[09:18] Join: Core_old joined #corewars [09:58] Join: Core_old_ joined #corewars [10:01] MSG: Ping timeout: 255 seconds [10:42] MSG: Ping timeout: 255 seconds [11:20] Join: Core_old joined #corewars [11:59] Join: Fluffy joined #corewars [11:59] :) [11:59] ~ [12:00] x [12:01] I guess that nobody will want to read that in the logs [12:02] Well then, be thou welcometh in this fine channel of conversation [12:02] much better! [12:12] Part: Core_old left #corewars [12:14] MSG: Quit: Leaving. [13:34] nacht! [13:44] Join: Fluffy joined #corewars [13:44] :) [16:33] Join: Core_old joined #corewars [16:44] MSG: Ping timeout: 255 seconds [16:54] Join: fiveop joined #corewars [16:56] Join: John joined #corewars [16:57] Hi :-) [16:57] Hi John [16:57] Hi Jens [16:57] Hi John! Neverland is your warrior? [16:57] Yes [16:58] It's published [16:58] hm... hard to beat *g* [16:58] :-) [16:58] It is okay to send to the beginner hill [16:58] I wrote it when I was a beginner ;-) [16:58] aha... hm... hmmmm..... [17:00] my new scanner scored much better than little factory with wilmoo benchmark, but got a lower score on the hill [17:00] than L'F [17:00] :-( [17:01] If it scores better against the benchmark, I would think it's better overall. [17:01] Just an unlucky set of warriors on the hill for you [17:03] but i am getting better :-)) [17:03] ares: http://www.ociw.edu/~birk/COREWAR/DRAFT/HILL/neverland.red [17:03] hmm... should i open that... wouldnt it be like cheating? *G* [17:04] (klick) [17:04] ares: huh? [17:04] neverland is small - wow [17:05] It's the smallest I could make it without too many compromises [17:06] ... and easy to understand [17:10] aaah! [17:10] Hey, what happend to the nano-hill? [17:10] I really should pay more attention [17:11] gnik took over! [17:11] ares: http://www.ociw.edu/~birk/COREWAR/TINY/HILL/lonegunman.red [17:11] I really hope that the "new" benchmark will work [17:12] ares: that oneshot is 6 lines and fairly easy to understand. It also does well on the beginner hill. [17:12] just reading the source [17:14] There version sent to the beginner hill didn't include the decoy maker [17:14] There EQU The [17:37] MSG: Quit: Harald *= 0; [17:37] Join: ares joined #corewars [17:46] Join: Nenad joined #corewars [17:46] :) [17:46] Hi Nenad :) [17:46] Hi Nenad :-) [17:46] anyone besides me problems with icq? [17:47] Don't use it [17:47] one of my 2 accounts just died [17:47] I can login just fine [17:48] my 2nd works, too [17:58] * Fluffy waves [17:58] Part: Fluffy left #corewars [17:58] * John waves too late [17:58] *g* [17:59] * Nenad waves even later [17:59] ares: I see that you've discovered Neverland :) [17:59] how can you see that? [17:59] it used to bother me as well, when I was a beginner [17:59] ? [18:00] ah [18:00] in the logs, of course :) [18:00] yeah... *g* i allready started thinking about a captain hook warrior [18:00] John has a way of making nasty scanners. [18:01] Neverland isn't nasty :-) [18:01] LOL [18:01] well, it is for a beginner warrior [18:01] * ares still loughing [18:01] It was a couple of months before I wrote Zooom and became a pro! [18:02] now THAT was a coool warrior [18:02] but somehow I think that cds without zooom trick tend to score better lately [18:03] at least that's the impression that I have gotten after experimenting with scanners... [18:05] hey - who killed the lone gunman? [18:05] ares: I suggest you start with papers, then play with scanners for a while, and then do some stone/papers and stone/imps... and then -> whatever you like most :) [18:05] ? [18:05] about one hr ago john sent lone gunman to beginner's [18:06] seems to be removed again [18:06] Yes, I don't want to flood it [18:06] i dont like papers very much... especially without limits [18:07] I like paper ;-) [18:07] hm - . [18:07] Makes yummy scanner food :-) [18:07] lol [18:08] heh. there are some that make you break a few teeth, though [18:08] pwi are NOT an easy food [18:09] btw, why did you go for Mischief in round1, there was almost no danger of encountering imps [18:09] and scn are more vulnerable to stn than cds [18:10] I expected to see hsaish scanners only in pspacers [18:10] Because I like Mischief ;-) [18:10] lol [18:11] I have to say that my favorites from HSAish group are Razor and Qutrum [18:12] just admit it -> in HSAish scanner you didn't have to change scanning steps, so it was less work :)) [18:12] I'm not bothered about a bit of work changing steps [18:13] I know that I would be... good step changes scores dramatically [18:13] and it takes time to find one + ini value + cnt + ... [18:14] I didn't want to publish a slight variantion of a warrior I've already published [18:14] a lot of work even for an optimizer, and it's ususally even faster to find the values by hand, however paradoxical that sounds [18:15] for all the tourney rounds out of the standard coresizes, it is really difficult to optimize, since you'd need first to create a banchmark big enough and diverse enough [18:15] :-) [18:16] Then I'm not at a disadvantage because I don't optimize [18:16] probably :) [18:17] so, since almost have the time has passed, I guess you've already made a warrior for one of the cores? :) [18:17] *half [18:17] Hmmm... no [18:18] :) well, don't worry... there will be more time given for the next two rounds, anyway [18:19] probably 3 weeks for round3 and 4 weeks for round4, but that's not a final decision yet. [18:47] Join: Fiz joined #corewars [18:47] hi [18:47] anybody here? [18:47] am on web client [18:47] :))))) [18:48] hi [18:48] don't know if it works [18:48] long time no see [18:48] hi Nenad [18:48] yes, am super busy at work [18:48] still on business trip [18:48] I send Corewar greetings from Kuwait now [18:48] wow [18:49] :-S [18:49] how did you end up there? [18:49] company send me [18:49] well, I hope you find some cw time soon :) [18:49] I will return to Germany on Thursday [18:50] I was sad you hadn't sent and entry for the first round of the tourney [18:50] yes, I hope so. I guess I missed the first round :-( [18:50] Hi Fizmo [18:50] hi John [18:51] how r u? [18:51] Fine thanks, are you? [18:52] I've been busy tweaking my webpage [18:52] I am ok [18:52] cool [18:52] I should also update my page, but somehow I didn't find time [18:53] Don't worry, there's not much to update [18:53] there is never much to update [18:53] too quiet overall [18:53] :-/ [18:53] :-( [18:53] but I hope this tourney will at least add some dynamics [18:54] yes, yes [18:54] so sad that I missed the first round [18:54] you still have time to catch up [18:54] 7 more to go [18:55] :) [18:55] :-P [18:57] first I must be back home in one piece :-P [18:57] anyway, Kuwait is more relaxed than I expected [18:58] :) [18:58] lot of different nationalities living here [18:58] you'll have to post some pictures [18:58] ohh, yes, that's a good idea :-) [19:01] not much traffic on 94nop :-( [19:02] No, lots of traffic on the nano hill though [19:02] I think the next warrior in the all time HoF will be a nano warrior! [19:02] ohh, really? I haven't checked [19:02] cool :-P [19:02] but maybe, because we set the hill too large [19:03] could be [19:03] It's not too large [19:03] it would be interesting to see 40 slot 94nop, though :) [19:03] Though obviously the average age is higher [19:03] SoV might still be there then [19:04] We've had a warrior age 500+ on the multi-warrior hill, which is only 10 slots [19:04] hahaha, 40 would be a bit too large I would think [19:05] but maybe 25 [19:05] yes, because almost noone makes an effort to make a complex pspacer for multiwarrior [19:05] and it was, as I remember [19:05] anyway it would be worth to think about the consequences if the hill size is increased [19:05] P-space works differently on multi-warrior [19:05] yes, I know [19:05] is pspace cleared (except cell #0) before a round starts, or is it kept? [19:05] it stores the number of surviving wariror [19:05] maybe start with the 94-draft hill first [19:06] there aren't much p-spacer [19:06] Fiz: almost no traffic there, though [19:06] ares: kept between rounds, cleared between warriors [19:06] 94t was my hill of choice for pspace -> enough space to combine anything and everything [19:06] thx [19:06] it's really funny. In the time of the Pizza hill it was the Hill #1 [19:07] the 94 draft [19:07] it should still be... but it seems that either we no longer make good pspacers or the optimized 94nop sbi, sai, pws have an advantage due to good qscans ... [19:08] most pspacers have trouble scoring well against strong pws [19:08] I think we don't make good p-spacer in the moment [19:08] probably :) [19:08] except Bulldozed :-P [19:08] modesty, of course ;) [19:09] hahahaa [19:09] :) [19:09] * John wonders what is in inversed's new qscan [19:09] yes, that is the question [19:10] I think if pspacer are also a bit optimized as 94nop warriors are, it shouldn't be a problem to enter the 94draft [19:10] optimizing the switch table? [19:10] otherwise we have almost a second 94nop [19:10] optimizing switching tables would be very nice [19:10] * Nenad wants to see more Combatra-type warriors at draft hill [19:10] but I haven't a tool [19:10] ? [19:10] optimax? [19:11] yes, but honestly I never tried to optimx switch tables so far :-P [19:12] there - you see that ssmd clear can enter 94nop :) [19:12] to optimize a p-spacer is a much more tricky task than a 94nop warrior [19:12] I realize that [19:12] :-P [19:13] optimizing pspacers won't be easier even next summer [19:13] Optimizing a p-switcher table is much quicker than optimizing the components [19:13] You don't need to run any battles [19:13] ? [19:14] how come? [19:14] hahaha, you got me with the Utterer TNG :-P [19:14] :) [19:14] nice score [19:14] Run each of your components against your benchmark warriors [19:14] well, it's the coreclear -> antiAfield specialist warrior [19:15] John: so? [19:15] Then you can run the p-switcher optimization based on the %wlt [19:15] if you're speaking of taking averages... I don't think it's realistic enough [19:16] I believe you do need to run battles indeed [19:16] and it's more difficult to find the right values [19:16] because of the nature of the search space [19:16] No, just a program which simulates the switcher [19:16] iteratively? [19:17] but what about brainwashing? and how do you decide on the next strategy based on an average score? [19:17] * Nenad is confused [19:17] I think it should be possible to calculate [19:18] approximately - yes [19:18] Okay from the beginning: [19:18] you can go through just nummerical instead of running cw [19:18] but what would be the variance, can yu tell me that? [19:18] Run each component of your p-switcher against each opponent [19:18] The put the %wlt into a table in a small program [19:19] you know how little the differences are in the end... the top 10 optimax table usually looks like 131. [19:19] 131.4 131.2 130.8 atc [19:19] *etc [19:19] and if this approximation you get isn't *very* accurate, [19:19] The program will generate switcher tables in whatever way - random, evolved, sequentially [19:19] you can easily end up with a worse version [19:20] Then run 1000 or whatever battles (x 250 rounds) with that switcher (simulating the switchers choices) [19:20] Whether it wins / loses / ties is base on the percentages in the table [19:21] hmm... well, in the end, there is only one way to really tell if it works... -> to test it in practise [19:21] I think it is worth a try [19:21] So it is very quick [19:21] oh, probabilistic? I missed the point... [19:21] but still -> what about BRAINWASHING [19:21] how do you simulate that? [19:21] that's part of the code [19:21] Brainwashing is simulated by a small percentage chance of setting the switcher to a random state. [19:22] hmmm... [19:22] well, I admit that it *might* work [19:22] :)) [19:22] And if 1000 battles isn't enough for you, make it 100000. [19:22] no, 1000 is enough [19:22] It is amazingly fast, because it doesn't need to run the battles. [19:23] (in a mars) [19:23] It does work. ;-) [19:23] but you'd have to run all the components with the rest of the pspacers code sitting idly aside -> not 94nop files... so that the chances of being found by a qscan are not affected [19:23] Yes [19:24] And here's the interesting bit [19:24] You can work with 50 different components [19:24] And include all the %wlt scores in the simulator [19:24] and then switch the components? [19:24] And it will eventually choose the components and switcher neccessary [19:25] but that contradicts what was said a few lines above [19:25] the %wlt need to be calculated with all the code included [19:25] so you'd have to recalculate for every subset [19:25] because components are, if nothing else, of different size [19:26] It's close enough with just the component and the rest of maxlength filled with junk [19:26] is it? what if you have a 2 component switcher? [19:26] one component can be either 10 or 30 lines long [19:26] My p-space is not affected by loses to qscans anyway ;-) [19:26] :) [19:27] heheheh, I hope I see a discussion of two authors who enters soon the 94 draft with some p-spacers :-P [19:27] Well, less affected ;-) http://www.ociw.edu/~birk/COREWAR/PSPACE/HILL/enough2.red [19:27] :) [19:27] well, I never optimized a ptable [19:28] just decided on the values following a hunch [19:28] Actually it has been done in the past exactly as I describe [19:28] ohh, really? [19:28] who have done that? [19:29] Two authors [19:29] John Metcalf and Dave Hillis ;-) [19:29] funny... never heard about them [19:29] you, Fiz? [19:29] :) [19:30] * Nenad is sorry to say that he has to go :( [19:30] :-( [19:30] * John waves [19:30] * Nenad waves [19:30] * Fiz waves [19:30] MSG: [19:30] Here's an example http://www.ociw.edu/~birk/COREWAR/PSPACE/HILL/mantraparcade.red [19:30] yes, I remember John [19:31] Hillis evolved the table, right? [19:31] Yes [19:31] Also http://www.ociw.edu/~birk/COREWAR/X/HILL/katv5.red [19:33] It did tend to sometimes generate tables which had dead-end starts. [19:33] starts EQU states [19:33] Once it switched to one, it never switched out. [19:34] In the end I added a condition to the program to prevent it [19:40] MSG: Quit: Lost terminal [19:40] Join: Roy joined #corewars [19:42] Hi Roy! [19:43] * John shakes Roy [19:44] hi! [19:46] hi [19:46] Part: Fiz left #corewars [19:47] Join: Fiz2 joined #corewars [19:48] hi [19:48] Hi again ;) [19:49] Sigh.. [19:49] hi, something seems wrong with the web irc client [19:49] I had a nice LED flashlight... but a button fell of, so I decided to ´fix' it [19:49] :-P [19:50] Now I put it back together and its always on, but when I press the button it goes off :P [19:50] Optimistic shining... [19:50] hahahaha [19:50] cool :-P [19:51] Now I have to tear it apart one more time :( [19:51] anyway here it is almost 11 pm. I think I will go to bed now [19:51] wish you nice time [19:51] ciao! [19:51] * Fiz2 waves [19:51] * John waves [19:51] bye bye [19:51] Part: Fiz2 left #corewars [20:14] * John waves [20:14] MSG: Quit: mov.i #1,1 [20:30] Nobody else around? [20:30] * Roy is bored [20:30] (no not Borat...fools) [20:31] * ares waves [20:31] writing manuals is boring, too [20:31] Then don't do it for a while :P [20:31] you know, why programmers allways mix up halloween and christmas? [20:32] No..? [20:32] OCT 31 = DEC 25 [20:32] lol [20:35] coffee.sys not found - sysop halted. [20:37] * Roy really can't wait till tomorrow [20:37] and why is that? [20:37] Im going to program something for my project with a ex vice-president of Apache [20:38] oh [20:38] what is it? [20:39] A connection between BEA Weblogic and Maven, not that intresting, but the guy is a real wizz, he was also invited for a talk with google, but in the end he wasn't smart enough :) [20:39] smart -> he didn't know enough algorithms, only a lot of experience [20:40] thats bad... sounds like me [20:40] For some reason google expects all its programmers to be very good mathemetichans [20:41] they have need for mathematicians... google is fast... [20:42] i heard about google in a scientific radio documentation... crazy maniacs at work there [20:42] Heh indeed, and lot of crazy stories about the googleplex too, a lot of swimingpools there! [20:42] on their hompeage they say, that they use some thousand low cost pc's [20:43] I bet thats a lot cheaper then a couple of high ass high-end machines, and you have much more seperation [20:43] do they have waterproof notebooks? sub(marine)notebook... [20:44] i'd love to work in a swimming pool *gg* [20:44] There are lots of things I would like to do :P [20:45] But it just doesn't pay very much [20:45] Bed-tester... chocolate-taster... gigolo :P [20:45] *g* [20:46] good ideas... we only need a financial concept [20:47] Hmm yes.. any ideas? [20:47] i am not a manager... i'd need one [20:48] Me neither, I have one for my current project, but thats a real manager... ignorant etc :P [20:48] We need some sales guy! [20:48] And a good R&D! [20:48] i tell you, if i finde some [21:30] Join: Roy_ joined #corewars [21:34] MSG: Ping timeout: 255 seconds [21:40] Join: Roy__ joined #corewars [21:43] MSG: Ping timeout: 255 seconds [21:58] MSG: Quit: Ik ga weg [23:22] MSG: Ping timeout: 255 seconds