CyberneticPony | 2015-05-19 23:05:18 |
I wonder if anyone has asked the question of whether the Archmage AI cheats or not.
What I mean by cheating is; does it have knowledge of the opponent's card pool? (It certainly doesn't play cards that do not exist OR modifies them in any way).
HeadphonesGirl | 2015-05-19 23:18:06 |
If memory serves correctly, I think Cooler or Estarh confirmed the archmage does take into account possible cards, but isn't given information a player couldn't have during the duel.
This was a long time ago though and I could be remembering wrong.
CyberneticPony | 2015-05-19 23:19:08 |
If memory serves correctly, I think Cooler or Estarh confirmed the archmage does take into account possible cards, but isn't given information a player couldn't have during the duel.
This was a long time ago though and I could be remembering wrong.
I don't think taking into account possible cards is cheating. If it's not cheating, then it must have a really high depth search, a good set of heuristics, or a combination of the two. Just want a confirmation because I am thinking of actually starting my own beginner's guide (as this seems to becoming a trend in this forum) and using Archmage as a potential opponent to demonstrate. If it cheats however, it really doesn't help because at that point you're better off learning how to exploit the AI's flaws and not how to actually play the game. In the case it does not cheat, it is pretty impressive they were able to create a pretty decent AI using no sort of typical shenanigans, just a good set of heuristics/high depth search/a combination of the two.
PS. Please fix this buggy post submission form at some point!
Modified by CyberneticPony on 2015-05-19 23:28:14 HeadphonesGirl | 2015-05-19 23:31:04 |
Yeah, what I meant to say was basically I think no it doesn't cheat, just does what a player would do as best it can.
Either way, I would still argue that the problem you posed holds true. Playing against the archmage probably does teach you certain bad habits. In my opinion the archmage is especially susceptible to solid resource game play, or extremely hard rushes, and only does well if you can't do either of those effectively.
CyberneticPony | 2015-05-19 23:36:12 |
Either way, I would still argue that the problem you posed holds true. Playing against the archmage probably does teach you certain bad habits. In my opinion the archmage is especially susceptible to solid resource game play, or extremely hard rushes, and only does well if you can't do either of those effectively. While I agree this is the case, I think this is still helpful as a starting benchmark; playing against players introduces a lot of variables that can often confuse someone in a guide. If a player can start beating Archmage semi-consistently, then I think it'd give them confidence in actual play.
Modified by CyberneticPony on 2015-05-19 23:36:33 Ruby456 | 2015-05-20 00:21:46 |
Archmage is LeveL 24 The BEST TACTICAL PLAYER IN THE History OF SPECTROMANCER .
He is Nott a Cheater !! Butt he Is good width some CLasses . if u choose ILLUSION You can Beat him easy ... Because the Combo waLL of ILLUSION AND acidic Rain and maybe Armageddons to ... Just must we Thinking to do a good Strategies vs TacticaL Archii and wins sometimes
against him !@! From the MidLe and the end of the Games : some 3 or 4 WaLLs of Lightning + ILUSSION WALLS AND Armageddons or Rain Stone and the STRONG Archmage Losse the Game .
Archmage is LeveL 24 The BEST TACTICAL PLAYER IN THE History OF SPECTROMANCER . He is Nott a Cheater !! Butt he Is good width some CLasses . if u choose ILLUSION You can Beat him easy ... Because the Combo waLL of ILLUSION AND acidic Rain and maybe Armageddons to ... Just must we Thinking to do a good Strategies vs TacticaL Archii and wins sometimes against him !@! From the MidLe and the end of the Games : some 3 or 4 WaLLs of Lightning + ILUSSION WALLS AND Armageddons or Rain Stone and the STRONG Archmage Losse the Game .
Btw, Is that anything wrong with your CAPS button? Or it just randomally turns on and off while typing?
Jeronimo | 2015-05-20 18:23:44 |
... Btw, Is that anything wrong with your CAPS button?Or it just randomally turns on and off while typing?
Ruby remarks only key words/letters with Caps-> Not in the sense of "importance", but in representation of a "loud reading": when you speak, you pronounce some words/letters higher than others... what Ruby does is giving his words "phonetic strenght", e.g. waLL -> listen a strong L with your imagination. The same goes with the length of them (e.g. not): Nott -> listen a strong N, followed by a long "tt" sound. Then as a cunning scientist, also enjoys experimenting with linguistic variations of English language. Ruby twistes words to his will. His supreme goal is the symbolic domination through syntactic morbidity.
The Encyclopaedia Britannica officers know your real name Ruby... Oh sure they do! Modified by Jeronimo on 2015-05-20 18:40:02 HeadphonesGirl | 2015-05-20 18:46:23 |
I don't know. I think 'wall' most of the emphasis is on the 'a,' not the 'l.'
Jeronimo | 2015-05-20 18:55:20 |
I don't know. I think 'wall' most of the emphasis is on the 'a,' not the 'l.' It's exactly what Ruby is doing. He is defying the conventional English concepts. He prefers to step beyond the normal speech, and goes with new pronunciations... His writing is not so random as you can detect many times he uses same word e.g. Butt -> "but" with strong B and long t.
He writes waLL... read it as normal wall but place emphasis in the final LL. It's Ruby talking to us as he is writing his post... It's his voice. He wants us to listen instead of reading. HeadphonesGirl | 2015-05-20 23:26:50 |
Ah, I see. It adds to his mysterious eccentricities.
Ruby456 | 2015-05-21 04:50:22 |
Jeronimo
HeadphonesGirL ,,, About Jeronimo i can TeLL You He wass , THE BEST PLAYER WITH THE MECHANICIAN CLASS !@!
Maybe STeeL he hadd this SKILL today
Maybe STeeL he hadd this SKILL today
Jero, Does your theory approves spelling MiStAkes? Or as you said, He is speaking to us rather than writing post, Spellings doesn't matter?
Modified by DKill on 2015-05-21 14:44:52 Cooler | 2015-05-21 18:29:56 |
I don't think taking into account possible cards is cheating. If it's not cheating, then it must have a really high depth search,
Search depth for Archmage is 3 semi-turns. Deeper search doesn't make it better, just worse.
Also deeper search makes it much slower which is crucial, since we need to run many-many millions of battles for constants tuning. Modified by Cooler on 2015-05-21 18:42:17 CyberneticPony | 2015-05-21 20:07:19 |
... Search depth for Archmage is 3 semi-turns. Deeper search doesn't make it better, just worse.Also deeper search makes it much slower which is crucial, since we need to run many-many millions of battles for constants tuning.
Only 3 semi-turns? That makes me feel a little dumb, but I guess there is a huge breadth space to account for.
I did find the "deeper search doesn't make it better, just worse" amusing; I wonder why that's the case.
I guess you might not need a huge search depth if it's the case that "a better state 3 turns in the future correlates to a better state 6 turns in the future"; ie inductive reasoning. Modified by CyberneticPony on 2015-05-21 20:10:27 Tendou | 2015-05-21 20:24:00 |
... Only 3 semi-turns? That makes me feel a little dumb, but I guess there is a huge breadth space to account for.
I did find the "deeper search doesn't make it better, just worse" amusing; I wonder why that's the case.
I guess you might not need a huge search depth if it's the case that "a better state 3 turns in the future correlates to a better state 6 turns in the future"; ie inductive reasoning.
I guess there is a factor to react to opponent moves and thinking and countiong beyond 3 turns may hurt flexibility in some sense.
Cooler | 2015-05-22 10:05:08 |
I did find the "deeper search doesn't make it better, just worse" amusing; I wonder why that's the case.
Level of uncertainty (I don't know better word, sorry) increases very quickly with depth. In open games (like chess) you always consider possible positions regardless of their depth, but here you have to consider positions that can't happen at all, and there is very little sense in wasting time on that. Good sophisticated position rate function is more important than brute force. CyberneticPony | 2015-05-22 13:19:28 |
... Level of uncertainty (I don't know better word, sorry) increases very quickly with depth. In open games (like chess) you always consider possible positions regardless of their depth, but here you have to consider positions that can't happen at all, and there is very little sense in wasting time on that. Good sophisticated position rate function is more important than brute force.
Thanks for that, this really helps me as an AI student! |