00:52:53amiller:fwd: fwd: #bitcoin-wizards http://i.imgur.com/Ww1APBo.jpg
00:53:36phantomcircuit:amiller, ha
00:58:50gmaxwell:amiller: buring ethylnol vapor off a wet dollar bill?
00:59:10amiller:yeah that's exactly what i thought of first
12:37:23weber.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
12:37:23weber.freenode.net:Users on #bitcoin-wizards: andy-logbot waxwing p15 Quanttek Starduster Ursium sipa jaromil_ Guyver2 cym hearn llllllllll RoboTeddy mortale todaystomorrow fanquake Hunger- justanot1eruser adam3us Dr-G HaltingState tromp mapppum TheSeven ebfull eslbaer_ dgenr8 bsm117532 melvster nuke1989 zibbo shesek SDCDev tromp_ koshii fierbuq postpre Graftec mkarrer Alanius Keefe go1111111 copumpkin jchp alferz skinnkavaj pi07r nsh bangsnap wiretapp1d forrestv Luke-Jr jgarzik rfreeman_w
12:37:23weber.freenode.net:Users on #bitcoin-wizards: OneFixt harrow tacotime Krellan K1773R nsh- phantomcircuit Guest53956 CryptOprah pigeons EasyAt iddo cfields nickler_ gwillen [\\\] grandmaster2 azariah4 samson_ bobke drawingthesun starsoccer midnightmagic mr_burdell Graet HM CodeShark Logicwax maaku Adohgg polyclef Muis LarsLarsen1 Sangheili Anduck xenogis zling_____ michagogo Eliel helo crescendo epscy mmozeiko Guest50253 Transisto [Derek] asoltys berndj-blackout BlueMatt digitalmagus7 sl01
12:37:23weber.freenode.net:Users on #bitcoin-wizards: weex Iriez abc56889 espes__ lechuga_ SomeoneWeird bbrittain nanotube rs0 davidlatapie Guest78271 hollandais grishnakh jbenet quackgyver poggy_ TD-Linux gmaxwell Meeh a5m0 BigBitz tjopper spinza catcow amiller Fistful_of_Coins dansmith_btc DoctorBTC danneu2 LaptopZZ_ Dyaheon- burcin optimator_ jcorgan [d__d] ryan-c kanzure petertodd UukGoblin wizkid057 nkuttler wumpus lianj Apocalyptic @ChanServ BrainOverfl0w throughnothing warren artifexd
12:37:23weber.freenode.net:Users on #bitcoin-wizards: pajarillo comboy roasbeef gribble Guest10516 phedny so kinlo
12:37:23weber.freenode.net:[freenode-info] if you're at a conference and other people are having trouble connecting, please mention it to staff: http://freenode.net/faq.shtml#gettinghelp
16:09:00warren_2:warren_2 is now known as warren
16:09:28BigBitz_:BigBitz_ is now known as BigBitz
16:10:01michagogo_:michagogo_ is now known as Guest900
16:10:25Muis_:Muis_ is now known as Muis
16:14:19artifexd_:artifexd_ is now known as artifexd
17:05:28o3u:o3u is now known as Fistful_of_coins
18:26:27danneu2:danneu2 is now known as danneu
19:51:21mappppum:mappppum is now known as mappum
21:15:10helo:helo is now known as james_russel
21:15:44james_russel:james_russel is now known as helo
21:16:53Guest900:Guest900 is now known as michagogo
21:20:44ben99:hello. hwas someone knowledge on GHOST?
21:22:31ben99:or rather: I have some general question on latencies and distribution of latency-skew. anyone studied this?
21:26:06pigeons:standard procedure is to go ahead and ask the question instead of asking if anyone knows about the general topic
21:27:55ben99:well, it's complicated and I'm not going to write an essay if nobody answers...
21:28:30nsh:funfact: you can re-use your detailed explanation of the problem/query when asking elsewhere
21:28:45nsh:there is very little to lose by asking effectively :)
21:30:03nsh:even if you speak into the wind, a clear and thorough formulation will often facilitate your own understanding
21:30:32ben99:ok, I'll try it this way: the idea of GHOST is to have what they call Subtrees. this should violate principle of global view of block-chains. one crucial problem with block-chains is that all nodes have to act on the same information
21:31:23ben99:if nodes can be thought of as a graph with each edge being a physical distance. then the block-time relates to the distances
21:32:11ben99:I haven't seen any study of this, except the original Lamport paper's and Nick Szabo's essay on advances in distributed security (2005)
21:32:23ben99:but is IMO crucial for scaling block-chains
21:33:20ben99:ethereum proposes to use GHOST, which is flawed
21:35:52nsh:how does the block-time relate to the [physical] distances between nodes?
21:36:40ben99:it takes time for information to travel. block-chain solves problem of creating a universal partial order of events (Lamport's partial order)
21:37:09nsh:that means the period between blocks is bounded from below by propagation delays
21:37:15nsh:but that's always going to be the case
21:37:45ben99:yes. TCP/IP latencies has high variance. for example imagine a broad-cast by radio-waves
21:37:51nsh:what you can do is have multiple overlapping regions of consistency is you want to increase the granularity, but that comes at the cost of increased complexity
21:38:22nsh:*if you
21:39:02nsh:jgarzik, i think, proposed some kind of continuous rolling reorg thing involving directed acyclic graphs but i forget the details
21:39:33ben99:as far as I can see the crucial problem is that block-chains are global
21:40:20ben99:so every node has to confirm all transactions. its exponential scaling by default so to speak
21:41:05nsh:local consensus isn't consensus. delegated confirmation isn't decentralized
21:41:37nsh:there are good reasons for bitcoin forsaking scalability to ensure distributed global consensus
21:41:40ben99:yes. not necessarily in principle
21:42:01ben99:say you had N regions, and each region has its own distributed consensus
21:42:21ben99:kind of a tree consensus
21:42:55ben99:where roots is not one node, but a larger number of nodes (100 e.g.)
21:44:38ben99:what confuses me is that some very smart bitcoin people like some of the ethereum people believe you can do local trees
21:44:59ben99:seems therefore some of this is not well understood
21:47:31nsh:andytoshi, are the logs not google indexed? :/
21:47:44nsh:oh, they're just not hosted on bitcoin.ninja
21:50:03nsh:oh, it was petertodd's idea, not jgarzik's (pardon me, both)
21:50:18nsh:( http://download.wpsoftware.net/bitcoin/wizards/2013/12/13-12-10.log )
21:53:41gmaxwell:ben99: I don't think the implications of ghost are not really well understood, when the paper came out I was able to come up with some nasty corner cases involving subtree withholding where you could creating tied forks that persisted for a long time with relatively little hashrate. Amiller had proposed ghost like things a couple years prior with the goal of making interblock times closed loop controlled by orphans, but most people ...
21:53:47gmaxwell:... didn't seem to consider it very interesting because driving up to minimum latency is a somewhat obviously centeralization pressure.