Some more points re. the Programmer market, which maybe generalize to the academic job market:

> Why? People mostly speculated that the junior/senior distinction is reality-based; it takes 2-4 years to learn to do the most important tasks. The failure to hire insiders might be because workers aren’t on board with this distinction and would expect a more linear raise schedule.

It’s actually a lot more rational than that. First, there’s a risk/reward tradeoff related to promoting people internally. Due to inertia and the general difficulty in letting people go, there’s a risk associated with promoting an insider. It might not be obvious that they might lack the required skills, or otherwise might not be politically viable as a promotion candidate. On a skill side, if they are promoted too aggressively, there's a higher perceived risk that they gamed the promotion-evaluation systems. Someone recently promoted to a senior position is a lot more difficult to immediately let go of (and it's impossible to demote internal folk), so if let go too soon after a promotion, it shows up as an embarrassing mistake on the side of a manager somewhere. On the politics side, this enforces a relatively conservative set of requirements for promotion (the candidate has to have visible achievements that are exemplary or above average for someone working at the new role level). Which means, internally, managers will err on the side of caution and only “promote after someone’s been performing very well at the new level for a while”, which is usually uncomfortably late for everyone involved, and then puts pressure for a larger jump in comp. There’s also the angle related to hiring strategies - one might be less risk averse, and hire 4-5 risky juniors for every 1-2 people who eventually make it to a senior position. So there’s some pressure to recoup that cost. Ie the cost of training and education for 4-5 people, of which roughly, say, 50% might be let go of or otherwise leave before making it to senior. So, especially at mediocre or badly managed companies, there’s a pressure to try and recoup that cost by having folk who are starting to perform at senior levels stay at the lower tier or lower comp within the same bracket, for much longer than appropriate.

Looking at it from the point of view of hiring an outsider into a more senior position, it can look a lot less risky. The senior person will have 3-6 months to prove themselves, during which period they can be let go (or alternatively, demoted; it's actually a lot more reasonable to demote a new hire under the right circumstances) with a lot less political cost or fanfare. The risk/cost of training and education is paid for by someone else. The evaluation process for new hires is usually perceived to be much more selective and harder to game. To de-risk things, some of their higher compensation can be in the form of a signing bonus that gets clawed back if they aren’t a good fit. Similarly, on a political de-risking level, it's more easy to fudge their experience and historic achievements, to make it sound like they’ve accomplished enough of the politically necessary, high visibility, projects. So it's easier, especially with mediocre management in place, to justify paying a premium when hiring externally for more senior positions

As a caveat - talented managers will usually find ways to make the comp and promotion curve much smoother, even when forced to work within the bounds of an unreasonable external framework

----

This pattern can actually get really bad at higher echelons in large engineering orgs. FAANG like companies notoriously tend not to have enough of the required high visibility projects or work for folk to get promoted to the higher echelons of the engineering org at a reasonable rate (and to then be compensated appropriately). It’s often is much easier/faster to get to the higher levels and force a compensation correction by leaving, achieving the politically required high visibility work or projects elsewhere (eg. at an early stage startup) and then be acqui-hired or hired back one or two levels above the one the engineer quit at