| --- Log opened Thu Oct 18 00:00:12 2007 |
| 00:15 | <@Vornicus> | ARGH |
| 00:15 | <@Vornicus> | so, a friend of mine has gotten me hooked on Harvest Moon. |
| 00:16 | <@Vornicus> | The first thing I want to do? /remake it so I can fix the thinkos/ |
| 00:58 | | gnolam [lenin@Nightstar-10613.8.5.253.static.se.wasadata.net] has quit [Ping Timeout] |
| 01:00 | | gnolam [lenin@Nightstar-10613.8.5.253.static.se.wasadata.net] has joined #Code |
| 01:00 | | mode/#code [+o gnolam] by ChanServ |
| 01:01 | | You're now known as TheWatcher[T-2] |
| 01:02 | | You're now known as TheWatcher[zZzZ] |
| 01:44 | | Vornicus [~vorn@ServicesOp.Nightstar.Net] has quit [Ping Timeout] |
| 01:46 | | Vornicus [~vorn@ServicesOp.Nightstar.Net] has joined #code |
| 01:46 | | mode/#code [+o Vornicus] by ChanServ |
| 02:34 | | gnolam [lenin@Nightstar-10613.8.5.253.static.se.wasadata.net] has quit [Quit: Z?] |
| 04:03 | | Chalcedon [~Chalcedon@Nightstar-2310.ue.woosh.co.nz] has quit [Ping Timeout] |
| 04:19 | | Chalcedon [~Chalcedon@Nightstar-2310.ue.woosh.co.nz] has joined #code |
| 04:19 | | mode/#code [+o Chalcedon] by ChanServ |
| 04:40 | | GeekSoldier|bed [~Rob@Nightstar-4537.pools.arcor-ip.net] has joined #code |
| 04:40 | | GeekSoldier|bed is now known as GeekSoldier |
| 04:41 | | GeekSoldier is now known as GeekSoldier|work |
| 05:47 | | ReivZzz [~reaverta@Admin.Nightstar.Net] has quit [Quit: reboot] |
| 06:23 | | Reiver [~Reiverta@Nightstar-3001.ubs-dsl.xnet.co.nz] has joined #Code |
| 06:24 | | Reiver is now known as NSGuest-1722 |
| 06:26 | | * McMartin makes libgmp earn its keep, verifying a useful property of a long-unsolved problem. |
| 06:27 | | NSGuest-1722 is now known as Reiver |
| 06:33 | | * McMartin will, of course, not solve it. BUt. |
| 06:34 | < Reiver> | ? |
| 06:34 | <@McMartin> | There is a famous problem called the 3n+1 problem. I just verified that the property it describes holds on all numbers from 1 through a million. |
| 06:35 | <@McMartin> | (Take a number, n. If n is odd, make it be 3n+1. If n is even, make it n / 2. If you keep doing this, will you always eventually reach 1?) |
| 06:35 | | * Reiver eyes that problem. |
| 06:36 | < Reiver> | So it's asking if you'll always eventually hit a number that is 2^n? |
| 06:36 | <@McMartin> | Yes. |
| 06:36 | <@McMartin> | http://en.wikipedia.org/wiki/Collatz_conjecture |
| 06:41 | < GeekSoldier|work> | ah, the good old collatz. |
| 06:41 | | * McMartin would like to produce a histogram of the length of the Collatz chain, and the maximum values reached. |
| 06:42 | < GeekSoldier|work> | I let it run on my server for a couple days. forgot to put a way to actually GET the result out of it though. |
| 06:44 | <@McMartin> | Heh. |
| 06:44 | <@McMartin> | *Main> map (length . wondrous) [1..1000000] |
| 06:44 | < Reiver> | McM: Won't that show a generally increasing-over-time value, though? |
| 06:44 | <@McMartin> | Nope! |
| 06:45 | < Reiver> | I mean, 8 has inherently more steps needed than 4, doesn't it? |
| 06:45 | <@McMartin> | Other values vary wildly |
| 06:45 | <@McMartin> | 20..30 is [8,8,16,16,11,24,11,112,19,19,19] |
| 06:45 | <@McMartin> | That 112 is not a typo. |
| 06:45 | < Reiver> | Uh... huh. |
| 06:46 | <@McMartin> | And, of course, 32 itself is 6. |
| 06:46 | <@McMartin> | In the 1 through a million range, most of the values seemed to stay in the 100-300 range |
| 06:49 | | * McMartin googles, finds someone who thinks he's proven the conjecture |
| 06:49 | | * McMartin finds someone else who thinks he's proven the unprovability of the conjecture |
| 06:51 | <@Vornicus> | *snrk* |
| 06:52 | <@McMartin> | The former has an attack on the unprovability proof, though. |
| 06:52 | <@Vornicus> | heh |
| 06:53 | <@McMartin> | Incidentally, Reiver, it's true that there is a minimum possible number of steps, that being log_2 (x) + 1, but even for 27 you have to get up to nearly 10^12 before you can guarantee that everything takes at least as many steps as 27 does. |
| 06:53 | < Reiver> | Aha. Right. |
| 06:56 | | * McMartin should really be working on his distributed trust and reputation system, but oh well. |
| 06:56 | < Reiver> | ...? |
| 06:58 | <@McMartin> | From a discussion elsewhere on how to do a global reputation system that's (a) actually global, (b) reasonably difficult to game, and (c) scalable to hundreds of thousands of participants. |
| 06:58 | <@McMartin> | (a) and (b) are contradictory, so this is in part a balance issue. |
| 06:58 | | * McMartin plans to make a solution, then check the literature to see what he can find on it and how his solution stacks up. |
| 07:06 | < Reiver> | ...hrm? |
| 07:07 | <@McMartin> | This is a pretty long-standing problem, and I haven't done any research on what's already been done |
| 07:08 | <@McMartin> | So I'm doing a first stab at it in the dark, and then doing the research to see how my ignorance stacks up against what's already been done |
| 07:08 | | * Vornicus wonders what your solution would be. |
| 07:09 | <@McMartin> | Assigned trust values to people you know, computed ones as a weighted average for one step out |
| 07:09 | <@McMartin> | After that it gets Really Heinous, so then it uses a most-trusted spanning tree to work out a maximum possible trust for everyone with any trust chain. |
| 07:10 | <@McMartin> | You then use their (non-trust-based) opinions as a weighted average to get snap reputation. |
| 07:10 | <@McMartin> | I suspect there's a way to solve the heinousness problem, but my stabs at it all diverged. |
| 07:11 | <@Vornicus> | I think I've seen one sorta like that - How does it handle trustmongers? |
| 07:11 | <@McMartin> | Trust is one-way, so unless you're connected to one, they don't affect the reps you see at all. |
| 07:11 | <@McMartin> | If you are, they're easy to notice and you can directly distrust them. |
| 07:12 | <@McMartin> | And if they're five hops away, they're only producing a reputation change of 2-3% max, I think. I need to implement some experiments to check. |
| 07:12 | <@McMartin> | It may be that the divergences I was seeing was due to trustmonger feedback loops. |
| 07:13 | <@McMartin> | (I'm assuming here a trustmonger is someone who trusts a bunch of colluders 100% and tries to get people to trust him, so as to sway available reps.) |
| 07:13 | <@Vornicus> | Yes. |
| 07:13 | <@Vornicus> | Or someone who gives two-way trust to a bunch of colluders, for cash. |
| 07:14 | <@McMartin> | The fact that the only place where trust is actually additive is at the "direct acquaintance" level in my scheme, you should only see the effects if you're a colluder yourself. |
| 07:14 | <@McMartin> | In which case you can die in a fire, etc. |
| 07:14 | <@McMartin> | Every hop past "multiple people I know know and trust this guy" is purely multiplicative, and all numbers are less than 1. |
| 07:15 | <@McMartin> | There's also "Anyone with no trust rating gets a trust rating of, say, 0.5 / entire population, now give me the global reputation of this guy" |
| 07:16 | <@McMartin> | Which will of course have colluders all over the place, but the ultimate confidence of that result -- assuming nobody actually *in* your effective trust network knows the target from Adam -- will be really low |
| 07:16 | <@McMartin> | And, of course, "no trust rating" does not mean "has a trust rating of zero". |
| 07:16 | <@McMartin> | If you know trustmongers, you can trust them zero and they'll be Worse Than Strangers. |
| 07:39 | | * Vornicus reads about the collatz conjecture, gets down to the picture of the collatz fractal. |
| 07:39 | <@Vornicus> | Shiny. |
| 07:40 | <@McMartin> | Actually, the other problem with trustmongers is that you can't just automatically shut them down |
| 07:40 | <@McMartin> | Because, well, you're going to expect in a global reputation system to see Trust And Rep Cliques. |
| 07:40 | <@McMartin> | aka "circles of friends". |
| 07:40 | <@Vornicus> | Yes. |
| 07:41 | | * Vornicus trusts most people in this channel. |
| 07:41 | | * McMartin doesn't, at least not in a way to grant transitivity. Sorry, guys. ;-) |
| 07:41 | <@Vornicus> | as, probably, do the rest of you. At least to some extent. |
| 07:42 | <@McMartin> | Yeah. Whether a value of, say, .3 counts as "trust" or not depends on your metric. |
| 07:42 | <@Vornicus> | Then there's also the Internet Reviewer Problem |
| 07:42 | <@Vornicus> | where everything is either The Best Ever or The Worst Ever. |
| 07:43 | <@McMartin> | All Kittens Get 10s |
| 07:44 | <@Vornicus> | I wouldn't trust a kitten with my credit card. |
| 07:44 | <@Vornicus> | I'd get it back all chewed. |
| 07:47 | <@McMartin> | The problem domain here was actually MMOs, and in the "Is this person a jerkass or not" vein. |
| 07:47 | <@McMartin> | PUG reputation. |
| 08:11 | < Reiver> | transitivity? |
| 08:14 | <@Vornicus> | Transitivity: a trusts b implies b trusts a |
| 08:15 | < Reiver> | oh /right/ |
| 08:15 | | * Reiver had been pondering that issue in something himself. |
| 08:15 | <@Vornicus> | Er |
| 08:15 | <@Vornicus> | Actually that's reflexivity. |
| 08:16 | <@Vornicus> | Transitivity is A # B && B # C -> A # C |
| 08:18 | <@Vornicus> | For a given operator that returns a truth value. |
| 08:21 | <@Vornicus> | (a strict weak ordering on a data type - the kind required for use in comparison-based algorithms like binary search, any sort other than radix, and treesets/treemaps, requires anticommutative < and >, reflexive =, transitive all three.) |
| 08:21 | <@Vornicus> | (a strict strong ordering requires that = always returns false) |
| 08:22 | <@Vornicus> | (unless the two things are exactly the same, completely) |
| 08:52 | <@McMartin> | That's commutativity. |
| 08:52 | <@McMartin> | Reflexivity is A trusts A. |
| 08:52 | <@Vornicus> | oh, yeah. daah |
| 08:55 | < Reiver> | ...,df z,rgneb-yz A ypgoy AZ |
| 08:55 | < Reiver> | uh |
| 08:55 | < Reiver> | ...why /wouldn't/ A trust A? |
| 09:19 | | You're now known as TheWatcher |
| 09:29 | | Vornicus is now known as Vornicus-Latens |
| 10:05 | | MyCatSchemes [~rb6822@Nightstar-18324.cs.bris.ac.uk] has joined #code |
| 10:46 | | MyCatSchemes [~rb6822@Nightstar-18324.cs.bris.ac.uk] has quit [Quit: Swim, swim, hungry!] |
| 11:02 | | Chalcedon [~Chalcedon@Nightstar-2310.ue.woosh.co.nz] has quit [Quit: Gone] |
| 11:16 | | ReivZzz [~reaverta@Admin.Nightstar.Net] has joined #Code |
| 11:29 | | Reiver [~Reiverta@Nightstar-3001.ubs-dsl.xnet.co.nz] has quit [Quit: Obelix is now running. Goodnight, Asterix!] |
| 12:05 | < ReivZzz> | Transitiveness works a bit differently in [love, as compared to maths]. "I love A, and A loves B, which means I want to make a light snack out of B's internal organs." |
| 13:01 | | gnolam [lenin@Nightstar-10613.8.5.253.static.se.wasadata.net] has joined #Code |
| 13:01 | | mode/#code [+o gnolam] by ChanServ |
| 13:33 | <@ToxicFrog> | ReivZzz: because they know that the gummint has been altering their brain? :P |
| 13:34 | <@ToxicFrog> | But the point here is that A # A is reflexivity, as opposed to commutativity (A # B == B # A) or transitivity (A # B ^ B # C -> A # C) |
| 13:35 | < ReivZzz> | Oh, TF |
| 13:35 | < ReivZzz> | Cygwin was throwing up errors. |
| 13:35 | < ReivZzz> | I found them... hilarious. |
| 13:35 | <@ToxicFrog> | Oh? |
| 13:36 | < ReivZzz> | "Error: Can't open D:\cygwin/log/setup.log for writing" |
| 13:36 | < ReivZzz> | Gee, I wonder why? |
| 13:36 | <@ToxicFrog> | ...why? |
| 13:36 | < ReivZzz> | Look at the slashes~ |
| 13:36 | <@ToxicFrog> | No, that's actually a perfect valid windows path. |
| 13:36 | < ReivZzz> | ...oh. |
| 13:37 | < ReivZzz> | Then I have no idea why it errored out trying to do that. |
| 13:37 | < ReivZzz> | Other than I told it to install on G: |
| 13:37 | <@ToxicFrog> | At least on NT-based windowses you may freely mix / and \, although use of \ should really be avoided. |
| 13:37 | < ReivZzz> | And it had already told me that I needed to exit all Cygwin applications to continue the installation... of cygwin. |
| 13:37 | <@ToxicFrog> | Where did you save the setup.exe and the packages? |
| 13:38 | < ReivZzz> | G:\Downloads\Cygwin\ |
| 13:38 | <@ToxicFrog> | And yeah, that's also normal. It can't install properly if something's holding the cygwin DLLs open |
| 13:38 | < ReivZzz> | Yeah, but I dunno what |
| 13:38 | <@ToxicFrog> | Not an issue if you don't already have it, but if you're installing over an old version it's seriously important |
| 13:39 | < ReivZzz> | This was throwing up errors on a clean install. |
| 13:39 | < ReivZzz> | I thus gave up. |
| 13:39 | <@ToxicFrog> | Weirdass. |
| 13:39 | < ReivZzz> | Very. |
| 13:39 | <@ToxicFrog> | What, exactly, did you do? |
| 13:39 | < ReivZzz> | Hence the giving up~ |
| 13:39 | < ReivZzz> | Uh |
| 13:40 | < ReivZzz> | Doubleclicked the installer, told it to grab the defaults, left it to it? |
| 13:40 | < ReivZzz> | (Changed the install path to the appropriate place too, but you know what I mean.) |
| 13:40 | < ReivZzz> | It got to 7% of installation, having successfully downloaded, before it went blooey. |
| 13:43 | <@ToxicFrog> | So...you didn't change the package storage path? |
| 13:43 | <@ToxicFrog> | Or the package selection? |
| 13:44 | < ReivZzz> | I changed the package storage path to G:, but that was about it. |
| 13:44 | < ReivZzz> | And didn't touch package selection; I couldn't remember what I was supposed to want. |
| 13:45 | <@ToxicFrog> | Everything. |
| 13:47 | <@ToxicFrog> | Anyways. Time for labs. Seeya. |
| 14:25 | | Attilla [~The.Attil@194.72.70.ns-11849] has quit [Quit: <Insert Humorous and/or serious exit message here>] |
| 14:43 | < Doctor_Nick> | gutsy is out |
| 14:55 | | Attilla [~The.Attil@194.72.70.ns-11849] has joined #code |
| 15:05 | | You're now known as TheWatcher[afk] |
| 15:15 | <@gnolam> | Cowardly is in. |
| 15:20 | < Doctor_Nick> | why, you! |
| 16:00 | | GeekSoldier|work is now known as GeekSoldier |
| 16:00 | | GeekSoldier is now known as GeekSoldier_ |
| 16:01 | | GeekSoldier_ is now known as GeekSoldier |
| 16:01 | | GeekSoldier is now known as identify |
| 16:02 | | identify is now known as GeekSoldier |
| 16:50 | | MyCatSchemes [~rb6822@Nightstar-23320.cs.bris.ac.uk] has joined #code |
| 17:12 | | Raif [~corvusign@Admin.Nightstar.Net] has quit [Ping Timeout] |
| 18:14 | | AnnoDomini [AnnoDomini@Nightstar-29508.neoplus.adsl.tpnet.pl] has quit [Ping Timeout] |
| 18:20 | | AnnoDomini [AnnoDomini@Nightstar-7082.neoplus.adsl.tpnet.pl] has joined #Code |
| 18:20 | | mode/#code [+o AnnoDomini] by ChanServ |
| 18:40 | | Syloq [Syloq@NetAdmin.Nightstar.Net] has joined #code |
| 18:41 | | Syloq is now known as Syloqs-AFH |
| 19:35 | | GeekSoldier is now known as GeekSoldier|bed |
| 20:32 | | Chalcedon [~Chalcedon@Nightstar-2310.ue.woosh.co.nz] has joined #code |
| 20:32 | | mode/#code [+o Chalcedon] by ChanServ |
| 20:39 | | Chalcy [~Chalcedon@Nightstar-2310.ue.woosh.co.nz] has joined #code |
| 20:39 | | mode/#code [+o Chalcy] by ChanServ |
| 20:39 | | Chalcedon [~Chalcedon@Nightstar-2310.ue.woosh.co.nz] has quit [Ping Timeout] |
| 20:39 | | Chalcy is now known as Chalcedon |
| 20:39 | | MyCatSchemes [~rb6822@Nightstar-23320.cs.bris.ac.uk] has quit [Client exited] |
| 21:56 | | gnolam [lenin@Nightstar-10613.8.5.253.static.se.wasadata.net] has quit [Quit: Blargh.] |
| 22:01 | | You're now known as TheWatcher |
| 22:01 | | GeekSoldier|bed [~Rob@Nightstar-4537.pools.arcor-ip.net] has quit [Ping Timeout] |
| 23:27 | | You're now known as TheWatcher[T-2] |
| 23:30 | | Forj [~Forj@Nightstar-2310.ue.woosh.co.nz] has joined #code |
| 23:30 | | mode/#code [+o Forj] by ChanServ |
| 23:33 | | Pi [~sysop@Nightstar-24414.hsd1.wa.comcast.net] has quit [Ping Timeout] |
| 23:35 | | You're now known as TheWatcher[zZzZ] |
| 23:45 | | Vornicus-Latens is now known as Vornicus |
| --- Log closed Fri Oct 19 00:00:19 2007 |