03:44:09tacotime:bitfreak's thing is apparently finally coming out: https://bitcointalk.org/index.php?topic=643758.0
03:45:08tacotime:although i still fail to see how it [finite mini-blockchain] offers advantages over something like maaku's utxo compression method.
05:13:44dsnrk:tacotime: oh no, more naming issues
05:14:32dsnrk:bytecoin now collides with another altcoin in name, and it's PoW collides with bitfreak's cryptonite name
15:43:37andytoshi:after chasing some links i got to http://bitfreak.info/mbc-wiki/index.php?title=Mini-blockchain and http://bitfreak.info/mbc-wiki/index.php?title=Proof_chain from tacotime's post
15:44:47andytoshi:it seems a bit confused, there are statements like "requires the full blockchain because that's the only real way to determine the full balance of all the addresses" and later they describe how they do a PoW on just the hash of the block header instead of the full header to save space to avoid storing "all the block data"
15:52:16andytoshi:but it might be just bad writing.. conceptually what they are doing is storing the last N full blocks then doing SPV from there back to the genesis. so now overwriting N blocks means totally hijacking the history but at least you have to do it in realtime to out-prove the main chain
15:58:34tacotime:andytoshi, yeah, it can be kinda of confusing. basically there are three components: a ledger, a header chain, and a tx chain (attached to the header chain)..
15:59:35andytoshi:seems overengineered.. if they added a commit to the utxo root to every block, then modified the client to only fetch data for the last 10000, wouldn't that accomplish the same thing?
15:59:41tacotime:Basically miners mine on the header chain and produce blocks with a tx tree I don't think they validate (not totally clear on this part)
16:00:02tacotime:Then the network looks at the PoW and tx tree and checks it against the ledger (UTXO set)
16:00:37tacotime:then updates the ledger accordingly. Then they just use the ledger and discard old blocks as they come in.
16:00:49tacotime:andytoshi, yeah
16:01:09tacotime:that's why i was saying i didn't really understand what it had over schemes like maaku's ultimate utxo compression
16:02:25tacotime:you can do it in bitcoin as well by OP_RETURNing the utxo root in the coinbase, as you'd imagine that whoever is signing the coinbase is the same person who did the work for the block, and so should be equivalent in security.
16:02:37tacotime:then you don't have to hardfork or anything.
16:04:06tacotime:i told him this though and he said that it wasn't what he had in mind or something along those lines.
16:04:37tacotime:So maybe I'm missing something, but I don't feel like I am...
16:07:12maaku:gmaxwell's ultimate utxo compression ;)
16:07:32tacotime:oh. :P mea culpa
16:08:37andytoshi:can one of you post a link? bct and the /topic links give several candidates..
16:09:29maaku:andytoshi: link for what?
16:09:51andytoshi:the utxo compression scheme you guys are talking about
16:11:09maaku:andytoshi: here's a good overview from amiller : https://bitcointalk.org/index.php?topic=101734.0
16:12:10maaku:it's just block/coinbase commitments of a deterministic hash-tree of unspent transaction outputs
16:13:09andytoshi:oh, cool
16:13:13maaku:gmaxwell gets dibs on the idea itself, although it was explored and refined by many people. etotheipi coined 'ultimate blockchain compression' which is how most people know the idea, although it is a bit of a misnomer
16:13:37maaku:for the last year i've been researching & prototyping it
16:14:07maaku:eventually settling on a binary prefix tree structure with a compressed representation
16:17:19maaku:tacotime: and yes, as far as I can tell andytoshi is right: you get the same behavior using utxo commitments and SPV validation past a certain number of blocks back
16:17:52maaku:except with utxo commitments the client gets to choose how much history to validate, and the network isn't vulnerable to rewriting the ledger with a massive reorg
16:18:18andytoshi:this notion of dropping block headers and just PoW'ing the hash makes me nervous.. how can you compute work done without knowing the difficulty changes?
16:18:25tacotime:yeah. bitfreak kept insisting his system was different when i talked to him though, but i couldn't figure out how. i meant to go over it on skype with him a while ago, but the whole thing fizzled out as i got bogged down with mc2 and monero.
16:19:23maaku:aside, "ultimate blockchain compression" should refer to snark verification of history :P
16:19:32tacotime:heh.
16:19:34maaku:you can't get better than that
16:19:37tacotime:yeah.
18:02:17Kawhi_desu_:Kawhi_desu_ is now known as Kawhi_desu
18:11:33Guest85836:Guest85836 is now known as qwertyoruiop
20:07:59Guyver2:Guyver2 has left #bitcoin-wizards
20:21:34[nsh]:[nsh] is now known as nsh
20:21:56nsh:nsh is now known as [nsh]