Profile of Reverb

Jump back to quoted post25 Aug 2022, 09:40 AMHannibal

That's also not data, that's a claim.

Anyway, I think a reasonable one. According to stats that SiphonX posted a couple of years ago, there's huge differences between modes. 2v2 features probably around 40-50% or AT, that mode would absolutely be hit hard unless player counts stably grow to at the very least double the number.

3v3 has approximately 20%-ish of AT teams, although bear in mind that his data is actually biased towards having a higher ratio of randoms, so it could be more. Nevertheless, 3v3 could probably take the hit if the game grows a little.

4v4 surely can take it, there's apparently not many ATs, although I assume there is a high ratio of 4 people just playing a couple of games before being ranked, but overall the majority will be randoms.


Since the majority of my games are AT games, obviously I am biased towards keeping AT games playable. However, claiming that just splitting the playerbase will make it better for the majority is oversimplifying quite a bit.
Most complaints at least here on the site are mostly about 3v3 and 4v4, and not necessarily about AT vs RT. This makes inherent sense, since these are the largest mode regarding player counts, but it is also a sign that the matchmaker itself is the major problem. Second, AT win rates are definitely higher, but not "unwinnably" higher. This definitely needs fixing, but can be tweaked by just artificially upweighting the ELO of ATs so that they get matches with more skilled randoms.
Next, even having only 80% of the player available in the case of 3v3 (assuming similar player counts as current CoH2, and CoH2 has never been in better shape player wise than now), will diminish the player base overall. Matchmaking in 4v4 can be very shoddy already, despite mostly randoms being around there. That's again mostly a matchmaking issue. And finally, it is doubtful if splitting the modes would overall help the player base. You'd likely get more randoms into the games since they are overall less frustrating, on the other hand many players probably would play the game less or not at all if some modes they like are basically dead. At least for me I can guarantee that I would not play CoH3 if I knew that 2v2 and 3v3 AT are functionally dead modes, I wouldn't buy the game for the random games I do.

Splitting consequences of splitting the modes is mostly speculation, but the easiest way to go about it is probably really to just upmatch ATs until their winrate matches the ones of RTs. There will still be odd issues at both ends of the ladder, but those already exist now, so no harm done.

Jump back to quoted post23 Aug 2022, 16:20 PMTickTack

Elo comes from a desire to match people up with similarly skilled opponents. Is it even relevant in random team mode?

How much worse would it be to have just random matching? Newbies and vets flung together, it's certainly more realistic.

If vets want to be tested they should be able to queue specifically for an elo-rated game, solo or as a team.

If there are more newbies than vets in the population, then probability dictates that, over time, random teams will be a couple newbs and a vet. Not the curb stomping vet Vs newb train that game designers seem to fear.

In short, I hate elo for random play, but like it when I want to test myself.

Elo seems more fashion than science. It's often implemented so thoughtlessly

Jump back to quoted post21 Aug 2022, 09:16 AMHannibal

This system probably won't work at all, because the only way for randoms would be up. What could happen though is to weight wins and losses accordingly. However, if you already have discovered the needed weight to make it fair, you can just set up a proper match up in the first place (i.e. lower skill arranged team vs higher skill randoms). They should have tons of data of CoH2 to make a decent guess for CoH3, how much stronger an arranged team is, just because they are "arranged" and not random.