Didn't see this on the roadmap....
So the basic idea here is that at the end of the fight we get a stat breakdown of what our bots did.
So at the end of the fight we get something like:
End of fight stats could really help us figure out what our bots are doing in a non-results orientated way. I think this might be useful to enable us to benchmark and compare AI strategies.
I don't know about everyone else, but I often get my bots to fight themselves and make one change; and then I presume that whatever of my teams won has the better AI. that is of course a results driven assumption that might not actually be true. With the stats I might get a better picture of whats going on.
We could start out small (e.g list above) and add more detail over time. E.g in the long-term having an ability to see how often an action/node triggered during a fight could be useful. Maybe some of my nodes are never firing and I haven't noticed. Thus, maybe the stats help us detect bug and/or redundant code.
So the basic idea here is that at the end of the fight we get a stat breakdown of what our bots did.
So at the end of the fight we get something like:
Code: Select all
Shield/Health Damage done: X/X
Shield/Health Damage taken: X/X
Confirmed Kills: X
Resources collected: X
Resources secured: X
Distance Travelled: X
etc
End of fight stats could really help us figure out what our bots are doing in a non-results orientated way. I think this might be useful to enable us to benchmark and compare AI strategies.
I don't know about everyone else, but I often get my bots to fight themselves and make one change; and then I presume that whatever of my teams won has the better AI. that is of course a results driven assumption that might not actually be true. With the stats I might get a better picture of whats going on.
We could start out small (e.g list above) and add more detail over time. E.g in the long-term having an ability to see how often an action/node triggered during a fight could be useful. Maybe some of my nodes are never firing and I haven't noticed. Thus, maybe the stats help us detect bug and/or redundant code.