow their careers.
We strive for transparency and don't collect excess data.