[00:13] MSG: Ping timeout: 240 seconds [00:19] MSG: Quit: blaaaarghhh [04:23] Join: Mannerisk joined #corewars [07:16] Join: Metcalf joined #corewars [08:56] Join: sh0ne joined #corewars [08:56] Hello [09:02] MSG: Remote host closed the connection [10:17] Join: sh0ne joined #corewars [10:18] Hi Sh0ne [10:19] Hello Metcalf [10:40] MSG: Ping timeout: 240 seconds [10:56] Join: ipeev joined #corewars [10:56] хело [11:00] Hi ipeev [11:00] MSG: Client Quit [11:01] * Metcalf frightened off the newbie :-( [11:04] Join: sh0ne joined #corewars [11:30] Join: CoreOld joined #corewars [11:44] MSG: Ping timeout: 240 seconds [12:10] Join: sh0ne joined #corewars [13:04] MSG: Quit: Swallowed by Supermassive Black Hole [13:22] Corewar is on the front page of programming reddit http://www.reddit.com/r/programming/ [13:22] twice! [13:23] MSG: Quit: Trillian (http://www.ceruleanstudios.com [15:26] MSG: Ping timeout: 240 seconds [15:33] Join: Davse_Bam joined #corewars [16:01] Join: Fizmo joined #corewars [16:07] Join: Metcalf joined #corewars [16:07] Hi :-) [16:07] hi hi [16:07] ;-) [16:08] I have an idea [16:08] Hi Christian [16:08] I wondered if you'd be here [16:08] What kind of idea? [16:08] a kind of crazy idea :-P [16:09] actually we have a kind of "static" hills [16:09] they perform just in case somebody send something [16:09] They're the best kind [16:10] but I think it would be also interesting to have a dynamic hill [16:10] which performs on fixed times, no matter if new programms are sent or not [16:11] lets say once per week or so [16:11] the problem would be that it would tend to forms clusters [16:11] Do you mean like at automatic random submission? [16:11] no, I think about a sumo tournament style system [16:12] just battle your close neighbours [16:12] I still don't get it [16:12] Ah, I see :-) [16:12] if you win more than you lose you will climb up the hill and facing another "better?" warriors [16:12] Yes, that might be interesting [16:12] but to make it even more challenging [16:13] what about the following [16:13] having such kind of "sumo-style" hill and tournament scheduling [16:14] that for every fight between two warriors, some parameter may change [16:14] depending on if your fighter takes the white or black stone [16:14] for example the Max. Process could be either 8, 64, 800 or 8000 [16:15] randomly [16:15] Hmmm... that might take away some of the science of corewar [16:15] one could also think about a delay [16:15] that one of the two warriors start 1-10 cycles later [16:16] the science of corewar is figured out already [16:16] Join: fiveop joined #corewars [16:16] the delay could be stored in p-space, so that the warrior knows if it started first or slightly too late [16:16] Hi Fiveop [16:17] well, it would bring some dynamic into the game [16:17] Starting late is a good idea [16:18] overall it would mean that a warrior not necessarily wins over another warrior, but that there is also a random factor [16:19] and maybe draft warriors and even LP-like warriors could co-exist [16:19] just a bunch of ideas actually [16:19] A random factor sounds good [16:19] bad [16:20] and the player could watch day by day or week by week how their warrior climb up the hill [16:21] I like the sound of starting late. I don't like the sound of max processes or coresize changing [16:21] with the random factors you brevent such a dynamic hill to fall into a steady state [16:22] prevent [16:22] Starting late could be used as a handicap, letting newbies play on the same hill as pros [16:22] no coresize changing, I also don't like [16:22] I still like the idea of a diversity hill [16:23] well, it could be also a kind of "aging" tool [16:23] every 100 age it starts one process more late [16:23] additionally to the random factor [16:23] just an idea [16:25] and the hill size would have just a minor effect, as a warrior would just battle the opponents ranking near to it [16:25] him [16:26] maybe a power ow 2 that the fight schedules could be simple calculated, such 64, 128 or 256 [16:27] or even larger [16:29] well, with an "aging" a new beginner warrior could even win against an old pro warrior [16:29] Join: OoS joined #corewars [16:31] 259 downloads of nanoWarrior so far! :-) [16:32] MSG: Ping timeout: 240 seconds [16:32] nice -:) [16:32] the newNano Warrior is really good!!!! [16:33] Thanks, sf and Neo will be pleased to hear that too [16:34] The paper/clear looks really interesting [16:34] could it be also usefull for Tiny or larger hills? [16:34] You should try the nano hill again :-) There's benchmarks and stuff at http://corewar.co.uk/nano.htm [16:35] hahaha, yes, yes [16:35] based on that I can prepare a nice testsuit for optimax :-P [16:35] (maybe, now that i know C, i'd try to do the modifications i wanted on YAce-nano) [16:45] John? [16:46] what you think about the hill idea? [16:46] I like the idea of only battling neighbours, and once per week [16:47] I like the idea of delayed startup as a handycap [16:47] But I'm not keep on the idea of random settings [16:47] yes, maybe too much at once [16:47] Do you remember how Olympus used to work? [16:48] a new warrior started to battle against the bottom warriors [16:49] 1. Run the newcomer against 50 programs taken at random (with an even spread) from the current rankings. [16:49] 2. Re-do the rankings, taking these 50 battles into account (in addition to all previous battles). [16:49] 3. Compute a set of battles: the top 25 against every other, and each warrior against the 10 above and the 10 below it. [16:49] 4. Run the battles of this set that haven't been run already. [16:49] 5. Re-do the rankings. [16:49] 6. Repeat steps 3-5 until there are no new battles in the set. [16:51] the advantage of Mt Olymp is, that you can see the scores against other warriors [16:51] at least it was available log time ago :-( [16:52] long [17:17] fizzer! [17:20] Hi Bvowk [17:20] greets Met [17:21] how goes? [17:21] hi bvowk [17:21] about my dynamic hill idea [17:22] my idea is that it is divided into blocks of 8 warriors which fight each other [17:23] the 1st climbs up two block, the second one bloch, 3-6 keep in the block, 7 drops one block and last drops 2 blocks [17:24] that would be just the most simplest solution [17:25] interesting. [17:25] got any code? [17:25] more complex is the way how it is done in sumo, but it is also more complex how to calculate the new ranking based on the last results [17:26] ohh, I just explained my ideas about a dynamic hill [17:26] why don't you write it up and I'll play with it in my evolver [17:27] wow, you would do that? [17:27] I've found the diversity hill works really well. [17:27] :-) [17:27] cool [17:27] so your idea sounds interesting :) [17:28] my idea is a big dynamic hill [17:28] but without having to fight all the battles [17:28] every warrior just fight it's direct neighbours in the ranking [17:28] if scores good it will climb up the hill [17:29] if it scores bad it will drop down [17:30] the way how the ranking is handled in the professional sumo tournaments is very interesting, as it keep the ranking quite dynamic [17:30] and no steady state could be formed (at least I hope) [17:31] the down side is, that it seems to me very tricky to calculate new rankings [17:31] bvowk? [17:32] yes? [17:32] how deep you're inside the emulators, such as pmars or exhaust? [17:33] I'm thinking of a kind of delay before the second warrior is starting [17:33] (hmm.. damnit, ive deleted that one document i had, this will take some time) [17:33] hrm. [17:33] you want to add a delay? [17:33] Metcalf> Starting late could be used as a handicap, letting newbies play on the same hill as pros [17:33] I still like the idea of a diversity hill [17:33] well, it could be also a kind of "aging" tool [17:33] every 100 age it starts one process more late [17:33] additionally to the random factor [17:34] well, with an "aging" a new beginner warrior could even win against an old pro warrior [17:35] at least it would give an additional random factor [17:35] ;-) [17:36] actually it's just an idea [17:38] 7/quit [17:39] MSG: Quit: humhum [17:39] hmmm [17:39] John? [17:41] maybe not a randomization of Max. Processes, but assigning in the warrior the prefered Max. Processes [17:42] i still think its not proper to screw up with the core-settings on an diversity-hill [17:42] unless you were talking about sumo-hill [17:42] where warriors with same Max. Processes fight at this conditions, while with different prefered Max. Processes fight random [17:42] It might be a good idea to run some tests and write an article about it like LAchi did for the diversity hill [17:43] yes, tests should be done [17:44] does anybody have markov-analysis of nano? [17:44] just wondering [17:44] We need to encourage players to submit to the hills we've got first ;-) [17:44] or even a flat table of opcode regularity [17:46] yes, you're right John [17:47] but most of the hills are really on the top noch, so it is pretty difficult to enter without a lot of efforts [17:47] Bvowk is the most likely [17:47] (nevermind, just table'ed samples of 50 nano warriors.. maybe Metcalf should NOT run it on infinity ;) [17:48] Should be too hard for a flat table [17:48] Maybe if we had a look at the hills, and published some ideas in r.g.cw [17:48] E.g. mention that a scanner would do well on '94nop at the moment [17:49] 250 lines, 129 mov's, 51 spl's [17:49] And mention that a good paper/clear would go straight to the top of nano [17:51] Mizcu: I suspected that would be the case [17:52] Oos: i finally got Will's build_freq_tables to work [17:52] :-) [17:58] I think the most under-rated hill is SAL's Tourney Hill [17:59] with a max. length of 300 it opens a lot of possibilities [18:00] What about the fortress hill? [18:00] Maybe 300 gives too many possibilities [18:00] yeah, the fortress hill is really cool [18:00] I should re-review it [18:01] When players haven't got much time to spend on corewar, 300 lines is too long [18:01] Neo's smart p-switiching concept would be interesting to try [18:02] you're right [18:02] I think I should focus on the Fortress Hill [18:03] BTW, I can' [18:03] t see you on the fortress hill, John :-P [18:07] tiny LP would be also worth to try [18:07] No, I tried it a few times and didn't have much luck :-( [18:09] I have massive numbers of nano warriors [18:19] maybe evolve for tiny LP? [18:21] 6 lines expanded to 25.. [18:36] our algorithm teacher would hang me if hed see how im coding.. [18:37] MSG: Ping timeout: 240 seconds [18:46] Join: OoS joined #corewars [19:07] yay, i screwed up something [19:09] MSG: Ping timeout: 240 seconds [19:14] typical, compiles fine, but windows throws an error [19:14] hrm. [19:15] I've got a massive collection of tiny warriors too [19:21] Join: elkauka joined #corewars [19:26] hi el kauka [19:31] hi fizzer [19:31] :-) [19:31] how r u? [19:32] fine, thx ;) [19:34] Join: OoS joined #corewars [19:36] I still running some tests about my ideas [19:41] MSG: Ping timeout: 240 seconds [19:42] 94draft vs. LP [19:42] on which max. process they are most equal [19:59] it's somewhere between 50 and 100 [20:03] i just suddenly started hating C [20:05] :-P [20:08] for some reason, 91 <= 2 [20:44] MSG: Quit: ChatZilla 0.9.84 [Firefox 3.0.4/2008102920] [21:09] i just accidentally math [21:54] MSG: Quit: Read error: Connection reset by peer [22:08] Join: Metcalf joined #corewars [22:09] Hi :-) [23:45] MSG: Quit: blaaaarghhh