If you do lots of quality work, but most potential hires can do the same thing slower, then you are productive but easy to replace. (A reasonable amount slower, not a ridiculous factor.)

At the less extreme end it's having a bunch of internal knowledge that isn't documented and is important to keeping things going. Or having business skills that whoever is in charge of hiring doesn't know how to hire for. Or knowing a programming language or framework that's fussy to learn and only 2% of your normal hiring pool knows.

Very different from just being fast.

The guy I linked has a very niche skill. And he talked about bringing in a Bulgarian with a similar but different very niche skill.

Neither one of those is more productive than the other.

But neither one can be replaced with the other.

That's an irreplaceability advantage, without a productivity advantage.

And you might find that a good bricklayer is actually more productive, but he can be replaced pretty easily with a different bricklayer.

Those niche guys would be in extreme demand even if they slowed down 3x, because there is no way to replace them.

Here's a hypothetical: You're in charge of purchasing supplies for making a machine. Then 9 other people build it. The whole team is pretty productive, but you think it's fair to say you do 10% of the work.

Then it turns out some critical part can only be bought by you. You've become irreplaceable, but it's hard to argue that your productivity has changed.

Then you decide to demand an assistant since you're irreplaceable. You start pushing more and more of your purchasing work onto your assistant, until you're only handling a couple products and you come in one day a week and leave at lunch. You're still irreplaceable, but your productivity has fallen off a cliff.

I'm not really talking about the top handful of people.