Earlier quoted context omitted.
[Let's use "Alice" as the name of our hypothetical security researcher.] If the Bad Guys can get the exploit from someone else, then Apple equally could pay someone other than Alice to disclose it to them. Your premise assumes the work is fungible. If the entire set of people (including Alice) who are capable of finding these vulnerabilities conspired to withhold their work and push the White Hat market clearing pric…
You're missing the point. Apple isn't matching the market's bid for these vulnerabilities. Clearly, people besides Miller can find Safari flaws; the difference is, when Miller finds them, we know they aren't being sold to organized crime. By your logic, if he stopped, he'd be making things better for the Bad Guys; he's therefore obligated to do the work.
Let me put it another way.
There are two markets for exploits: the legitimate one, and the criminal one.
Charlie is participating in the legitimate one. He's going to get paid what the sole counterparty wants to pay him. We can argue about what the counterparty should pay him, but that's up to Apple (in this case), and there are a lot of different things that might enter into their calculation.
An argument that uses the value of the exploit in the criminal market in an attempt to set a value in the legitimate one only makes sense in one of two cases: (a) you're going to take your work and sell it over there, or (b) you claim that someone else either has already discovered or will soon discover the same exploit independently, and will choose to sell it on the criminal market, and therefore the value of your work should reflect the danger of that happening.
In the first case, you're engaging in blackmail.
In the second case, it's just not a very good argument -- because the chance that each element in the chain of reasoning about the value (it's about to be or has already been discovered by someone else, it's going to end up on the black market, it's a substantial risk for a 0-day, etc.) is not true represents a probability that reduces the overall value of your exploit in the legitimate market. Plus, there's the additional reductions in exploit value that come from the vendor not actually caring that much about fixing problems until they're in the wild, or having already found the issue and decided that the particular problem isn't worth fixing for a variety of non-technical reasons, or any one of a dozen other external factors.
Working on spec and then demanding that the vendors match the exploit values that the criminal market is paying is just a Bad Idea, morally and practically.