#kothorg

[6:05:21]digital.wilderness : <@UBZCDSVKN> Is this the one you're looking for?
[11:56:53]john.kipling.lewis : Okay... that's incredible. How did you find that?
[12:04:10]john.kipling.lewis : 1997... now I feel old again. :rolling_on_the_floor_laughing:
[12:06:57]roy.van.rijn : Now I’m firing up my old evolver… just to see if it can create something that does something (written in Java, is very stupid, but quite fast)
[12:07:38]roy.van.rijn : I mostly use that code for optimizing warriors, for example the warrior created/submitted yesterday that uses some more advanced code to calculate candidate step sizes
[13:09:27]john.kipling.lewis : I wish I had thought of an GA for optimizing step sizes. That would have been smarter (and useful) than what I created. (in perl 5!) It just makes random Redcode noise and then recombines it with "successful" version with some mutation rates. Nothing amazing ever came from it... that I know of.
[13:15:20]roy.van.rijn : That is what my ‘evolving’-code does, mostly… I just started it and the best is has done up to now is: ``` ;redcode ;name a7cf7452-12b9-408f-be28-cf70666e90f0 ;author Roy van Rijn ;assert 1 START spl.ab #242 , }2 mov.i }115 , {432 mov.i >115 , {500 djn.f #0 , <-30 ```
[13:24:44]john.kipling.lewis : Dude... that's ... very close to this: ` ;redcode-94 ;name invicta duo ;author John K. Lewis ;assert 1 step equ 5211 spl 1 spl 1 spl 1 ant spl @0,step mov }ant,>ant pic mov 64, step+64 add #6117, step-1 jmp -2 for 60 dat 0,0 rof index for 7 dat index, 10-index rof `
[13:27:59]john.kipling.lewis : OH... <@UBX2E0GR4> is for/rof implimented on the site? It didn't seem to work for me.
[13:38:06]roy.van.rijn : Hey John, about that warrior, can I ask a couple of questions? The ‘paper’ part of 5 lines long, but you are starting 8 parallel processes by doing spl 1/spl1/spl1. Instead you could do: ``` spl 1 mov -1,0 mov -1,0 ``` Other improvements can be experimented with, have you tried duplicating the two ant-lines making it a double copying paper? Or changing the jmp to a djn-stream?
[13:56:38]john.kipling.lewis : The idea was to "bomb/clear" the extra spaces below the paper with dat code from empty space. That's the reason for the extra processes.
[13:58:02]dougajmcdonald : <@UBZCDSVKN> Speaking to Gareth, for/rof is implemented but only in a simple way
[13:58:09]dougajmcdonald : so for example, this should work:
[13:58:18]dougajmcdonald :
[13:58:30]dougajmcdonald : but I don't think labels with for/rof do
[13:59:20]roy.van.rijn : <@UBZCDSVKN> That doesn’t really add much I’m afraid, but the bomb-step add interleafing with the next copy is a fun thing
[14:10:52]john.kipling.lewis : <@UBQGHU2CC> I was fascinated with 4 line micro things. Papers, bombers, scanners... etc. I loved the idea of a 4 line thing you could "throw" around the core and spawn out a process.
[14:22:48]john.kipling.lewis : Here's an example of a 4 line code that kills scanners. ``` ;redcode-94 ;name Agony Killer ;author John Lewis ;strategy Jump Jump Jump ; ; spl 1 mov.i <2, <1 jmp @0, -7 dat #0, #0 ```
[14:23:12]john.kipling.lewis : I think this was the birth of the other code above.
[14:25:02]john.kipling.lewis : Yes, the missing link between:
[14:25:10]john.kipling.lewis : ``` ;redcode-94 ;name invicta ;author John K. Lewis ;assert 1 spl 1 mov -1,0 mov -1,0 ant spl @0,1221 mov }ant,>ant mov >ant,>ant mov >ant,>ant jmp ant,{ant ```
[14:27:45]roy.van.rijn : The idea of adding to a different paper can be used to create something fun, I’m going to experiment with that if you don’t mind :wink:
[16:44:02]john.kipling.lewis : I'm sharing to share. Let me know what you come up with!
[17:21:17]roy.van.rijn : This is the paper I’m experimenting with now: ``` spl 1 spl 1 spl 1 pStep1 equ 6846 pStep2 equ 2001 pStep3 equ 5450 pStep4 equ 4496;5931 pStep5 equ 7597;1443 pStep6 equ 1590;6558 pap spl @0 , }pStep1 mov.i }-1 , >-1 spl @0 , >pStep2 mov.i }-1 , >-1 add.x #pStep3 , -pStep2+1 mov.i #pStep4 , {pStep5 mov.i #pStep6 , <pStep2-1 djn.f -3 , <pStep2+pStep2-2 end ``` Not a lot left of the original… I admit, but the idea of using pStep2 (now twice) is the same
[21:11:40]john.kipling.lewis : How is the warrior doing? What meta does it beat, what is it losing to?
[21:19:05]roy.van.rijn : <- inspired by it
[21:37:23]john.kipling.lewis : Interesting!
[21:42:41]roy.van.rijn : (it is the paper mentioned above with a qscan and boot)