Replay System + Input Tracker from '/bot teach'

Suggestion type:
FGunZ


Suggestion description:

Hey, i started testing macros on bot system when something came up to my mind. bots react pretty much different when a player is macroing which makes easier to identify. Would be by any chance a good idea implementing a input tracker into replay system such as '/bot teach' to ease/improve report analysis?

It's just a guess (bc i'm not at home to test) that replay system tracks player inputs and position in the same way bot system do, since one is based on another (only difference would be the timeline control that could maybe modify the expected output)... then when a macro is used you could visually see in the replay this player going slow during inputs or pretty fast right after he/she used it.

Also this would open a door to a machine learning algorithm identify possible macroers in the future if it's sufficiently accurate (it's what i'm aiming to do for fun)

i know its none of my business how u deal with them, i'm just supposing it could help.

Also i didn't really thought how macroers could hide it better from this probably "ass solution" i came up with since i started playing around with macros a week ago.

Bots would behave differently when a macro like a speedy dagger slash is used, but when a macro that repeats an advanced k-style is used, a player can easily hide it by making the moves timed humanely. On the other hand, a player recording a move is also capable of making the bot mess up when repeating it due to imperfections in its timing implementation. I don't think this would be an easy task at all to automate and a human eye is most likely much better and time efficient. A machine learning algorithm can definitely automate this process to a relatively good degree but obviously that has to be written, tested, trained, and confirmed against judgement from gms, which takes time and probably not even worthwhile time-wise for a game without thousands of players unless a player takes it upon themselves to do.