I've been on both sides of this one. I'm going to speak mostly from the side of one doing the recruiting, because that was where I had the chance to actually le

I've been on both sides of this one. I'm going to speak mostly from the side of one doing the recruiting, because that was where I had the chance to actually learn why people were sought or avoided, accepted or rejected. When you're a candidate, you only have your own conjectures about why you get yeses and nos, but when you're involved in recruiting, you get a front-row seat for how stupid companies can be. Companies think that if they set requirements 3 levels above the actual work, they'll get it done really well, very fast, and reliably. That's not true. It doesn't work that way. It's the opposite. Companies are afraid to hire at-level because they can't admit the truth about the quality of work they have to offer. Most of what they are doing is not very interesting, and their environments are often broken, but they're terrified that if they hire at-level for that work, they'll poison the environment with inferior programmers. Reality: the culture was already damaged by the crappy work. It's a sunk cost if that's the kind of work you have to do, so fucking accept it and hire at-level. There's also a social-climbing element to it that is pretty disgusting. I've seen this with startups. They hire the people they can get-- a few are bad, a few are great, most are OK-to-good-- but then, after they're more established and richer, they decide that none of those people are "good enough" to grow with the company, so they decide to upgrade their crowd and hire shiny, new people. Familiarity breeds contempt, so eventually they are used-up shoes and newer, shinier people need to be brought on. (This is why, if you're an early startup hire, you need to lock in a VP-level title even if you're a full-time programmer. You can never get full assurance that people won't be hired above you from outside, but the title makes it harder. It puts a limit on that sort of shit.) There's also a ridiculous pansy-ass risk aversion involved. "We can't do machine learning unless we hire a machine learning expert with 20 years of production experience." Bull-fucking-shit. So many companies refuse to get out there and do something interesting until they can hire a Messiah, and people anything like that Messiah want to work for companies that are already doing interesting work.