[FabHK]

So, let's play that out. Suppose 90% of miners (hash power) collude (as GP posited), for example validating blocks with a larger coinbase (mining reward). The "good" nodes don't accept these (but the "bad" miners can trivially have many nodes that do accept them). But now the "good" hash power drops to 10%, so only every 100 minutes a "good" block is mined (suppose this happened just after a difficulty adjustment), and that goes on for 20 weeks. In the meantime, the bad nodes and miners carry on as before, pretty much, but with their malicious change incorporated. If they're bored, they can devote some of their hash power to double spend attack the "good" chain. Which chain will come out on top?

Sure, it's unlikely. But not impossible. Just as runaway inflation in a reasonably managed fiat currency is not impossible. Just unlikely. (And, yes, hyper inflation has happened historically. Similarly, rewriting of the immutable Ethereum chain and BTC forks have happened historically.)

[Sohcahtoa82]

A malicious set of miners that decides to give itself a higher block reward will never have their blocks accepted by the "good" nodes. It's quite simply not possible. The blocks will be invalid and therefore rejected outright. It doesn't matter how much hashing power is behind them.

> If they're bored, they can devote some of their hash power to double spend attack the "good" chain. Which chain will come out on top?

That's a different attack entirely.

[eraad]

Is this just a hard fork? This exists, I believe its called Bitcoin Cash or Bitcoin "Satoshi Vision".