00:00:16c0rw|afk:c0rw|afk is now known as c0rw1n
00:15:31midnightmagic:lmatteis: I didn't see anyone answer your comment about embedding p2pool within bitcoin core. Yes, lots of people have said it, and gavin has mentioned he would be completely in favour of a C++ conversion of p2pool for inclusion with bitcoin-core. I suspect a separate daemon would be most palatable, but the answer is yes. And I suspect if it were cleanly-written, it would likely not have muc
00:15:38midnightmagic:h difficulty getting ACKs from the
00:15:40midnightmagic:* midnightmagic stabs splitline
00:15:42midnightmagic:universal developer set.
00:16:13kanzure:what are the arguments against including the python implementation?
00:16:55midnightmagic:kanzure: probably dependencies. There were no arguments about it, as far as I know everyone else agreed a C or C++ implementation would be a superior direction.
00:18:15midnightmagic:kanzure: I suspect the fact that universal p2pool mining would push the canonical p2pool chain difficulty so high it would no longer be an effective reduction in mining variance and therefore unpopular. Probably manual subscription or semi-automatic subchain membership organization would solve that.
00:18:43gmaxwell:kanzure: so p2pool is ... not very healthy right now. It's more or less unmaintained, and the performance is not great. (e.g. uses several hundred megs of ram). AFAICT forrestv burned out on bitcoin when people with crappy centeralized pools were earning a hundred grand a month, and his major engineering feat bairly got enough for him to buy an asic miner (which then was never shipped).
00:18:47kanzure:oh; yeah i guess dependencies would be bothersome, even when bundled into a release.
00:18:54kanzure:also was unaware of forrestv burnout.
00:19:08gmaxwell:midnightmagic: it always results in variance reduction, right now the low hashrate on it overwhelmingly dominates, current users would still be better off with everyone using it.
00:19:35gmaxwell:(they wouldn't get paid in every block, lots of share variance, but much less pool variance)
00:19:43midnightmagic:forrestv is extremely cooperative as a developer, and a very high-quality human in general.
00:20:01midnightmagic:(at least what I've seen of him the last five years)
00:20:02jcorgan:not many of those
00:20:10belcher:too bad about the incentives around open source development :|
00:20:22gmaxwell:Forrestv is a great guy and frighteningly smart p2pool code is too "clever" for me to hack on easily; it's full of python metaprogramming.
00:20:30belcher:in that situation being evil is more profitable
00:20:34zooko:gmaxwell: ouch
00:20:50zooko:After we fix money, we need to fix economics.
00:21:03zooko:* zooko drank a glass of red wine
00:22:19kanzure:i thought aging was next on the agenda
00:22:23kanzure:maybe i have the old agenda
00:22:44zooko:Heh heh heh. We definitely need to fix that.
00:22:53midnightmagic:I like p2pool's code, personally, although the lambda stuff was the first time I've ever encountered it in python code. Spoiled by commercial LCD python programming I guess.
00:22:55zooko:* zooko joins #bitcoin-wizards-offtopic to fix aging
00:23:02gmaxwell:In any case, since p2pool is a consensus network itself, its a bit tricky to maintain! (though forrestv managed to upgrade it a BUNCH of times)
00:23:33zooko:forrestv: is it true you felt burnt out by the experience?
00:23:37zooko:forrestv: and what are you doing now?
00:23:46gmaxwell:oh he's in here?! crazy!
00:23:59zooko:* zooko peers at forrestv
00:24:02zooko:Looks pretty idle.
00:24:20gmaxwell:* gmaxwell misses talking to forrestv
00:24:26midnightmagic::-) that's why I used his name when I said nice things about him. :)
00:26:33zooko:If I order another glass of red wine, I'm /part'ing from this channel for the night.
00:27:01gmaxwell:zooko: oh you got one of those keyboard breathalizers, ... so much more useful than those fingerprint readers.
00:28:13zooko:Heh heh heh.
00:28:16zooko:That's a great idea.
00:28:23zooko:I wonder if I'll find one if I google for it.
00:28:28zooko:Maybe search kickstarter.
00:28:39kanzure:here's your term sheet
00:28:52kanzure:i spilled some coffee on it and i drew with some crayons on it, but it's still good i think
00:29:31zooko:I *am* actually working on an investment contract right now, while drinking, and IRC'ing.
00:29:38zooko:Maybe I should send some emails and tweet while I'm at it.
00:29:44gmaxwell:zooko: software is the new hotness, can probably use the camera to detect flushing.
00:29:51moa:got a bot to do that
00:30:17zooko:I wonder if there is a keystroke-timing equivalent to slurred speech.
00:30:27gmaxwell:via this DSP technique: http://people.csail.mit.edu/mrub/vidmag/
00:30:43dgenr8:didn't see this mentioned here in a quick search http://eprint.iacr.org/2015/578
00:31:42gmaxwell:dunno if it was mentioned here, I've seen it. The 20 minute stuff is well known (see #bitcoin-dev logs), made less of a concern by immune networks between miners, but will be improved in later versions.
00:31:54gmaxwell:I haven't read the full thing in details.
00:32:01kanzure:.title http://eprint.iacr.org/2015/578
00:32:02yoleaux:Cryptology ePrint Archive: Report 2015/578
00:32:13kanzure:"Tampering with the Delivery of Blocks and Transactions in Bitcoin"
00:33:49dgenr8:if we have an inv, and get a tx push for that hash, why not accept it?
00:35:41gmaxwell:dgenr8: I believe we do (did?) accept unsolicited TX pushes; thoug there is a good argument why not to: it strongly encourages crappy implementations that don't INV (and can't inv) and thus waste a bunch of bandwidth.
00:36:21kanzure:(in other off-topic research) "Meta-analysis of heritability of human traits based on 50 years of twin studies" http://www.gwern.net/docs/2015-polderman.pdf
00:36:58gmaxwell:kanzure: man, I misread that as meta-heritability and was momentarily perplexed.
00:37:50nsh:i just spent a few seconds doubting the utility of delineating traits in the first place
00:38:40midnightmagic:need a clean way to encapsulate a zotero single-paper archive.
00:39:00dgenr8:this problem would go away even if we accepted a push only in the case it was previously asked for from another peer
00:39:03midnightmagic::-( this plain-pdf on its own stuff is Not Da Best Kung Fu
00:39:50gmaxwell:dgenr8: I don't see how, peers shouldn't be sending unsolicited transactions in the first place; so how is a behavior going to be fixed by changing the handling of something that shouldn't happen?
00:40:27dgenr8:what are the improvements you mentioned?
00:45:06dgenr8:oh for one thing they claim the incoming inv buffer from a single peer is not de-duplicated. that should be easy to fix
00:47:48dgenr8:they are probably using multiple peers and just didn't state it
00:56:18amiller:that tampering with delivery paper repeats a bunch of stuff i already mentioned in http://cs.umd.edu/projects/coinscope/coinscope.pdf
00:56:24amiller:including that the inv buffer is not deduplicated
01:05:53c0rw1n:c0rw1n is now known as c0rw|zZz
01:10:01tdryja:tdryja has left #bitcoin-wizards
01:16:05dgenr8:ah. it's not even keyed by hash so, yep
01:20:30dgenr8:i'm thinking AddInventoryKnown should pay attention to whether it was already there and continue if yes
01:28:05dgenr8:that gets it down to just 2 minutes (which is still too long)
01:28:45dgenr8:amiller: haven't read that, now looking forward to it!
01:37:13moa:what's with all these AI movies these days, seems to be a deluge of them coming out
01:38:02moa:i'm sure one of the dialogs in ex machina was ripped off -wizards logs
01:40:53moa:... and I'll be running the Chappie fork from now on btw
01:41:38gmaxwell:moa: You didn't notice that the main computer genius guy looks like a buff eliezer yudkowsky?
01:44:26moa:well he was buff ... and an alcoholic
01:45:00moa:but his last line was a classic ... stumbling down the hallway wife knife protrduing from chest "effing unreal dude"
01:48:36gmaxwell:I liked the movie, for whatever is worth; also in accepting it for what it is. but hah. ending wasn't that awesome though.
01:49:51moa:yeah, take it for what it was worth .... but it makes one wonder unless AI has it's own "great filter" then the fermi paradox great filter might be AI, etc
01:50:01jgarzik:They had a private screening of Ex Machina at the Singularity Conference where I spoke last week. After the screening, a private Q&A with MSFT A.I. expert and another A.I. expert
01:50:06jgarzik:...and I missed it :/
01:51:04moa:murdered by the AI you created
01:52:48gmaxwell:moa: well that was part of the point; AI is potentially an existential risk to humanity. I think our expirence with Bitcoin informs that even the AI doomssayers are probably very naieve about how far away mankind is from being able to produce complex systems that do as intended, too.
01:55:16moa:fast-learning adaptive systems can go off the rails just as fast as they can achieve amazing things
01:56:51moa:AI need good sleep cycles and a naughty corner to consider the consequences of their actions
01:58:11moa:a trip to the zoo and some ice cream might not be sufficient on the carrot side of the equation though
01:58:42gmaxwell:moa: I think I may have mentioned before I tried to come up with a risk metric for adaptive systems. The idea, I thought is you can consider the universe a probablity space and then ask how unlikely an event the process could bring about, and use that as an estimate of danger. But after not-much reflection I realized simple processes like the phased lock loop that brings my local atomic clock
01:58:48gmaxwell:in sync with UTC from GPS is able to cause synchronization with error <= 10e-13 level quite quickly, which is impossibly good alignment from a probablity perspective... so by that kind of metric like ... everything would be very dangerous, so not that useful.
02:05:33moa:maybe a better starting point is that the universe is an incredibly dangerous place, everything is dangerous it is a miracle every second we stay alive?
02:06:03moa:life and living is the low probability event
02:06:34gmaxwell:well we already know life is dangerous!
02:06:43moa:yup, it's axiomatic then
02:07:29gmaxwell:I think that intution is not completely awful, e.g. it correctly detects a bunch of dangerous things (people) and not very dangerous things (rocks)... but unfortunately it also calls many things that are probably not very dangerous dangerous, like a phase locked loop syncing oscillators.
02:07:35jgarzik:bah, I have a 100% track record of not-dying
02:07:39jgarzik:seems safe to me
02:08:12moa:i could walk through the plate-glass window in front of me right now ... look around you and ask how many ways you could be dead inside a minute
02:08:46moa:if my circuits got scrambled all that danger comes into play
02:15:43moa:sanity is balancing a highly complex system inside a very narrow phase space of relative safety, surrounded by a domain of mostly danger life-ending motions
02:16:23moa:then you hop in a car and drive down the freeway at 70mph to spice it up :)
02:16:59moa:on your way to go wing-suiting in the alps for the weekend
02:54:03kanzure:gmaxwell: i think we need better strategies than that (e.g. we really should have not-all-your-eggs-in-one-basket redundancy so that if we turn out to be in a world where very dangerous things are highly probable and easily achievable, then we are less entirely screwed)
02:54:30kanzure:*such that we are less entirely screwed
02:55:41kanzure:((if your form of "dangerous" is such that it can always usurp any possible solution, then it's not a useful problem definition because it provides no utility for what to bother doing with our time, if anything specific))
03:08:40the_last:rusty: nice blog post
03:09:30rusty:the_last: thanks!
03:43:32moa:rusty: so ~9mins to propagate 8MB blocks?
03:43:46moa:e.g. on your outback link
04:13:32rusty:moa: heh, this is fifth largest city in Australia, 1.25M population. But, yeah :(
04:14:34moa:yeah, that was tongue-in-cheek ... 11Mb down 1 Mb up is pretty standard DSL
04:31:07rusty:moa: fastest available, unf. Doesn't seem an unfair basis if you're thinking about a global userbase.
04:31:14rusty:rusty has left #bitcoin-wizards
04:46:11moa:rusty1: I agree ... and I think many people will be surprised with how bad upload connectivity really is for network end users
04:46:51gmaxwell:it's also roughly my connectivity.
04:48:22moa:supposedly the old 700 MHz analog tv bands are going to support 4G equipment to us in the NZ bush sometime in near future
04:51:12Luke-Jr:moa: aren't they "open to all" now?
04:51:48Luke-Jr:* Luke-Jr ponders difficulty level of setting up a SDR to use an open channel for 4G <.<
04:51:49moa:ha, i wish ... they had an unseemingly fast auction and ignored calls to keep some of it open for public use
04:52:30moa:I was envisioning a mesh network using old UHF that every house still has rusting on their roofs
06:12:40jgarzik_:jgarzik_ is now known as jgarzik
06:44:55fluffypony:moa: I reckon cheaper satellite will be deployed before that happens
06:49:47moa:fluffypony: satellite comms is hard, tv antennas have been done for over 80 years
07:07:56jgarzik:fluffypony, moa: Q3 2016 for Dunvegan Space's satellites w/ blockchain on board...
07:08:16phantomcircuit:jgarzik, er how much bandwidth does that have?
07:09:25jgarzik:phantomcircuit, 1 Mb/s - not a comm satellite, more a server in space - http://dunveganspace.com/assets/bitsat-infosheet.pdf
07:16:35fluffypony:jgarzik: nice
07:17:24fluffypony:moa: I know, I was thinking more about Elon Musk's thing
07:17:48jgarzik:If you think you have a customer use case, let us know! jgarzik@dss.co This far in advance of launch, we are lining up people interested in storing data or running apps on the satellites.
07:18:05jgarzik:Run fun stuff on there like bitcoin oracles
07:18:21phantomcircuit:jgarzik, 1Mb/s theoretical or actual?
07:18:34jgarzik:phantomcircuit, 1 Mb/s theoretical max
07:18:37fluffypony:jgarzik: can it do two-way, or one-way only?
07:18:41jgarzik:2 way
07:18:48phantomcircuit:jgarzik, any idea what actual will be one way?
07:19:01jgarzik:actually more than 2-way since we have inter-satellite relay
07:19:06jgarzik:in addition to ground<->sat
07:19:18jgarzik:phantomcircuit, close to that based on bench tests
07:20:32moa:jgarzik: any limits on apps that can be run on your sats?
07:20:53moa:physically, legally
07:21:04jgarzik:moa: whatever the sat can run & doesn't get me arrested ;p
07:21:26fluffypony:* fluffypony reckons moa is designing porn-o-lites
07:21:58moa:jurisdictional arbitrage gets interesting in space
07:22:08jgarzik:our net can do worldwide broadcasting, once all 24 are up
07:24:16jgarzik:* jgarzik -> sleep
08:05:18wilhelm.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
08:05:18wilhelm.freenode.net:Users on #bitcoin-wizards: andy-logbot rubensayshi _biO_ antanst adam3us gill3s damethos AaronvanW SDCDev CoinMuncher Mably davi fanquake1 p15_ www jgarzik mjerr GAit nessence bliljerk_ [7] afk11 dgenr8 justanotheruser Dr-G Zooko-phone d1ggy_ cdecker pollux-bts waxwing moa isis kinlo nephyrin EasyAt phantomcircuit tromp hashtag_ rustyn badmofo afdudley Cory elastoma gmaxwell sparetire_ LeMiner cryptowest_ spinza dc17523be3 helo hulkhogan_ MrTratta gielbier triazo arubi
08:05:18wilhelm.freenode.net:Users on #bitcoin-wizards: melvster SubCreative Tebbo lmatteis Tiraspol GreenIsMyPepper mengine the_last koshii goregrind _whitelogger mkarrer go1111111 jmcn PRab heath mm_1 dasource superobserver paveljanik grubles kyuupichan Logicwax catlasshrugged_ Starduster_ Adlai yoleaux Relos [d__d] ebfull luny stqism ThinThread jouke cluckj Emcy Madars sl01 jrayhawk grandmaster OneFixt tromp_ alferz c0rw|zZz tucenaber sneak sadoshi andytoshi MoALTz maaku jcorgan sparetire bosma
08:05:18wilhelm.freenode.net:Users on #bitcoin-wizards: sundance bsm117532 CodeShark copumpkin K1773R ttttemp gwillen lclc STRML akstunt600 kanzure midnightmagic @Luke-Jr Iriez fenn gnusha bedeho wizkid057 akrmn mariorz mikolalysenko Meeh Krellan ajweiss so cornusammonis s1w poggy lnovy HM dansmith_btc catcow btcdrak Xzibit17 prosodyContext vonzipper adams__ yrashk CryptoGoon mappum artifexd Muis runeks kumavis platinuum jbenet yorick amiller fluffypony livegnik mountaingoat a5m0 Apocalyptic wiz
08:05:18wilhelm.freenode.net:Users on #bitcoin-wizards: wumpus Alanius iddo forrestv theymos Taek null_radix smooth narwh4l thrasher` otoburb Keefe weex pigeons sturles rasengan berndj harrow qawap stonecoldpat davout jessepollak huseby espes veox comboy stevenroose gavinandresen nickler ggreer harrigan scoria brand0 larraboj nsh jonasschnelli leakypat epscy cfields coryfields azariah warptangent TD-Linux crescend1 Zouppen binaryatrocity BananaLotus optimator Eliel mr_burdell throughnothing_
08:05:18wilhelm.freenode.net:Users on #bitcoin-wizards: Fistful_of_Coins Jaamg xabbix dignork petertodd richardus SwedFTP guruvan nanotube warren sdaftuar eric roasbeef morcos merlincorey [ace] jaromil Graet indolering ryan-c gribble d9b4bef9 starsoccer BlueMatt Anduck AdrianG BrainOverfl0w @ChanServ
11:17:34antanst:antanst has left #bitcoin-wizards
17:55:20jgarzik:Anything overly nerd-y going on in SF this weekend? I'm around with - shocker - no meetings or calls
17:55:27jgarzik:maybe I'll go see Ex Machina
18:11:11hearn:jgarzik: it's good
18:11:14hearn:jgarzik: if you like cerebral sci fi
18:11:38hearn:jgarzik: the google analogies are a bit too direct imo but other than that it's quite engaging :)
19:18:40justanotheruser:justanotheruser is now known as justanotherusr
19:54:20bosma_:bosma_ is now known as bosma
21:06:23Mably__:Mably__ is now known as Mably
21:15:51cpacia:cpacia has left #bitcoin-wizards
21:40:42leakypat:jgarzik: what is the status of bip100 ? Is it officially a bip now?
21:41:47jgarzik:leakypat, BIP draft coming in a few days
21:41:59jgarzik:leakypat, It's officially a BIP when it's in the BIP drafts repo
22:11:48leakypat:jgarzik: great, thanks
23:05:10c0rw|zZz:c0rw|zZz is now known as c0rw1n