And of course that’s exactly what the major mining consortiums will do: Collude to increase the limit when it’s reached. After all, the rewards are in the mining, not the transaction fee.
You don't need anyone else to agree to start mining your own fork / altcoin. However, even if most of the miners change the protocol, it doesn't mean that the users will.
The incentive structure discourages this. If half the hash power goes off to mine Bitcoin Infinite (the fork with uncapped Bitcoin generation), then mining the original Bitcoin would be twice as lucrative because there's half the competition. This gets better the more miners leave. If 95% of miners leave for Bitcoin Infinite, then mining Bitcoin becomes 20x more lucrative. There's the matter of the difficulty adjustment, but that has been historically dealt with using emergency difficulty adjustments.
Meanwhile, the miners who are mining the fork have real expenses (electricity, equipment), which need to be paid, and if there isn't sufficient demand for their Bitcoin Infinite coins, they'll quickly go bankrupt. Some napkin math:
* If the hash and economic power are both split 50-50, then there will be little impact in profitability, but also little impact on Bitcoin
* If 90% of the mining power goes to Bitcoin Infinite and only 25% of the economic power follows them, then the miners can expect to see a 72% drop in revenue. If their original profit margins were 20%, they can expect their profit margins to drop to -66%. If we use the price/generation rate of just prior to the last halving (May 2020), then that would represent a loss of $9.9M per day (not including opportunity costs/lost profits).
The only incentive to mine then is to earn the transaction fee. These fees ($8) are currently orders of magnitude smaller than the mining reward of 6.25 coins ($125,000 at $20,000/coin). Either fees have to dramatically compensate by rising stratospherically or miners will depart for greener pastures. Yes, difficulty level drops eventually, but by then a large number of users have also followed miners off the network, in which case the price will drop too. As the price collapses, there will be a selling frenzy, further reducing the price. In this scenario, Bitcoin may settle to something as low as $10 per token.
Once miners depart en masse, network resilience will drop precipitously. That makes it an attractive target for someone to stage a coordinated attack, which would destroy remaining residual trust in the network and bringing about the end of Bitcoin original.
This is a common misconception about how consensus works. Everyone on the network has to agree, anyone who doesn't agree by definition won't be on the same network. They won't be able to receive the same fork of bitcoin that everyone else is sending.
Forks are occurring all the time at the apex of the chain. They wither and die slowly as 50%+ miners accept successive blocks as the next block. That’s why most exchanges and users wait for six blocks on average to deem the transaction settled.
I don't see why that can't happen to Bitcoin if the network participants (ideally normal users but practically the miners) and major stakeholders can benefit from it.
or just wait for a lull in the transaction volume and pay pennies per tx? eg https://blockstream.info/tx/3b9ce26a827b014e5f6cda461f5fa112... from last block only paid 14 cents.
"Simplified Payment Verification is for lightweight client-only users who only do transactions and don't generate and don't participate in the node network. They wouldn't need to download blocks, just the hash chain, which is currently about 2MB and very quick to verify (less than a second to verify the whole chain)."[0]
It's not clear what benefit the user or the currency gains from having 90% of users running full nodes instead of, say, 50% or 10% doing so. In any case, when people in poor nations have phones that can manage uninterrupted 24/7 connections to the bitcoin network, they'll probably also be able to afford 2 TB of storage (currently costing about $50) which would be enough to store 10 years of 4 MB blocks.