01:18:01wallet42:wallet42 is now known as Guest45798
01:18:01wallet421:wallet421 is now known as wallet42
08:05:19verne.freenode.net:topic is: This channel is not about short-term Bitcoin development | http://bitcoin.ninja/ | This channel is logged. | For logs and more information, visit http://bitcoin.ninja
08:05:19verne.freenode.net:Users on #bitcoin-wizards: andy-logbot rusty terpo justanotheruser moa hashtagg_ koshii wallet42 damethos grandmaster hktud0 b_lumenkraft x98gvyn NewLiberty priidu orik arubi jaekwon TheSeven nullbyte fanquake spinza zooko antgreen Dr-G2 shesek helo jonasschnelli copumpkin sadoshi Tjopper ryanxcharles RoboTeddy p15 nubbins` adam3us c0rw1n Transisto jgarzik btcdrak prodatalab_ d1ggy_ JonTitor go1111111 gielbier yorick harrow r0ach weex Pan0ram1x dgenr8 GAit SwedFTP cluckj
08:05:19verne.freenode.net:Users on #bitcoin-wizards: PaulCapestany Hunger- lmacken dc17523be3 sqt face HM Luke-Jr luny gribble yrashk Xzibit17 artifexd kumavis adams_ platinuum pollux-bts otoburb a5m0 vonzipper catlasshrugged Madars huseby midnightmagic BlueMatt throughnothing kefkius GreenIsMyPepper dasource warren TD-Linux mariorz binaryatrocity hguux___ fenn wizkid057 Starduster rustyn ebfull ajweiss LeMiner waxwing airbreather maaku thrasher` Tiraspol SubCreative starsoccer Adlai jaromil
08:05:19verne.freenode.net:Users on #bitcoin-wizards: Anduck melvster iddo aakselrod MoALTz_ bosma sneak poggy PRab bliljerk101 amiller raizor Chillum wiz nsh luigi1111w kyuupichan Logicwax d9b4bef9 tromp runeks phantomcircuit EasyAt hashtag_ lechuga_ luigi1111 isis nanotube yoleaux gmaxwell andytoshi berndj gavinandresen dignork AdrianG s1w livegnik optimator fluffypony Meeh cursive dansmith_btc morcos guruvan BananaLotus bedeho heath roasbeef_ Fistful_of_Coins dardasaba comboy_ tromp_
08:05:19verne.freenode.net:Users on #bitcoin-wizards: stonecoldpat afdudley espes__ pigeons eric jessepollak sipa warptangent ryan-c mkarrer_ phedny so BrainOverfl0w MRL-Relay azariah btc___ @ChanServ brand0 davout NeatBasis mr_burdell CryptOprah leakypat K1773R indolering veox Eliel Graet gnusha jbenet mappum Keefe Oizopower Krellan Taek epscy null_radix sdaftuar Alanius nickler gwillen wumpus kinlo cryptowest_ coryfields_ Zouppen cfields Muis catcow kanzure petertodd [d__d] jcorgan Apocalyptic
08:05:19verne.freenode.net:Users on #bitcoin-wizards: Iriez lnovy larraboj sl01 null michagogo forrestv STRML
09:08:57bosma_:bosma_ is now known as bosma
11:27:00rusty:Lightning Networks Part III is up. A bit dryer than the other two, but if you got this far... http://rusty.ozlabs.org/?p=467
11:51:51bedeho:rusty: This is a great write-up, will read through
12:15:10CoinMuncher:Very much appreciated! s/Hased/Hashed/
15:21:35runeks__:runeks__ is now known as runeks
15:30:14Guest39245:Guest39245 is now known as amiller_
15:57:14mkarrer_:Is a locktime > 0 standard (relayed)?
15:59:36fluffypony:is #bitcoin-dev a better channel for this?
15:59:50mkarrer_:will try it there
16:31:54maaku_:maaku_ is now known as maaku
21:27:52orik_:orik_ is now known as oirk
21:27:54oirk:oirk is now known as orik
22:07:54unlord_:unlord_ is now known as unlord
22:54:49bsm117532:Who has read this before? http://fc15.ifca.ai/preproceedings/paper_101.pdf
22:55:02bsm117532:It's an idea that's been in the back of my head for a while. They beat me to it.
22:55:19bsm117532:Basically, using a Directed Acyclic Graph instead of a chain (of blocks).
22:57:11phantomcircuit:bsm117532, there's no date on that paper
22:57:14phantomcircuit:i cant stand that
22:58:44kanzure:"i only read papers that include their git commit id in their header"
22:59:04kanzure:but also, the url has some hints
22:59:05fluffypony:kanzure: I thought you only read papers transmitted to you via carrier pigeon or morse code?
23:03:42bsm117532:Yes, see the top level URL, it seems to have been contributed to this conference: http://fc15.ifca.ai/
23:03:56bsm117532:In Puerto Rico. I wish I had gone!
23:04:39bsm117532:I can't find it published anywhere else, I think it's probably still in preparation by the authors, I'll probably write them an email tonight, because I'm thinking of implementing this.
23:04:53andytoshi:it's from late 2014, it is stuff the GHOST people did to improve on GHOST, they presented at FC15 in PR in january
23:05:08andytoshi:i was there but didn't follow it; several other -wizards were who might have a better idea of it
23:05:51bsm117532:GHOST did not change from a chain/tree to a DAG, AFAIK, but if you have refs to the contrary...
23:06:14bsm117532:This paper specifically includes both GHOST and the DAG improvements.
23:20:33bsm117532:Basically, each block has multiple parents, and you allow the same transaction to occur in multiple blocks.
23:22:14bsm117532:With those two, you can basically throw blocks out as fast as you want, and make the software assemble them into a DAG (which they are anyway unless you have a time machine or there's a hash collision). You wouldn't have orphans anymore, chain tips would track conflicting transactions instead, and non-conflicting transactions could be moved to another fork and re-mined into it.
23:23:01bsm117532:There's a ton of advantages to this. You could have a much faster block time, and put a serious dent in miner centralization.
23:36:00Taek:bsm117532: you'd still need some way to limit the total transaction volume per minute
23:36:14bsm117532:Of course. I'm oversimplifying.
23:36:17moa:bsm117532: what limits number of chaintips being divergent?
23:36:53bsm117532:e.g. let each node decide for itself what the minimum difficulty it will accept/relay/mine on.
23:37:35bsm117532:moa: good question. Been thinking about that, don't have a good answer right now. What's bitcoin's limit on tracking orphans/alternate chains?
23:37:54moa:seems like a DDOS issue also
23:38:21bsm117532:It's a DDoS of mined blocks, yes. So it's an expensive DDoS that requires mining hardware...
23:38:40moa:not if mined blocks can have diff=1
23:39:02Taek:It might be good to pair this concept with something like difficulty-per-kb
23:39:12bsm117532:No node would accept or relay those. I'm thinking each node would evaluate for itself what the hashrate is (with a fast hashrate it's easy and accurate) and advertise what it will accept.
23:39:42Taek:I'm pretty sure that would cause consensus risk
23:39:56Taek:you'd need a global rule determining which blocks were acceptable
23:40:03moa:and then you're back into longest chain is best so what else do you need to track?
23:40:42bsm117532:Taek most difficult chain is best.
23:40:45moa:it is an interesting approach though
23:41:09moa:different latitdue
23:42:02bsm117532:I sat down last night to start implementing this, got slightly discouraged by the depth of the required changes, and found this paper today...
23:42:11Taek:bsm117532: 'most difficult chain is best' doesn't help you if you've decided to ignore blocks smaller than diff=X, but the longest chain contains those blocks
23:42:46Taek:or rather, as a miner you get put at a disadvantage for ignoring them
23:43:36bsm117532:Taek: depends on how you combine difficulty from two blocks. If you do it such that D(x)+D(y) = 2*D(x) then the miner could have produced a block of the same difficulty in twice the time.
23:43:56bsm117532:One can imagine other ways to combine it...
23:44:24Taek:where D(x) < D(y) ?
23:44:38bsm117532:D(x)=D(y) is 1/hashrate for the miner.
23:45:01bsm117532:Point is, miner would gain no advantage by throwing blocks out twice as fast with half the difficulty.
23:45:37bsm117532:This part my idea is not fully fleshed out... not sure how much overlaps with the above paper...
23:46:58Taek:sorry if I'm misunderstanding something but having 2x the blocks might be an advantage if you know a minority of your competition can't keep up with the bandwidth
23:47:31bsm117532:Yes. The block rate becomes a dynamic parameter that reflects the bandwidth of participants.
23:48:43psgs_:psgs_ is now known as psgs
23:48:44moa:and issuance?
23:49:38bsm117532:mao you mean coinbase reward? Would have to change obviously. A fixed target block rate might be desirable anyway for that reason unless one came up with a different issuance scheme.
23:50:28moa:well difficulty/block rate/issuacne are all coupled in btc
23:51:17bsm117532:Let's just be really dumb about it and say you get a block reward proportional to the difficulty of the block you mined.
23:51:31Taek:that sounds reasonable enough to me
23:51:52bsm117532:That reward would grow in time with the hashrate but...
23:51:52moa:sounds exactly like hashcash
23:52:27bsm117532:Except for the DAG part...
23:52:43Taek:well, you'd still have an adjustor. It would just tally the total weight of the blockchain every X work and determine what the new reward per work is
23:52:44moa:yeah i meant in the more hashes=morecashes sense
23:53:33Taek:and then you throw on an extra limit that says blocks can have at most Xkb per work, and then gets retargeted at the same itme
23:53:53bsm117532:I really dislike having the network need consensus on the adjustment. If we got rid of orphans like this, each node can make his own decisions about what difficulty he will take, and adjust it as he sees fit. Most difficult chain still wins.
23:54:38Taek:with having a specific attack in mind, that just seems like something that would favor more powerful miners
23:54:40bsm117532:Needing consensus on the target brings clocks into the game.
23:54:57Taek:yes the reliance on clocks is unfortunate
23:55:06bsm117532:Taek: yes, that's why I mentioned that this proposal for the block reward was "dumb"
23:56:03moa:i think it is fundamental isn't it?
23:56:13moa:as in unavoidable
23:56:40Taek:moa: I don't see why it would be fundamentally necessary
23:56:47bsm117532:No, Satoshi conflated issuance and consensus. I could premine all coins and distribute them a different way if I wanted to.
23:57:12moa:seem to recall that from some previous analysis but forget the chain of reasoning ...
23:57:27moa:would need to revisit
23:58:19bsm117532:Mind you, I think that conflation was a good idea for initial distribution, but they're logically separate things. I think it also gave a lot of people the terrible idea that this "mining" activity was somehow "generating" value.
23:58:49moa:distributed timestamp server
23:59:05bsm117532:What about them?
23:59:54moa:that's the clock in the game