But to talk about the value of real code, the impression I get by looking at my friends, almost all big SV companies care about are how fast you can answer the almost cookie-cutter questions which can be culled from a number of forums. I know people had interns at MSR and pretty good publications but were rejected over those who semi-memorized those questions, but had almost nothing else to show off. Shameful process, which is partly the thing that puts me off.
Also to talk about the bluff. How can peer-reviewed research at decent venues be considered bluff. Or if you have contributed something which can be accessed by anyone, is visibly used by many, but is not on Github, how is that a bluff?
Is it only because you (as a recruiter) do not know what concrete contribution you (candidate) had in that specific effort? If that's the only thing, should not the right reference sort that out?
I'm not just talking out of my arse. I still remember getting that first job, and showing them that I had done things was a HUGE part of that.
I've also been in on interviews for dozens of candidates, and the ones that showed code were always much more likely to impress us. Without code, we had to guess how much you understood, and how much you were bluffing. Code doesn't lie.