When I was a senior at university, my friends and I all had a wall at our apartments covered with our rejection letters. Back then (if they gave you the courtesy of a reply) they were sent on paper in the mail. So yes, it's always been that way.
I myself have been on interview committees and voted to reject candidates who got hired, and turned out to be good. Why did I want to reject them? Don't really know, just got a bad vibe in the interview more than anything. We didn't do much technical screening at that company. Decisions were mostly based on past experience and references, and whether you seemed reasonably intelligent and sociable in the interview. I think that we hired a real dud maybe once, so it seemed to work.
It seems to me anecdata suggest the average hiring process lets through only a few duds.
It often feels to me like people are using that as a justification of the quality of their hiring practice. What's interesting is hiring practices vary greatly across the board but mostly in aggregate seem to produce this "we only ever got a few duds" result.
My hypothesis has now shifted to "most people can do most jobs". Or to restate it "There are only a few duds. Period. And hiring processes produce mostly random results. That's why we see a few duds come through."
I did this when I was a math undergrad applying for grad school. I applied to 15 graduate programs and got into 10, and rejected by 5. The surprising pattern was that I got into a graduate program if and only if it wasn't in the Northeast; selectivity of the program didn't seem too relevant. After graduate school, for my first job I ended up being on the graduate admissions committee at the Harvard Mathematics department and got to see things from the other side, where you receive 400 incredible applications, and can say yes to less than 1% of them. In that setting exactly what you wrote "Decisions were mostly based on..." applied just the same.
Over the years I've known various undergrad admissions people at another Boston area school. At least the way they used to do things was to basically have a matrix with quantitive academics/test scores on one axis and a score for soft stuff on the other axes.
Admit the "no brainers" in the upper right. Reject the "no ways" in the lower left. Then take a harder look at those in the middle. But the reality is that you could probably take an arbitrary subset of the middle group and they'd probably do nearly as well/contribute as much/etc. as the final selection you sweated over.
I've also been on the selection committee for various conferences over the years and that also often ends up working the same way to a significant degree.
As a nice twist on the rejection postings, while in college I visited a friend at Yale. He and his roommates had hung all three of their Harvard acceptance letters next to each other on the wall.
It's stunning how much energy, ink and tongue movement is wasted trying to justify most decisions, in every aspect of life, when in practice they boil down to vibes.
I’m a hiring manager and I think it’s a complete roll of the dice sometimes.
There’s people who came off really well in the interview who turned out to be terrible hires. And equally there’s been people who seemed woefully inexperienced but I liked their honesty so took a gamble on them and they turned out to be some of my best hires.
Then there is everyone in between.
You have all of these processes in place to help your practical judgement but at the end of the day it’s really just a lot of luck because it’s trivially easy to game interviews and some really good engineers are often really bad at interviewing. So pulling those who interview well but are rubbish from those who don’t interview well but are brilliant is very difficult at times.
Heck in one interview I got nervous and started rambling on about FreeBSD instead of Linux. I was easily qualified enough for that job but that was a terrible interview on my part and I honestly wouldn’t have hired me.
So that's the issue - you're hiring someone and you have no idea which version of them showed up to the meeting, and which one will be coming to work every day.
It's almost like we're trying to solve a taylor series of work efficacy (of an individual on a given team) by capturing a very small number of data points in stressful situations (interviews) and doing a hell of a lot of extrapolation.
I myself have been on interview committees and voted to reject candidates who got hired, and turned out to be good. Why did I want to reject them? Don't really know, just got a bad vibe in the interview more than anything. We didn't do much technical screening at that company. Decisions were mostly based on past experience and references, and whether you seemed reasonably intelligent and sociable in the interview. I think that we hired a real dud maybe once, so it seemed to work.