Economically, it would make sense for A and B to work for C, since they gain 50K more. Economically it makes sense for C to hire them since together they create wealth, and the 700K that C gets to keep is a reward for his willingness to risk hiring them and managing them into a productive team etc.
If A and B would like the risk, they could start pairing themselves together and attempt to create wealth independently. But they could fail. 10% chance of a 1000K is still just 100K on average (it just has an interesting distribution of the payout). The difference up to 1000K is a payout to C for the risk they took in growing their business up to the current level of scalability. (otherwise other companies would pay more for the value that A and B provide)
If A and B care only about technology and are risk adverse, the economy doesn't award them more than the 150K which they are already getting. For such people, creating wealth of 1000K for the companies where they work is not a result of their behaviour, but a reward for the founders that took the risks and had the vision to create organizations allowing that.
And obviously, doing a startup is no guarantee that your returns will be better.
Here's what I really mean by subsidizing, in context, and you can decide if it fits, or if I am suggesting something else...
If you buy into the idea that a developer, for example, can be 10x more productive than another, which I do (and I'm not one of the 10x guys at all), then you tend to think that those people should be paid a heap more than the average IT guy... but in fact, they are bound by the same corporate salary ranges as everyone else on the team... so you end up with the delta between the bottom guy and the most productive guy is at most 70K.
Therefore, thanks primarily to the top performers, our overall team performs well and everyone benefits similarly... but they really shouldn't benefit similarly since the amount of value creation is wildly different from developer to developer.
Just because you are 10 or 100 times more productive doesn't mean the company can sell 10 or 100 times more.
I suspect it gets diverted to senior management (founders?) and shareholders. Given we are commenting on this at ycombinator.com, I don't have a problem with it, since I'm hoping to end up in either or both categories at some point.
Your subsidy argument cuts both ways.