Point 3 is incorrect because there will never be exactly 21M bitcoins. It will end up being slightly [1] less than 21M.

[1] https://en.bitcoin.it/wiki/Controlled_supply

> It will end up being slightly less than 21M

Or a lot less. People lose their keys all the time, die mispay.

Definitely a lot less if you look at it in terms of lost keys. I'm speaking specifically about the scenario where no keys are lost, the total possible supply of bitcoin will still never be 21M.

If a majority of miners (hash power) decide to mint more, there will be more. The code change is a simple one. The politics and economics are, as always, much more complex.

If a majority of full nodes (validators) reject blocks from those miners, their chain will be orphaned and the block reward will fall to the subsidy they've created for themselves (where block_reward = block_subsidy + block_fees).

Bitcoin has value because this just isn't possible, in practice.

> just isn't possible

It's absolutely possible. Who runs nodes? There's no way to know that the majority of hash power doesn't already control a majority of nodes.

As long as you run your own node, it doesn't really matter who runs the rest. If a peer relays an invalid block to you, disconnect from that peer and defer to the chain with the most work.

Have you read the paper?

If a block is accepted by a majority of nodes and miners it isn't invalid.

Your own personal node is free to implement different logic, but it will be unlikely to interact in a functional way with the rest of the network.

The full node operators are typically holding Bitcoin, and I just don't see anyone willingly accepting a reduction in the value of their assets. Any such block would be viewed as an attack on their sovereignty.

There is a way to know since segwit was activated using a user activated soft fork despite not having a majority of miner support

Node control has worked in both theory and practice