At the end of a row of improvements stands a weaker AI?

User avatar
Gimric
Script
Script
Posts: 24
Contact:

At the end of a row of improvements stands a weaker AI?

Post#1 » 21 Apr 2017, 18:58

Sorry, if this was asked before:
This happens twice to me. I usually copy an AI, let's call it "Reaper", make some improvements (always my hope) on the higher version and let Reaper MK II fight against Reaper (I like AMBIDEXTROUS for that). If "MK II" wins, I copy it and so on, maybe up to "Reaper MK V". I was pretty satisfied after some hours of improvement on my best AI. After a while it came up to my mind to let Reaper (the original) fight against Reaper MK VI and shockingly "MK VI" was totally defeated.

Reaper MK II hits Reaper
Reaper MK III hits Reaper MK II
Reaper MK IV hits Reaper MK III
Reaper MK V hits Reaper MK IV
Reaper hits Reaper MK V -- !!!

At the end of a row of improvements stands a weaker AI? How is that?
I know:
Fixes break an AI more than bugs do
(Kanishka), but how can I explain this?
I got a new sensation
In perfect moments
Impossible to refuse
INXS

pier4r
Skynet
Skynet
Posts: 3390

Re: At the end of a row of improvements stands a weaker AI?

Post#2 » 21 Apr 2017, 19:27

http://www.reddit.com/r/Gladiabots/wiki/players/pier4r_nvidia_shield_k1 -> Gladiabots CHAT, stats, insights and more ;

mcompany
Autonomous Entity
Autonomous Entity
Posts: 872

Re: At the end of a row of improvements stands a weaker AI?

Post#3 » 21 Apr 2017, 19:34

(In regards to the last forum link you sent, dammit pier for stealing my post). I think the last forums link is actually the most likely to be the cause of what is going on. Simply beating one AI does not guarantee that the AI is necessarily better, but rather, that it could just beat your own AI, and most of the times there is something that may go overlooked in a ”fix” that might make the AI more likely to lose in more situations, or you might run into a situation that is much more of a rock-paper-scissors deal. Since what wins isn't the only way to determine the strength of an AI, but rather how the AI wins does

User avatar
Kanishka
Skynet
Skynet
Posts: 1421
Contact:

Re: At the end of a row of improvements stands a weaker AI?

Post#4 » 21 Apr 2017, 19:49

*bow*
Fixes break an AI more than bugs do. :ugeek:

Gladiabots Off-Topic Chat


My Stats: Kanishka_RN3;Kanishka_MiPad

User avatar
Gimric
Script
Script
Posts: 24
Contact:

Re: At the end of a row of improvements stands a weaker AI?

Post#5 » 22 Apr 2017, 00:39

:idea: Thank you pier4r and mcompany, very kind and informative -- thats a lot of stuff to think about! :idea:
I'll try to test my new AIs in more different situations (and on the "field of glory" -- see you on the other side).
Merci.
I got a new sensation
In perfect moments
Impossible to refuse
INXS

ultral
Script
Script
Posts: 13

Re: At the end of a row of improvements stands a weaker AI?

Post#6 » 24 Apr 2017, 22:08

Gimric wrote:...

Reaper MK II hits Reaper
Reaper MK III hits Reaper MK II
Reaper MK IV hits Reaper MK III
Reaper MK V hits Reaper MK IV
Reaper hits Reaper MK V -- !!!

At the end of a row of improvements stands a weaker AI? How is that?
...


:lol: the same here. I have reached quite high rank and then added some "improvements". The result is going down.. but you made my day.

harthag
Automaton
Automaton
Posts: 197

Re: At the end of a row of improvements stands a weaker AI?

Post#7 » 24 Apr 2017, 22:12

Perhaps a slight variation would help:

Reaper MK II hits Reaper
Reaper MK III hits Reaper MK II and Reaper
Reaper MK IV hits Reaper MK III and Reaper MK II and Reaper
Reaper MK V hits Reaper MK IV Reaper MK III and Reaper MK II and Reaper
...etc...

User avatar
Ritter Runkel
Neural Network
Neural Network
Posts: 498

Re: At the end of a row of improvements stands a weaker AI?

Post#8 » 24 Apr 2017, 22:41

I think the main problem when testing against your own AI is, that you find tweaks to trick your old AI. Information you can get from this type of test is, of your AI can be tricked easily.

Maybe you should check with the AI you like most (reaper or reaper mk v) why it loses against the other ones. Sometimes it's only bad timing ... Nothing to worry about. Sometimes you find something to fix.

Btw,I use the same map if I want to test improvements in my attack and in my regrouping module.

MaDRaY
Hello World
Hello World
Posts: 1

Re: At the end of a row of improvements stands a weaker AI?

Post#9 » 25 Apr 2017, 04:17

Usually is just "luck", try all maps!
A good exemples is my last victory against
mrchris, is AI is better than my "nex" , my AI was an experimental one ( I was testing tags)
But somehow my bots were good haha
I really belive that making an AI is a static study :what wins more often?
What tipe of defense is usually more successful?

User avatar
Gimric
Script
Script
Posts: 24
Contact:

Re: At the end of a row of improvements stands a weaker AI?

Post#10 » 27 Apr 2017, 12:16

Thanks a lot!

@ultral: it's mostly two steps back and one step forward. Double the nodes for one little advance. In the next days I'll be back in tournament. :)

@harthag: You're right... This was obvious, but I didn't realize it. *sigh*
It's as always: "Before the success the gods have set sweat." :D

@Ritter Runkel: I learned a lot from the answers to my question. I'm now testing different maps and different opponents.

@MaDRaY: Haha! I should do this as a full-time-job to check out every possibility! :geek: Usually I "fix" AIs on a daily base and on a weekly base i check out how the "newbies" in all 8 maps whether they pass or fail.
I got a new sensation
In perfect moments
Impossible to refuse
INXS

Return to “Questions”

Who is online

Users browsing this forum: No registered users and 1 guest