2.75 Playtest Results
 

For curiosity I let the bots play themselves for 2.75 hours, 4v4 and these are reults. Great Work Stefan.

www.skirecs.dynu.com/playtest.jpg

Re: 2.75 Playtest Results
 

cool, i wonder how the newer bots would perform there :)

Re: 2.75 Playtest Results
 

Which newer bots?

Re: 2.75 Playtest Results
 

look @ progress thread. I am working on a next WIP version.

Re: 2.75 Playtest Results
 

i tested accidentaly :D wip 9 (3005) for 7 hours in 5on5 on one map (i don;t remeber wchich one though it was 3 days ago) (i went to sleep and forgot to turn off machine)
but the server never crashed, aditionally the team scores were like 235/234 !!!
but there were lags when i joined in the morning - i guess it was coudes by the experinece data...
yesterday i did the same (yawn..) on proper mapcycle - the all works fine

i just lack of some good waypoints thogh

running linux steam with amx 0.9.8
still more tests to go

Re: 2.75 Playtest Results
 

nice, the bots navigation should improve severely compared to the first hour of playing. I just re-noded de_dust and saw some bots just want to get on a bit to high ledges, but after some time you see they do not try that anymore :)

true - seem they behave more natural after few meinutees of fresh play, but as soon they get skill they get mostly the most safe way with the least encountered problems in the past :/

so tehy are learning but not always as we want to :D

Re: 2.75 Playtest Results
 

i have already solved the 'randomization' issue. I see the important area definitions are working extremely well. You can force bots to be dynamic this way. I will create files for most maps for the next release.

Re: 2.75 Playtest Results
 

someone should make a playtest in this way :
mix bots on a server
5 realbots vs. 5 pod-or ivp-or whatever bot
and let this server run for some hours
that would be interisting i think
but i have no time for this

Re: 2.75 Playtest Results
 

problem with bots vs bots is that the winner is not always the best or funnies bot to play against.. reactions aim and so on..

Re: 2.75 Playtest Results
 

yes but it could be a little competition for the developers who made the deadliest bot

Re: 2.75 Playtest Results
 

i don't think you should look at the scores to determine a winner. In fact yo ushould not determine winners at all.

You can however compare in what aspects bot differ, do some bots use sounds well? Are some bots good at objectives? Do they manage to get around the level quickly without problems? Etc.

Do mind that most things like navigation stuff is pretty relative, as most of the time bots are in combat.

Re: 2.75 Playtest Results
 

You should primarily look at the win ratios. Which team actually won more often, not just got more kills.

It doesnt matter if your ratio is 4:1 if you lose every round... and once you die you have no money to buy back that precious awp. You know what I mean?

That's the problem with playing online, so many pussies on each team which will camp out and hide, and screw over those people that actually do want to complete the objective. Then they buy awps, die, and the cycle repeats. How boring.

Bots fix this by having no real sense of self, they are programmed to do something and they will do it. If they are programmed to accomplish the goal, by any means necessary, they will do so. Many bot-makers get too caught up into making the ultimate stealth killing machines, and forgetting that somebody has to play the Ts once in a while.

Re: 2.75 Playtest Results
 

and all along i thought they programmed a bot that will play like human..

Re: 2.75 Playtest Results
 

my goal for the bots is to make them play as i experience players online. Currently the wip version (you see as in 'not released') are quite stealthly when they hear someone. But not all the time! I admit they prefer stealthy stuff most of the time, but that is how it is on on-line servers. Put in lower skilled bots and the bots won't care a lot about stealth though.

Anyway, when i play and i hear someone coming most of the time i stop, walk (or duck) to a position to get his position fast and shoot. Most of the time i can guess when he will 'be coming around that corner' so i shoot before i see... and sometimes people say i am a cheater (yeah,12 year old kiddies :))..

When playing against my bots now, i know i cant run around, because when i do the bot will hunt me down (he will go after the sound origin, but in stealth mode...). I have had a few really 'OMFG IT SCARED THE #()*$&@# OUT OF ME" moments because bots where already silent and rushed around the corner. BAM, headshot...

i like it :) I think when WIP 10 should play against WIP 9, WIP 9 would be slaughtered.

Re: 2.75 Playtest Results
 

i think im gonne install hl and cs again.. maybe this weekend.. problem is i dont own a cd key for hl.. :/ maybe i can borrow one..

Re: 2.75 Playtest Results
 

yeah the sound feature of the bots will be really cool i can imagine when you look at RACC dev.-sreenies you can see he will listen to sounds too that will help a lot to make the bots better i think