I interview a lot of candidates. I just can't imagine to make a hiring decision for a dev, let alone a manager that manages other managers, based on 20 minute discussion.
Regardless of the size of the org you need 45 mins to get good signal.
20 minutes might work for a “what questions do you have” sell call.
But any company making hiring calls on this model, that’s a yellow flag right there
At what size of org it stops being important who is going to be heading it?
But what if others in their 20 minute discussions with the candidate ask the questions you would have asked if you had spent longer interviewing them?
If the hiring decision is based on the feedback from all the interviewers I could see having many of those interviews be short interviews where the interviewer just concentrates on finding out one important input for the group decision working, provided that there are enough interviews to cover all the important things and if there has been some planning on the part of the company to coordinate who covers what in the 20 minute interviews.
I have no idea if Stripe does the necessary coordination to make that work, but the fact that several of the interviewers started out mentioning they would only need 20 minutes suggests that it was some sort of organized thing.
As a manager/leader of that organisation they will have an important role that can mean difference between those hundreds of people bringing huge value or huge loss to the company.
So your responsibility is to figure out how much time to spend with the candidate. You can choose anywhere between "just hire first person to apply" and "spend a year grooming an employee to see if they can do the job".
And you want to tell me that 20 minutes is the right answer here? That out of entire continuum of possible choices you say that the optimal return (performance of manager) on investment (cost of conducting interviews) lands at approximately 20 minutes -- less time than you take to have a lunch?
I get that he had couple of these sessions but still... it sounds like giving the job to a first person that looks the part.
The former gives more flexibility to alter the questioning on the fly, such as to delve more deeply into some area than had been planned. The latter gets more people to spend time with the candidate.
A mix of this could be the best of both worlds. Have several short interviews mixed with some long ones. If one of the short ones turns up something that seems worth going in depth on that can be handled in one of the long interviews.
As you say, it is very hard to attribute a bad recruiting process to something that is non-structural...no matter how many thousands of people you hire.
Anyone who can do anything with zero variation should definitely drop what they are doing now and make it their new business.
If you are in software, recruiting is your business. You have no other real assets. So categorising your hiring process as a random variable makes no sense. You should have processes in place that ensure non-randomness...again, is Coca-Cola out there selling tons of rat piss, and just saying: "Tough luck guys, this is a hard business"...no. If you don't have processes to ensure that outcomes in the core parts of your business are not random, you don't have a business (I used to work as an equity analyst, I have heard this kind of thing from CEOs over and over...I never recommended investing in such business, I have never seen a company that was run that way succeed).
I've seen odd things first hand with processed food from the grocery store. I've bought sealed packages of food that were all dried out and stale. Or that looked fine but gave me...indigestion. The weirdest thing I've seen recently were some mints where some of them randomly were solid chocolate, no filling. Oh, and a frozen dessert had a sealed cardboard box, but the plastic covering inside was open.
How does that sort of variation happen? I'd imagine that the better your process is, and the less variation you have, the larger proportion of your failures will be "unknown unknowns" that are just weird.
I acknowledge the conclusion that the interview process is f-ed up could well be correct.