It´s interesting how we can read the same thing but not read the same thing.
I don't think he meant what you think he meant. Otherwise bitcoin would have no issue right now with, say, 20mb or 100mb blocks. But it does. They won't work in real life. Unless of course you put Bitcoin in 5-10 data centers and then, yes, it'd work. Until the gov comes down and raids the data centers. That would be "awesome"...
Now that i think about it, we might never need a fee market.
Say we don't have fees.
What's stopping me from instering 500 Terabytes of spam for the lolz into the blockchain.
That is not what i said. There still will be fee, but it won't be the main factor in constraining tx output. Fee today is what you (and btw was intended as by satoshi) said exactly: spam prevention. Not to mention, if 1 block reaches 1000MB, that is around 250,000 tx x 0,0001 fee, which is 2,5 BTC per block. You can either raise the number of tx/block, or cost of fee /tx, or both to reach mining income. Why you speculate that only fee cost/tx can go up, not tx/block as well to pay miners?
Why would be only 5-10, or only 500 nodes? There is no factual data to support that claim at all. There are more than 6000 nodes today, run by voluntaries. Why those with a few hundred bucks and good connection won't run one in the future, especially as we can speculate hardware keeps getting better and cheaper? Who thought we will streaming HD+ videos today for giggles?
Again, no experimental data available to argue there will be less nodes in the future if we scale up the network - which is logically contradicting, 100x more users would suggest 100x potential node runners. FUD speculation.