[04:34] Join: Mizcu joined #corewars [06:06] MSG: Ping timeout: 255 seconds [07:36] Join: fiveop joined #corewars [10:00] MSG: Read error: Operation timed out [10:02] Join: asw joined #corewars [11:08] MSG: Read error: Operation timed out [11:11] Join: asw joined #corewars [16:17] hrm. [16:20] hrm [16:22] whats up miz? [16:23] hacking on anything? [16:23] yes, my head on the table [16:23] heh [16:23] too much sleep today [16:26] bah. [16:46] Join: John joined #corewars [16:46] Hi [16:46] hi John [16:46] heya! [16:47] Hi Mizcu, Bvowk [16:47] Did you see my new website: http://patchworkpixels.co.uk ? [16:47] yup [16:48] Sorry Mizcu, I don't think the name you suggested would have been effective. (having "ripoff" in the URL might put people off ;-) [17:06] * John is curious about eNeRGy [17:24] nope [17:25] Join: Core_old joined #corewars [17:34] Hi Core_old [17:36] hi! [18:07] Join: Nenad joined #corewars [18:07] hi [18:07] Hi Nenad [18:14] Join: Miczu joined #corewars [18:14] wtf, i say [18:15] Hi Miczu [18:15] Do you know Mizcu by any chance? [18:15] quite yes [18:16] what an nice young man he is [18:16] MSG: Ping timeout: 255 seconds [18:16] Nick Change: Miczu changed nick to Mizcu [18:16] shh.. i am tricking him ;) [18:17] well, you look awfully alike [18:17] you're sure you're not relatives or something? [18:18] quite not, my friend [18:20] but seriously [18:20] been watching my little bunch of code lately? or should i just dump it to John ?;) [18:20] ? [18:21] to Nenad, the line was supposed [18:21] :-) [18:22] just had a little interesting idea i couldnt develop any further and i gave it to Nenad - who just ran out of time the time i gave it to him [18:22] yes, well I was away for a few weeks [18:23] and without a pc as well [18:23] but haven't forgotten :) [18:23] will look at it now (equ within a day or two) [18:24] well, i am not in a hurry anyway [18:24] Topic Change: Mizcu sets topic: http://corewar.co.uk - http://www.corewar.info - http://www.koth.org - how about IRCT at sunday if we got the people? [18:25] today is? thursday? [18:25] y [18:25] sure :) I'm in [18:25] If I'm here, I'll be in too [18:26] at what time should it be? [18:26] 20h CET? [18:26] 19 GMT [18:26] is that the same or not? [18:27] CET = GMT+1 [18:27] Topic Change: Mizcu sets topic: http://corewar.co.uk - http://www.corewar.info - http://www.koth.org - IRCT at sunday if we got the people, 1900 GMT [18:36] Time to go [18:36] * John waves [18:36] by-by [18:37] MSG: Quit: mov.i #1,1 [18:38] * Nenad waves [18:38] have to go as well [18:38] multimode [18:38] MSG: [18:38] bye to you too, Nenad [19:58] Mizcu, something on your mind? [20:05] http://koti.mbnet.fi/mizcu/optimax.png [20:05] thats from an old, failed project [20:06] optimizing can make some really unexpectable scoring in some cases [20:06] though actually i know exactly where that weird scorer comes from [20:09] the most common "unintentional" tactical change optimizing is from anti-imp papers [20:10] usually papers have 8 processes; with the anti-imp papers, the optimizer finds stepsizes where copies overwrite eachother near end of battle so that some papers get more processes [20:11] that increases the size of bomb-carpets and extra dat's, thus making it more aggressive and better against imps [20:13] and as an interesting note, Hurkyl once met an interesting thing while optimizing he's stone/imp. One of the created warriors was suprisingly resistant to some type of anti-imp papers [20:14] but with little change in the numbers it lots the resistance [20:14] It was because one of the stone's bombs hit the spiral somewhat early in the battle, thus splitting one spiral into two [20:15] and as specialized anti-spiral bombs killed one spiral, the second part survived [20:17] so there are things beyond numbers in optimizing [20:18] (if Inversed is reading logs, please dont see me as an optimax-maniac even with what i just said) [20:28] it is better to optimize such anti-imp papers so that instead of trying to find such overwriting from random numbers, is to write the paper so that it overwrites always regardless of the step you are optimizing [20:34] but you have to know that the overwriting is the mechanism behind the anti-imp power before you can start optimizing with that in mind [20:47] what, no comments? [20:51] MSG: Quit: Trillian (http://www.ceruleanstudios.com [21:06] tries to go sleep --> [22:49] Sorry, I had my attention taken elsewhere. [22:51] I found some papers on the mathematical explanation for the "special constants". They mostly confirmed my suspicion. There's nothing whatsoever special about the number. The number is only special given a certain algorithm (e.g. if you divide core into smaller and smaller units, evenly, versus some other bombing tactic). [22:52] So really it's a discussion of the optimal implementation of a particular algorithm, which is a very reasonable thing to talk about, I just think it's peculiar to be talking about it in terms of changing only constants. [22:56] I don't know what inversed thinks, but personally I think if optimizing works (and I'm sure it does lead to improvements at least in some cases), it's a reasonable thing to use. I don't understand the supposed "manly" programming tactic of doing things in the least intelligent, and/or most difficult way possible. If the warrior works better when run through optimax, I'm all for it. [22:57] I just don't trust things I don't understand real well. I was trying to wrap my head around why certain numbers would be "special". The mystery goes away when it's worded "some algorithms work better than others, and algorithms occasionally involve constants". The only thing that /aught/ to be "special" would be statistical frequency of some feature. [22:59] Which I'm sure does effect certain things like taking out the typical imp patterns, but as I mentioned before I suspect it doesn't work better when someone intentionally tries to screw up your assumptions by doing things in an abnormal (i.e. not average, and/or not optimal) way. Such a change quite likely will make them more vulnerable in general, but will help them fight "optimized" warriors. [23:10] I believe it is probably worthwhile to have a optimal response to typical patterns. This means "on average" you will win. You may lose against oddball opponents. To handle the oddball opponents in the best way possible, it would /probably/ (this is my guess) be best to avoid having your strategy depend on typical patterns where possible. [23:11] That should let optimizing help your warrior (it does better against the average code), but not hurt it (it copes as well as possible with non-average code). [23:28] I have noticed, that, everything else being the same, someone that always plays the optimal strategy can sometimes be made to lose, or at least be prevented from winning (a tie), by someone that expects that, and plays in such a way that takes advantage of that predictability. [23:35] There's a term, but I can't recall what it is, for a species that evolves such that it is very well suited to its current environment, but its method of survival is so dependent on specifics of its current environment, that they tend to become extinct when the environment changes (e.g. global climate change, a new species is introduced, a virus kills something they depend on, etc). [23:37] Species that have a harder time, in general (i.e. "are less optimal for their current situation"), are often more adaptable, and thus go on to survive (because their survival mechanism, while it may still not be optimal, isn't specialized so much it can't handle the new conditions).