--- Log opened Fri Mar 15 00:00:22 2013 00:17 < amiller> jgarzik do you know how i'd go about getting a dataset of all the work done at that time 00:17 < amiller> like shares from pools 02:02 < warren> didn't know this existed 02:02 < amiller> yeah 02:02 < amiller> it's the rocket science research central 02:02 < amiller> we're going to the moon 02:02 < jgarzik> not until amiller blabbed about it anyway ;p 02:03 < warren> and not logged, apparently 02:03 < amiller> this is more of a shunt away from #bitcoin-dev than an exclusive panel jgarzik :p 02:08 < amiller> it could be logged, no one has seen the need to bother 02:09 < amiller> i have logs if you want a copy 02:09 < amiller> warren,^ 02:15 < amiller> good evening 02:15 < cads> 'sir 02:16 < amiller> also see pm 02:16 < amiller> i think you're barking up the wrong tree with the suggestions about AI but it's probably not important 02:22 < amiller> cads ah i'm not sure what question this really answered but i feel like linking to a bunch of papers on the topic of interestellar economics 02:22 < amiller> http://people.csail.mit.edu/rivest/fc97-paper.pdf 02:23 < amiller> first of all ron rivest made a short opinion piece at the first FC conference speculating how future money would be based on computational power and it would have a lot in common with voting 02:24 < amiller> i like a really famous old paper in distributed computing about protocols that scale arbitrarily well http://groups.csail.mit.edu/tds/papers/Lynch/jacm88.pdf 02:24 < warren> shamir? 02:24 < amiller> shamir what? 02:25 < warren> -dev mention 02:25 < amiller> http://eprint.iacr.org/2012/584.pdf 02:25 < amiller> this is a paper by shamir and a colleague dorit ron on an empirical analysis of bitcoin 02:25 < amiller> it was really lame because they didn't really answer any of the questions they posed and those questions weren't worhtwhile in the first place >:| 02:26 < amiller> it has no theory or math or anything in it which is strange given that the authors are reputable academic crypto/math people 02:27 < amiller> still hopefully it's seen as 'breaking the ice'? 02:28 < amiller> the basic idea of partially synchronous networks is that there are some algorihtms that work even if you don't know what the latency is across the whole network 02:29 < amiller> the basic technique is something like exponential backoff - no matter what the real latency is, you 'find it' eventually/quickly 02:31 < amiller> so you said something about wanting to understand the basic limits of global consensus 02:32 < amiller> also it's useful to think about space and long distances if it puts coping with big latency in scope 02:33 < amiller> otherwise you can fantasize about intercontinental network splits or whatever 02:33 < amiller> but a practical reason is that better dealing with latency also implies better use of more efficient networks in a normal environment 02:33 < amiller> already there are other smaller bitcoin knockoffs like litecoin (which now has its own silk road, uh oh) that basically differ only because they go faster 02:34 < warren> amiller: I looked at their SR-clone, it is wayyy faster and a nice web design at least. 02:35 < warren> amiller: I'm concerned of their recent x10 price spike, and possibly more coming, only to be smashed by their complete lack of developers and an unmaintained client. 02:35 < amiller> there's been some thought into relativistic finance that is less interesting than bitcoin but still goes over some basic ideas: http://www.alexwg.org/publications/PhysRevE_82-056104.pdf 02:36 < amiller> that's the most famous one that basically says you want to place your hedge fund headquarters at the right spot between orbiting financial centers you want to possibly arbitrage between... 02:38 < amiller> ah dammit i can't find the last paper i wanted to show 02:39 < warren> I'm sorry for contributing to -dev devolving away from -dev. 02:39 < warren> I can't help it sometimes. 02:40 < amiller> yeah http://www.biosystems.physik.lmu.de/paperpdfs/money_momentum_p_a.pdf 02:41 < amiller> there's a view of how different currencies might interact 02:41 < amiller> it's an analogy to particles and anti particles and conserved quantities thereof 02:42 < amiller> it's interesting in part because it's about transferable credits associated with a particular issuer 02:42 < amiller> so as a financial model it's closer to opentransactions or ripple or even colored coins 02:43 < amiller> i think the two particles interacting is the right way to think about what would happen if coins between multiple chains needed to be transferred 02:45 < warren> you mean randomness 02:45 < warren> two manic depressive entities on each side of a trade 02:49 < amiller> not randomness 02:49 < amiller> more like localized atomic interaction 03:07 < andytoshi> amiller: i think coins across multiple chains is the right way to think about particles ;) 03:10 < amiller> there are some really interesting forms of money 03:10 < amiller> early forms of accounting basically 03:10 < amiller> since multiple currencies is really more about accounting than anything else 03:10 < amiller> http://www.jstor.org/discover/10.2307/40697984?uid=3739704&uid=2&uid=4&uid=3739256&sid=21101834916641 03:10 < amiller> the orginally tally, the 'split tally', is a lot more interesting 03:10 < amiller> than the tally you're probably used to just drawing things on boards 03:10 < amiller> it's a split tally, that sort of resmebles bitcoins scriptsigs and scriptpubkeys 03:11 < amiller> where one piece is needed to be presented to redeem the value promised by the other 03:12 < amiller> there's also an old kind of money called the bulla 03:12 < amiller> https://en.wikipedia.org/wiki/Bulla_(seal) 03:13 < amiller> it was a way of sealing something valuable inside a lump of clay with seals on it so that it remained tamper proof 03:13 < amiller> so it's like carrier money 03:14 < amiller> old testament technology 03:15 < andytoshi> very cool 03:15 < andytoshi> i wish i had time to read up on all this stuff 03:18 < cads> thanks for the papers amiller 03:18 < amiller> glhf 07:43 < warren> So I want to learn how to programmatically interface with bitcoind. I figure it might be helpful to implement a SD-clone on testnet in order to encourage "realistic" garbage tx's to test the blockchain handling as we move forward with 0.8.1 hard-fork testing? Note that I disapprove of SD. I don't post this in -dev because I'm not sure about the legality of running a fake test money gambling service. 07:57 <@gmaxwell> "no no, your honor, testnet money is _double fake_" 07:57 < sipa> is "faking" nilpotent? 08:12 < warren> I realize that might be a stupid worry. I just don't want to broadcast intent like that for something that I would actually do if I'm not sure. 08:12 < warren> about the legality 08:13 < warren> and maybe this won't even be helpful? If so I'll do something else. 08:13 < warren> I just figure people actually USING testnet would be a good thing. 09:07 < warren> gmaxwell: and dude. I'm really sorry I've been annoying you these past few weeks. My stress has been really high trying to do this thesis all this time, and I have to learn better to deal with stress than to unproductively discuss crazy and unproductive things and make dumb jokes. When I'm past this a month from now, I really would like to learn more about Bitcoin and do things that are helpful. I would appreciate guidance if this curren 09:07 < warren> t idea would be helpful or not. 09:09 <@gmaxwell> I don't think it's worth doing if you're not also interested in trying more efficient variations. If you just want to generate txn on the network you can do that without running a service. 09:16 < warren> oh. I figured random user behavior that mimicked the real network would be helpful for a simulation, but I guess it doesn't matter if we're only concerned with the quantity of transactions from block to block. The thing is I'm not really interested in thinking about or helping gambling to be more efficient. I think they should stop spamming the chain entirely, do most things in-house, and only payout when they withdraw. 09:28 < warren> OK, I'll think of something else more useful. 09:34 <@gmaxwell> warren: sure but you could artificially generate the 'random user behavior' in a way that was pretty faithful with basically no more effort, plus the advantage of repeatability and no risk or ambuity about legality. 09:34 <@gmaxwell> e.g. it's only really interesting if the users are the subject of the experiment instead of the system. 09:39 < warren> gmaxwell: Watch the bitcoin mempool and reproduce those txn in testnet of similar quantity, KB size, age and fee behavior? 09:42 <@gmaxwell> yea, even make a model with parameters fit from the actual network. one fun test would be to also try on a private testnet with difficulty 0 (comment out the POW check) and see how fast it can run. 09:43 < warren> Oh. Accelerate the simulation. Don't wait for real-time mining. 09:44 < warren> gmaxwell: was testnet in a box fixed? 09:45 <@gmaxwell> warren: testnet in a box is fine— but testnet in a box doesn't have the pow disabled. You can just isolate some regular testnet nodes and remove the pow check. 09:45 < warren> gotcha 09:45 < warren> g'nite 21:30 < andytoshi> has much thought been given to how an altchain could pay people to run full nodes? 21:30 < andytoshi> or bitcoin, for that matter.. 21:47 <@gmaxwell> andytoshi: the best I have along those lines is "validation is mining", where you have a POW which is a memory hard function based on performing UTXO queries. 22:58 < andytoshi> i wonder if simply charging for IBD would do it 22:58 < andytoshi> though i suppose there's probably no way to get bitcoins for payment until you've got a blockchain :P 22:59 < warren> IBD? 22:59 < sipa> initial block download 23:05 < warren> gmaxwell: how would that POW be transmitted and kept track of? 23:06 < warren> oh. I see. nevermind. 23:22 <@gmaxwell> andytoshi: charging for IBD == ~no one validates. In general the whole idea of blockchain consensus's security involves assuming that an attacker can't partition the network "because information wants to be free"— hard to hide the best chain. 23:27 < warren> gmaxwell: was your earlier comment a suggestion that an alt chain to pay for validation would be a good idea? 23:32 <@gmaxwell> warren: No. Try reading andytoshi's question again. 23:33 < warren> argh. sorry. back to thesis. 23:53 < andytoshi> warren: i'm also working on a thesis, i understand ;) 23:53 < andytoshi> wait, not a thesis, a term paper.. --- Log closed Sat Mar 16 00:00:26 2013