EDIT: to be clear no-one, including myself, thinks the blocksize must
never change. Rather achieve scalability first through off-chain transactions, and only then do you consider increasing the limit. I made a
rough guess myself that it may make sense to raise the blocksize at a market cap of around 1 trillion - still far off in the future. Fees in this scenario would be something like $5 per transaction, or $1billion/year of proof of work security. (not including the inflation subsidy) That's low enough to be affordable for things like payroll, and is still a lot cheaper than international wire transfers. Hopefully at that point Bitcoin will need less security against takedowns by authority, and/or technological improvements make it easier to run nodes.
We won't get to $10 billion, let alone $1 trillion, if it costs $5 to make a bitcoin transaction. I refuse to believe that you truly do not understand this. Bitcoin will not capture enough economy if it is expensive to use because it won't be useful.
Nobody is going to convert their fiat currencies twice in order to use bitcoin as a wire transfer service with a marginal (at best) savings. And they will have to convert back to their fiat currencies because bitcoin is to expensive to spend or to move to an off chain transaction service (to spend there). Expensive is not decentralized.
Go fucking make PeterCoin, give it a 10KB block size, and see how far you get with that horseshit.