All of the arguments that "nodes do matter" have the logical fallacy of taking the "intended way the network SHOULD work" as the "actually technically resulting technical operation".
I agree with all you say. However, the situation is even more dire. Turns out
we have been lied to all this time. The reality is that no entity is a node, if that entity does not mine. Indeed, this is the true, original, and proper definition of a
node.
This is what I'm trying to point out. I don't know if it is "lying" or "indoctrinated" or whatever.
But the argument that bigger blocks would lead to centralization, while the centralization already took place, always left me astonished: no-body ever reacted to that. It is not so much that I absolutely want bigger blocks or whatever - it is that the argumentation is fallacious, and I have a hard time believing that the experts saying so, can't make the same obvious reasoning than I did here, and did several times, just to be greeted with the counter argument "but everyone knows that full nodes matter / keep the network honest / .... " and if that doesn't work, that "I'm a paid shill " or something of the kind. So, are all these people self-deluded ; or do some of them know this but don't want it to be acknowledged ?
I have difficulties with fallacious arguments from experts.
That said, full nodes are not totally useless, but their only use is for *their owner* who is the only one who can *check for himself*. But with that knowledge, he can do nothing else but acknowledge "that he has been had" or "that he hasn't been had", but that's about it. The other advantage of a full node, for his owner, is that his owner can send out his own transactions, and nobody can know that HE was the one sending that transaction, as, being a full node, he would also send all transactions of light wallets connected to him. So there is some kind of deniable anonymity of the IP address that sent out a transaction.
But that's about it. These can be sufficiently good reasons to run a full node, BTW.
Of course, it is true that full nodes DID HAVE power of filtering "bad blocks" when mining nodes were connected only to the P2P network to other mining nodes, and didn't talk to themselves directly. Then the P2P network could stop them from building a chain with which the full nodes didn't agree, because they would not receive one-another's blocks.
But as I explained several times, a miner pool would be crazy to wait for other miners' blocks through the P2P network, while he can just configure his node to connect directly to the other miner pool's node and get it faster, wasting less hash rate. As this is mutually beneficial, I don't see why mining pools wouldn't do so.
When I look at the LOW orphan rate, I arrive at the conclusion that miners know one another's blocks in about 1.5 seconds on average, which indicates that those blocks cannot really hop from node to node in between. 1 MB blocks, transmitted and verified within 1.5 seconds seems impossible to do over a random P2P path, where each node receives the block, checks it, and sends it out again.