Matchmaker Pool Feedback Thread

@waffelznoob said in Matchmaker Pool Feedback Thread:

Just a reminder that these low rated players who have maps learned by heart, still are low rated

And low-rated players that do NOT have maps learned by heart, are also low rated. I'm not sure how this is significant?

Thanks tons for the reply btw.
I've lost a bunch of games from not knowing map-specific 'tricks'. Like huge reclaim fields in certain spots, or edge-building (cliffs) to allow expansions without transport craft.
While much of this is a learning curve, the idea that new players should be extra-vulnerable to 'I don't know this map' losses seems odd to me.

I can understand it in most games, of course. But in a game with auto-generated maps in the ladder pool (which is absolutely AWESOME!), I can't see why the feature is reserved for high ELO.
It's great for veterans, sure, but it's great for new players, too! For aforementioned reasons, as well as variety, and testing a player's ability to adapt to new situations quickly.

I just think random maps sound awesome, and it sounds like they should be bestowed on new players, too. (I can't think of a reason not to, and can even think of a few reasons that it is extra-important!)

If you couldn't find the huge reclaim field in a premade map, you wouldn't find it in map gen either. That's just knowing ctrl+shift is important in looking at a map when you spawn in and are unfamiliar with it.

Edgebuild is still kinda fair (as a complaint) just because the build range circles for units is still not integrated into base game iirc so someone with the ui mods to press shift and see if their engineer/acu can reach on top of a cliff is going to have that competitive advantage. But that's still something you can find out before fully playing a map the first time. Assuming you do have that ui mod, anyway.

The logic of building up to map gen is that lower skill brackets need to "learn the lessons" of some maps in a more controlled environment. This map style forces X or makes Y more optimal because of Z or W reasons. Map gen has so many externalities that it just increases the noise and therefore the pains of learning said lessons because you can't actually isolate certain things in a map as causing certain interactions.

@ftxcommando your suggestion do disallow big rating spreads is crazy. Imagine you manage to convince a friend to try this game only to then be denied the option of playing with them. Are faf promotional efforts that good at attracting new players that you're ready to sacrifice this basic human interaction just to have ratings converge a bit faster? How are you gonna justify banning smurf accounts after you make that the only way of playing alongside your beginner friend?

If a change to tmm is really warranted because of this problem, my vote is for configuring the matchmaker such that it tries to find a similarly structured opposing team, rather than what you suggest. If a 1500 + 1500 + 100 premade is in queue, it's ok to match them vs 1400 + 1400 + 300. TMM is unavailable half the time simply because not enough players are searching for a game, any further restrictions on team compositions would only make that worse, and for bigger games, your proposal constitutes a harsher restriction than mine. But there's a delicate balance between wait times and match quality that needs to be struck, and if the scales are tipped too far towards the latter, people stop queueing because of the wait times, compounding the issue, so I'm not convinced the problem justifies the cost.

As for custom lobbies, in addition to the argument above about friends playing together, there's the fact that for many noobs, astro lobbies are the only option. No matter how much we wished it wasn't the case, if a noob hosts any other map, their lobby won't fill unless the title is "all welcome". This is because other new players rarely get in the mood to try something else. When they do, the friction imposed on them should be minimal.

Custom games can be reasonably well balanced if 2 noobs join, especially if the map layout allows the host to put them against each other, and it's far preferable to hoping a noob is willing to try hosting, and they're in the mood to pick a map they never played before, AND 7 other noobs decide to join. Not gonna happen until the nr of players online is an order of magnitude higher.

Besides, since when is global rating such a treasure to cherish when It can be gamed so easily? Why put up such outrageous barriers to play just to protect this flawed system from a bit of variance, when it's been the general guiding principle of faf development for some time to steer players towards ladder/TMM and league standings anyway?

Tbf mapgens tend to be simpler than designed maps, so they may indeed be well suited for low rating players (i'm mainly thinking the 7.5km size for low rating 1v1 ladder).

Mapgens are very well suited for low rated games. They are way less intimidating. The simple fact of the matter is that new players have repeatedly expressed their discomfort with having to learn maps as well as game mechanics at the same time, and even though sometimes they communicate that discomfort poorly, no amount of quibbling over what "build orders" actually are is gonna make their reluctance go away. If the goal is to get noobs to play ladder instead of custom games, they should be included more for that reason alone.

I'm not sure teaching players specific game mechanics should be the main goal with ladder map composition. Even though we call it ladder, the vast majority of people actually stop climbing it as soon as the rating system zeroes in on their skill level, and it should primarily be fun to play at your given level as opposed to a challenge to learn boat or transport micro or whatever, at least in my opinion.

Maybe once ladder/tmm games become prevalent, the focus could shift, but for now, my understanding is that custom lobbies still dominate.

@sylph_ said in Matchmaker Pool Feedback Thread:

@waffelznoob said in Matchmaker Pool Feedback Thread:

Just a reminder that these low rated players who have maps learned by heart, still are low rated

And low-rated players that do NOT have maps learned by heart, are also low rated. I'm not sure how this is significant?

They are low rated despite having an advantage in map knowledge. Assuming you are equally rated, that means they must have a disadvantage elsewhere, or else you wouldn't be equally rated

I also personally thought learning maps was fun. It is the one thing you see improvement the fastest in. Make a huge mistake in your BO on theta passage? Ok, try something else next time
Make a huge mistake in your mapgen BO? Too bad, you're getting an entirely new map next game and you'll make the exact same (or some other) mistake

This extends past build orders into general gameplay, too

frick snoops!

@phong Hmm, yeah, disallowing/punishing high rating spreads for people queuing together does sound like a bad idea.

It's partially a problem of low player sizes and partially a problem of the ranking system having to rate many different setups with the same number.

Two 1.5ks do not play the same as one 2k and one 1k afterall.

For custom games this is less of a problem because the rating is way less meaningful for one, but also because you can see all players in your game before the game starts.

For tmm I think the best way is to further punish high rating spreads in the search algorithm and to maybe find a way to make the queuing with a lower rated player uneccesary for high rated players.

Maybe once you hit like 1.8k you get a counter on your matchmaker tab showing you how many players of your rating bracket are currently queuing/in a game to avoid the frustration of queuing for 30 min in a full queue w/o finding a game? Dunno

@cheeseberry I think the problem is that people have too high expectations of the matchmaking system, not realizing how crippling the small player pool is to any such system. We can't afford to have perfect matchups when half the time there are 0 people in queue.

The league system is exactly what was needed but I think development of it stopped at a critical point. If those shiny badges replaced rating numbers everywhere in the faf UI, lobby, scoreboard etc, then they might actually become desirable, and we'd finally stop referring to people as "the 1500" like they're wearing prison jumpsuits.

If the devs could count on queues containing hundreds rather than a handful of players, the matching algorithm could easily be tweaked to address all these issues.

@waffelznoob I think learning new maps can be fun, but I also can see how, a new player especially, struggling with the basic game mechanics, might want to postpone or minimize engagement with specific maps. It's a perfectly natural instinct to focus on skills that generalize, and since the game has such a steep learning curve, mechanics naturally rank higher than map-specific knowledge. New players feel inexperienced compared to their ladder competitors, even if they know nothing about them or their familiarity with bespoke maps. This is why the perception of a level playing field can make ladder more attractive even if it's based on an unjustified lack of confidence.

There wouldn't be an "if" I would just not allow those teams to form by automatchmaker nor be a premade.

@ftxcommando if no one gives a shit about new players we'll all run out of people to play with sooner rather than later. You're short-sighted.

Nothing quite like that new player experience of being used as a slave to farm 500 rating because rating systems falter at large variances and giving this 300 trueskill player 900 rating which will now take 3 times as long for him to reach his actual rating due to the impact of sigma.

Thanks for your long term strategic vision.

Your buddy somehow managed to convince you to install this old-as-balls game but you can't play with your friend because you're too new and some asshole decided that's problematic for some reason. do you A: grind alone until you reach their level or B: open your steam library and play something else?

Fuck rating, it's never gonna be perfect. Fun is more important. It's a game.

Also your proposed solution to the problem is shit when it doesn't need to be.

And I don't see the externalities of this system impacting me via the form of having to deal with 2000s that are actually 1500 or 1600 and kicking them because I don't feel like taking the tragedy of the commons cost of dealing with him ruining 30 games before he has a proper rating while also assuming he doesn't go and farm the old rating back again as fun.

Likewise I don't see games where players 1000 rating above others are around as serious games, nobody is playing them seriously. They're just messing around because nobody wants to tml the 0 rated dude's mexes, arty drop him, strat rush him, etc. The ones that do play it this seriously are the ones then are then either stuck in this loop of only playing these games because that's where their rating came from, risk getting banned for rating manipulation by deranking elsewhere and giving 400 rating to others to spread the problem around, or just ruining 2 dozen lobbies of another game variant. People are just playing as chill as possible. Like just play unrated games if you want to show a guy the game. Or coop, or ai games. Tons of options available beyond harming the integrity of the rating system.

Fuck the rating system and it's integrity. It's farmable and you need to address that before nitpicking this. Realize that before the matchmaker is any good at all it needs player count. You're not gonna achieve that without some compromise at the start.

Addressing it is stopping the problem of large variance games lol

How about you enjoy the game in whatever curated custom lobby you desire instead of telling others to play unrated? Do you even realize how hard it is to join a game as a 0 rated player? People enter the lobby and say don't kick me before even saying hi.

This isn't some uniform issue. There are plenty of new players that get pissed off at getting a game where the rating range is huge, particularly since you can get situations where the 0 rated player gets the slot that is supposed to lose against say a 1200 and the game becomes "who can gank the enemy new player faster." I read complaints from both ends about not getting games or getting shit games from every level of FAF.

I'm not even aware what the odds of not getting a game would be, I don't really see why it's a problem at low level lobbies. It isn't that difficult to get another <500 rated player to join or just launching with 2 players less. Most of the costs here are centered at high rated players that have no option but to join an 800+ game or whatever equivalent. Or queueing with 500s in tmm as a 2k. However if I had more security in being able to get an 1800+ match in tmm which would come with not having to worry about 2ks trying to curate a match via partying with 500s, I would also be more likely to queue. So that calculus is also not that simple.

It ultimately boils down to player counts. If they were high enough these problems would be easier to address. And if you're throwing new players under the bus to "fix" the issue now, you're giving up the opportunity for a better fix further down the line.

Your comments about games where the lanes are imbalaned also makes me wonder why you suggested custom lobbies need this restriction when that's precisely where the problem can be minimized by having 2 noobs face each other in a lane with manual balance.