Quote from: Peter Todd on February 24, 2014, 02:40:01 PM

Quote from: d'aniel on February 24, 2014, 10:13:59 AM

Quote from: Peter Todd on February 24, 2014, 08:35:38 AM

If you think merge-mined chains represent proof-of-publication in a world of large pools, you misunderstand what the idea is. Fundamentally merge-mining is insecure without the participation of at least a very large fraction of the mining hashing power, which negates the scalability argument for merge-mining.

Don't forget about all the non-mining full nodes that would avoid having to carry the extra burden. Miners are getting financially rewarded for it, so it's much less of a problem for them. And hashing costs dominate up to pretty large scale anyway.

That's the problem! If miners are the only ones with the data, then it's not proof of publication. Of course, this shows a flaw in Bitcoin, but at least for now we've been able to paper over that flaw...

If anything, not being forced to curate the "garbage dump" in addition to the Bitcoin blockchain would enable more people to run Bitcoin full nodes. I understand that a lot of participants are unwilling/unable to fully verify, but hijacking the Bitcoin blockchain for a zillion other often unrelated and often unwelcome applications just makes the problem worse.

I get your point though about parasitic users being incentivised to be, well, parasitic.

Also, even if there is ultimately no technical means to prevent blockchain hijacking, this doesn't mean social pressures can't work to some useful degree. Well-respected developers being vocal about it at the very least this gives less abusive alternatives a PR advantage, or can correct the most egregious abuses - like e.g. mastercoin's initial use of non-prunable outputs.

Quote from: Peter Todd on February 24, 2014, 02:40:01 PM

TXO commitments are just a small part of solving scalability - they only help with long-term storage and actually make bandwidth scalability significantly worse. They do appear to be a good approach to blockchain sharding - tree-chains makes use of them - but the people who have been claiming they represent some scaling breakthrough misunderstand the technology.

Right, I calculated a little while back something like a ~7 fold increase in bandwidth (for authenticated prefix trees), but my understanding was that they are useful because they enable partial verification/fraud discovery and distributed block construction, which spreads the increased bandwidth load over a much larger number of participants. Is this a misunderstanding? Or just an overestimation of the number of extra participants?

Quote from: justusranvier on February 24, 2014, 05:15:05 PM

Back in ancient times, there was a P2P filesharing network called "Mojo Nation" that attempted to use market based resource allocation for bandwidth, storage, and content indexing.

It didn't really work out for several reasons, but it's probably worth dusting off, fixing its price discovery problems, and repurposing for Bitcoin.

I guess I'm ancient, as I'm actually aware of Mojo Nation

Zooko et. al. were talking about implementing accounting a while ago for their more modest successor project, Tahoe-LAFS, though I'm not sure if they're still planning to do that.