Let's assume we model Bitcoin swarm behavior with game theory. In this case 1 BTC is the payoff and if I increase it to 2 BTC this is a more attractive payoff. If I program a Bitcoin miner to maximize his BTC profits, he will opt for the 2 BTC chain.

Of course, you are completely right that Bitcoin miners do not work that way. They are human beings and not just game theoretical agents. They know about dilusion and inflation and therefore will not readily switch to the 2 BTC or the 11 quadrillion BTC chain. Thus, we cannot model Bitcoin swarm behavior with game theory alone. We must look at human consensus and at economic aspects. So let us do so...

However...we now realize that this provides no final answer as well. Most states, share holders, companies, federal reserve banks know about dilution and inflation, nevertheless they, as human beings, still decide to accept a MINOR form of if...a bit LARGER form...since it worked out...again a bit LARGER...and...ooops...we have a financial crash.

So, wy should Bitcoin be more stable than the dollar, when the Bitcoin system inherently has the same problem (human beings making bad decisions) as the dollar.

I must admit that I get some pleasure from thinking about this theoretical issue and trying to find an answer which Bitcoin mining cannot compensate. I am interested whether this model is correct or my line of reasoning has a flaw. I am interested which mathematical / behavioral mechanism could eventually provide for some ultimate stability in Bitcoin (which, as mentioned above, it does not have in it's current form). Compare with the dollar: It HAD a gold standard - until the president / fed abolished it. What would prevent miners to abolish the current limitation system in BTC...and opt for a 51 BTC per block chain (or, rather, 14 quadrillion in the end).