I actually disagree, at least as far as the site’s favorability is concerned. an auto captained open pug should probably never have worse than .55-.45 calculated favorability. you can just do 1 to 1 swaps after choosing the players and roles
to even out the ratings between each team. there are only 32 combinations to try before figuring out which one gives the closest to 50-50 favorability. of course the ratings themselves being inaccurate will throw this off but provided sample size the ratings will move to approach more accurate numbers. ive seen the auto picker pick teams that it calculates are 75-25 and this should literally just never happen. it wouldnt be hard to implement this either, and it would be fully independent of whatever logic they are using for picking the players / roles in the pug.

fun tournament join up for the next one

https://soundcloud.com/solidsoundfm/djwitchgirlmiracle-producer

Shoutout to my beautiful friends and also the 2 middle aged dudes playing rock band in the corner

you can do first come first serve and still avoid favorability that bad. have the system get the pool of 12 ppl that will be in the pug (and even what classes they will be on, if you want to make optimization simpler) and then have find the combination of players such that the total rating is the closest to even. if you lock the classes, there should be only 32 ways to slice it which should take literally no time even if you use just brute force to find the most even favorability.

Shooting rockets at gullywash choke

I feel like looking at demos at a speed where every tick lasts nearly a second can throw off your perception. TF2 doesn't actually record POV demos at that high of fidelity, the aim is just interpolated from the position it was on the previous tick. it seems pretty possible for this to trick you into finding "impossible, cheating" aim from something that in actuality could be explained through natural causes.

below is my artist's depiction of where his crosshair actually is at the time of each tick. the second diagram is how it looks on the POV demo due to interpolation and you can see the hard stop to the flick and downward motion. the third diagram is how it could have actually been aimed, and you can see overflicks and acceleration begin in the opposite direction at the end of each direction of the flick. obviously I can't know if this was the case, but you can't know that it wasn't either, since the data just isn't there.

https://i.imgur.com/LKt1Kcn.jpeg

I would also like to say that his aim lingers around the medic's head for around 2 ticks and doesn't instantly start snapping back to where it started; this is consistent with aim deceleration and just having decent click timing. as for the aim returning to where it started; this just happens for a lot of people who aim like this, sometimes. finding a couple clips of this isn't exactly evidence, you need a larger sample size.

slowmos from ur own video. yea he moves his mouse a little more in the medic shot but the other shots demonstrate the same aiming habit, it's just a slightly larger flick. a flick where he knew he was dying in like 0.1 seconds and just went for a yolo shot btw (anyone would aim harder in that situation). I used to flick like this myself before i improved my mouse control and yea you don't always return exactly to where you started your aim, but sometimes you just do. that's human variance for you (not to mention he actually returns his aim to slightly lower than it was before the flick)

https://youtu.be/rf5Hxh4-W1M?t=156 (aim actually doesnt move in a straight line when he flicks up and left, but it does move in a straight smooth line down and right to where he started)
https://youtu.be/rf5Hxh4-W1M?t=165 (rubber banding motion, altho doesn't go exactly back to where he started)
https://youtu.be/rf5Hxh4-W1M?t=189 (rubber banding motion, goes in same direction as where he started and like 80% of the distance)

not saying that he isn't cheating, he very well might be and some of the clips in this thread are weird. none are conclusive though. I looked at the old videos too, where people were commenting on the weird flicks/180s/target switching, and I thought that all of those decisions seemed reasonable enough from the information he would have in game

hes not banned by anticheat because nothing posted in this thread is even close to conclusive proof that he is cheating. to me it looks like he aims the same on most every shot and sometimes he hits really crazy looking ones because he is flicks hard and fast. in that comparison video you can see the same mouse movement, flick back and forth on the shots he misses as well as the medic shot at the end. the flick on the medic at the end also isnt nearly as much mouse movement as it looked like to me at first, the medic counter strafes and the movement of the scout in front of him makes it look like a bigger flick than it is, I would just diagnose it as a lucky shot. not saying that he is not a suspicious player overall but if rgl banned off these clips they would be stupid

I am kinda curious what the picking algorithm actually is. It doesn't seem to follow traditional logic of how you would implement things like this (highest elo in the pug gets picked last, etc)

united kingdom has mike. also new age players like toemas and papi and stuff

2 pieces of toast puked one of em up

https://youtu.be/v3tYrZVrX8g?si=apaKEsZL3KITswCc

hello bump i need a team... if u need a last minute player hit me up

as a noob i think quake is fun but i can only bring myself to duel friends. and usually they only want to play for like an hr, so maybe the game isn't that good sadly