You are correct. You have to pay someone a tiny amount: the cost to make executing your transaction, storing the blockchain, and mining a block worthwhile. That cost is far less than the monopoly rents that beneficiaries of network effects collect. It is cents per transaction.
What about user interface and user experience? That's a key area that all the successful companies you've listed do very well, and something that decentralized approaches traditionally don't do well.
The typical user doesn't care how it works under the hood--they just want their room/driver/goods and services quickly and with a minimum of fuss. Being good at this will remain difficult and expensive even in a world where Ethereum dominates. Perhaps it competes better on the infrastructural level, against the likes of credit card companies, banks, and cloud services than against polished b2c applications?
There's actually no justification for charging for this.
The "cost" of making the transaction includes development costs, which you've had $18m to pay for (isn't that enough?) and computer cycles, which are distributed anyway.
So what you're actually doing is adding a commercial chokehold and red tape into a system designed for the open distribution of value and information.
You're simply aiming to become an alternative monopoly.
Thiel approves, of course. I think the idea is a very bad one, because I believe in removing all friction and cost of entry to the absolute maximum extent possible - not in automating red tape with contract code.
