Welcome to Smashboards, the world's largest Super Smash Brothers community! Over 250,000 Smash Bros. fans from around the world have come to discuss these great games in over 19 million posts!
You are currently viewing our boards as a visitor. Click here to sign up right now and start on your path in the Smash community!
Thats the flaw I'm pointing out. We have observed CPU not using replay data. However, you cannot use that as proof that they CANNOT EVER use replay data.
Actually, I was thinking about it working another way. Brawl has a very detailed system in handling who's responsible for what KOs (it even records people stuck under falling blocks as your KOs), so I think it might keep a count of the attacks that deal the most KOs. The CPU might then learn it. This could also works with moves that deal the most damage, and other useful stuff.
We have seen accounts of CPUs(snakes) awkwardly using dash attack, then Usmash, and others like Peach constantly roll before they throw turnips; just because CPUs may "learn" sequences doesn't mean it'll also have the exact same timing, which often messes up the AT. Also, CPUs seem to only recognize "Killing" sequences, as those are directly related to a KO. I doubt they would be able to tell if other techniques, particularily one involving short hops, were useful if they don't KO very often.
, so I can't really give an honest opinion of what I would perceive from viewed gameplay and actually playing. 
If you ever find that video definantly post it!
Well, if they learned just by copying moves, they'd still never be able to catch up since they can't mindgame in the slightest.
this wud explain why falco cpus chaingrab 2-3 times then spike or down c-stick most the time theres two cg's before an edge and then the player does a jump spike so ocassionally the cpu messes up the timeing and dsmashes. it works for falcon punches too if u did enuff falcon punches eventually it would work like how they learn crouch tuanting is 100 percent ko. it teachs the cpu that falcon punch = 100 percent success therefore falcon punch often = automatic win to the cpu
i like how u criticize us even tho there is VIDEO PROOF that they can learn
Oh ho ho, but you forget, no one has yet to even try to disprove our theory outside of Scotu and "No. You're being stupid." Former hasn't replied and latter is baseless.
O, wat now, "CPUs can't do ATs" people?
That doesn't mean anything. The game could very easily just save what "level" of learning the CPU is at along with the replay.
I don't think they have preset levels, because its unlikely they have "glide-toss" level and "falcon punch spam" level. Its probably something more like the "custom" moves are saved onto the replay, and any other time, the "standard" CPU takes over.
So you're proposing that there's a hidden level along with the displayed level.
Sorry for the late reply, but if this is the case, the wii either needs to save the full set of "learned behavior" with each replay or each replay will look different on different people's wiis.
Thats basically the jest of it. However, I want to point out that it would only record the "learned" behavior, and as an example, would switch off from a minute of un-recorded "standard" to a few seconds of "learned" behavior, so it would only have to save maybe 15-20 seconds, not 3 minutes, of actions. Furthermore, I believe that the chains of "learned" behavior are quite small (jump, Falcon Punch) so the overall file size difference would be rather small.
