01:08:01 | fanquake_: | fanquake_ is now known as fanquake |
06:26:27 | arowser1: | arowser1 has left #bitcoin-wizards |
08:05:14 | card.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 |
08:05:14 | card.freenode.net: | Users on #bitcoin-wizards: andy-logbot speshul wallet42 vmatekole chris2000 damethos llllllllll AaronvanW koshii jtimon justanotheruser NewLiberty mmozeiko RoboTeddy moa mirclient TheSeven Graftec stonecoldpat devrandom forrestv OneFixt_ Dr-G3 fanquake jaekwon ebfull eristisk aburan28 samson_ waxwing Luke-Jr SDCDev gavinandresen erizo Adlai wiretapped napedia SomeoneWeird Sangheili rfreeman_w d4de^^ spinza go1111111 Logicwax rdponticelli midnightmagic hollandais copumpkin |
08:05:14 | card.freenode.net: | Users on #bitcoin-wizards: atgreen artilectinc emsid Anduck gnusha_ espes__ epscy CryptOprah tromp_ pi07r asoltys NikolaiToryzin Graet Nightwolf tacotime iddo mortale Guest21768 DoctorBTC zenojis Max_H3adr00m EasyAt wizkid057 gribble MRL-Relay shesek altoz HaltingState Emcy dansmith_btc2 nanotube Starduster burcin Meeh LarsLarsen optimator_ wumpus Grishnakh berndj [d__d] livegnik nsh Dyaheon Fistful_of_coins Krellan harrow throughnothing abc56889 lechuga_ phedny @ChanServ |
08:05:14 | card.freenode.net: | Users on #bitcoin-wizards: fluffypony mr_burdell Apocalyptic kinlo starsoccer so ahmed_ Gnosis Keefe pajarillo roasbeef ryan-c otoburb helo [Tristan] TD-Linux catcow danneu UukGoblin poggy coryfields kgk btc_ crescendo amiller Eliel zibbo_ jbenet mappum yoleaux firepacket jrayhawk_ Hunger-- Muis Kretchfoop weex bbrittain tromp michagogo artifexd sl01 @gwillen BlueMatt smooth CodeShark digitalmagus BrainOverfl0w BigBitz petertodd bobke andytoshi a5m0 heath kumavis |
08:05:14 | card.freenode.net: | Users on #bitcoin-wizards: phantomcircuit K1773R Alanius hguux dgenr8 lnovy LaptopZZ_ _2539 kanzure Taek gmaxwell sipa pigeons [\\\] coryfields_ warren comboy [Derek] Iriez |
08:09:55 | speshul: | speshul has left #bitcoin-wizards |
09:50:38 | Pan0ram1x: | Pan0ram1x is now known as Guest89609 |
10:59:03 | mircclient: | mircclient is now known as skinnkavaj |
11:12:30 | fanquake_: | fanquake_ is now known as fanquake |
11:38:09 | phantomcircuit: | phantomcircuit is now known as Guest16666 |
12:48:40 | c0rw1n_: | c0rw1n_ is now known as c0rw1n |
12:56:40 | c0rw1n_: | c0rw1n_ is now known as c0rw1n |
14:02:08 | nuke1989: | nuke1989 has left #bitcoin-wizards |
15:12:09 | Taek: | Is there a discussion somewhere on incentive attacks? For example, what happens if someone puts a bribe out for X btc for miners who produce blocks following a particular fork? |
16:12:53 | gmaxwell: | Taek: it's been discussed on bitcointalk before several times... I'm not sure what string to best search for, there was also a recent bitcoin-development thread where it came up. |
16:15:09 | gmaxwell: | The answer is... today, nothing.. probably also not much in the future; but we're not sure. Those kind of attacks, at least when bitcoin is the only asset being reversed, are only rational for the miners under very short term thinking.. "congrats you got the bitcoin, but have throughly eroded trust in the system". Automatically acting on them requires a fair amount of software sophication that no one has bothered with. |
16:16:12 | gmaxwell: | It's one of the reasons I consider pools and hosted mining to be greater risks though, since they transfer control of hashpower to people without the long term investment in the equipment. |
16:39:16 | rdponticelli: | rdponticelli has left #bitcoin-wizards |
18:13:55 | hearn__: | hearn__ is now known as hearn |
19:26:21 | hearn: | http://eprint.iacr.org/2014/838 - not exactly bitcoin related but still pretty amazing |
19:26:32 | hearn: | 100x speedup for homomorphic encryption, when a GPU is in use |
19:26:45 | hearn: | they used it to build a Bayesian spam filter that operates over encrypted messages |
19:26:55 | hearn: | * hearn is suitably impressed |
20:05:47 | gmaxwell: | wow, FHE ... actually doing something interesting! |
20:16:23 | hearn: | for small messages |
20:16:31 | hearn: | but yes - great progress. and scales perfectly as you add GPUs |
20:31:07 | gmaxwell: | hearn: sadly, its not the FHE scheme that has the decryptor... the combination of the two gives you program obfscuation; e.g. a program that will only sign certian transactions, from which you cannot extract the embeded private key. |
20:31:24 | gmaxwell: | which I think is a task of vaguely similar computational complexity to filtering a small email message. |
20:31:51 | hearn: | without iO? |
20:31:58 | hearn: | i didn't realise you could build an obfuscator with fhe |
20:32:47 | gmaxwell: | input to FHE is no big deal; that just works (give people encrypted zeros and ones to use as inputs. Outputs is the hard part. Google 'candidate indistinguishability obfscuation' ... basically they come up with a decryptor that can decrypt the output of fhe. |
20:32:57 | gmaxwell: | (of a particular fhe scheme) |
20:33:08 | gmaxwell: | so the composition is obfuscation (hopefully) :) |
20:33:53 | gmaxwell: | ) |
20:38:50 | andytoshi: | the decryptor was NC^1, and they had an iO candidate for NC^1 ... the FHE was to bootstrap to poly circuits |
20:38:53 | andytoshi: | is that what you're talking about? |
20:51:15 | hearn: | gmaxwell: my understanding was that iO was not based on FHE exactly but rather this multilinear jigsaw puzzles trick applied to matrix branching programs |
20:51:18 | hearn: | or that was the starting point anyway |
20:51:29 | hearn: | perhaps it's now FHE applied to matrix programs |
20:54:51 | gmaxwell: | what they do is come up with a decryptor for FHE that can work in a very narrow kind of circut (NC) that they can obfscuate. |
21:00:42 | andytoshi: | it's the NC obfuscator that uses multilinear jigsaw puzzles ... you are both right :) |
21:02:25 | gmaxwell: | oh sorry, sometimes I sound like I'm disagreeing. |
21:22:51 | hearn: | ah yes i remember now, the final step where they leverage it up to full blown obfuscation |
21:29:54 | gmaxwell: | yup |