I think it means to build something that the shareholders wouldn't want to expend internal resources to duplicate, but it's hard to tell for sure.
If @daltonc's app.net doesn't massively undercut FB's App Center revenue plan (or overwhelm it's distribution plan) somehow, what's the point?
I think your cannibalization concept is brilliant, but I don't think app.net is completely vulnerable to it. Caldwell's made it clear that he's not trying to find success of Twitter and FB's magnitude, that sustainable profitability is more important for this venture than unbounded growth and huge winnings, and he's not trying to steal Twitter's customers, whatever that would mean. But unless Twitter becomes wildly profitable I don't see them giving away ad-free unrestricted API access just to eliminate a small gadfly.
I think your concept has a lot of merit, but I'm not sure it applies in this case.
a) dev builds software on the platform that doesn't disrupt the platforms economics, distribution, or organizational limitations. b) dev builds software on the platform that does disrupt the platforms economics, distribution, or organizational limitations.
You seem to be saying that if he had done a 'proper job' he would have pursued strategy b. But surely that is precisely the case where you can expect to be bullied. It is strategy a. where you develop software that does none of those things, but adds value to the platform by adding value to the users of the platform where you are least likely to be bullied. In these circumstances you are no threat and you are complementing the platform.
Strategy a can only be pursued when you are not dependent on the platform and hence are much harder to bully.
I mentioned elsewhere that, personal opinion only, of course, that this expectation is the sign of bad mindedness. Think of the logical implications of this point.
And, when I don't like the tone of a meeting, I politely leave. Facebook's offices are physically difficult to enter, not leave.