Noob matchmaker queue

@ftxcommando We need the rating system because people don't want to play stacked games. They really, really don't. If they didn't care about rating, they would just ignore it. Anyone is free to host a lobby where teams aren't balanced based on rating. Nobody does that and just about nobody would join, if someone did.

@maudlin27 you claim to see a greater variety of strategies on Astro compared to 1v1 ladder, and it is because of this:

@blackyps said in Noob matchmaker queue:

you can do basically anything in lower ratings if you do it somewhat well.

Generally speaking, high level ladder is played at a much higher level than "high level astro" (if there even is such a thing). The strategies used by top players on ladder are refined and optimized over collectively tens of thousands of hours of experience, and lower level players will seek to emulate those strategies. You see a preponderance of strategies on Astro because it is being played at a much lower level and literally nobody knows what they are doing. If there were top players motivated to play astro as much as ladder, you'd likely see a set of "optimal" strategies emerge, and as lower level players sought to emulate that you'd likely see a decrease in strategic variety.

I also don't understand the logic proposed both here and in the PC election thread for removing the global ranking system.

I think it makes sense that fun and casual game modes such as Astro crater, FFA, Phantom, etc. are unranked. Imagine those game modes actually contributed to your rating. People would exploit those game modes to gain rating! Imagine being able to host an All Welcome Astro crater lobby, put any 0 rated players (that is, the 1500±1500 player with no experience) on the other team, and have a teammate just give you his eco; if you are a half decent player, you would probably win almost every game and gain a large amount of rating while doing so.

Oh wait—that already happens. I saw this player with 2300 rating a while back, and I wondered who he was. Must have been a great player I somehow missed, so I went through his replays. Turns out, he had played Astro Crater all welcomes exclusively, and every single game had his ally dump him mass for a very fast all-in Czar rush. I don't even think that strategy is particularly good; if you know it's coming, and have 1 teammate give you his eco, it's pretty easy to counter. But he was able to get to 2.3k doing only that. Come to think of it, if I played a ton of all-welcome astro craters, manually stacked the lobbies, and had both my teammates walk at the enemy after giving me an initial set of buildings every game, I could probably get to 2.5k global or more.

It's just an easily exploitable system that decreases the integrity of global rating as a whole. In theory global is already pretty useless for balancing on astro or in general because of how easily exploitable it is, but thankfully right now most people aren't like this guy or Suzuji (who got 3k rating from playing Dual Gap all welcome). In my opinion it's probably better just to keep these unranked so people can actually play for fun, especially since they aren't being played very competitively anyways.

@BlackYps Fair point, I expect I could get a much better variety in 1v1 with lower APM if I forced myself to play more relaxed and use weaker strategies (by virtue of my rating dropping giving me weaker on average opponents). That said though I'm naturally competitive so find it hard to avoid trying to play as best as I can to win.

@archsimkat Re your point on variety being due to a lower level of play on Astro, while I acknowledge ladder has a far higher skill ceiling than astro, the variety difference isn't simply due to a lower level of play on astro. E.g. For astro if I want a reasonable prospect of winning I'll need to scout my enemy regularly, micro key battles, determine my opponents vulnerabilities and exploit them, and determine how to counter my opponents strategies; by contrast 1v1 ladder most games around my rank my opponent will be doing t1 land spam, and I can do reasonably well initially if I just do t1 land spam (and am likely to do poorly if I don't do t1 land spam), with my strategic options reduced more to thinking where do I send my com and where do I try and raid and/or send my land spam based on where their land spam is.

Re the global ranking, your post actually highlights the value of the ranking system. If someone is consistently winning with a strategy (in this case having a teammate who gifts you your base on astro) then their ranking rises to indicate the strategy/strategies they use are likely to win. People will naturally look at ranking and expect the opposing team to in aggregate have a similar ranking to balance out, and the better the opposing team the less likely such a tactic is to work. People also generally join a game to play it, not gift their stuff to their teammate, so it's likely such a player will play with the same person for alot of those games making that other person's rank also increase (in turn requiring a better opposing team to avoid an unstacked game).

If by contrast you remove the ranking, that 2300 ranked player will appear the same as a low ranked player, allowing them to easily stack the game in their favour, resulting in far more unbalanced games and alot of inconvenience for people that want to play non-stacked games (who now have to resort to outside analytical options on each person that joins the game to figure out how good they are, and if you join the game as the last person or just dont want to spend ages ahead of any game researching your opponents, you're far less likely to realise you're about to waste your time on a stacked game against you). As arma mentions, people don't want to play stacked games, and on average you'll get fewer stacked games with the rating system than without.

E.g. glancing at a few replays of the highest rated astro player I could find (Harper, about 2.2k), in those games with good game quality (90%+) the games didn't appear stacked and Harper won some and lost some. If anything you're at a disadvantage at such a high rank, as there's not much difference between a 1600 player and a 2200 player on astro due to the lower skill ceiling, but there's more of a difference between a 600 rated player and a 1200 rated player.
Contrast this with the few replays I could find with a <50% game quality (there were far less of them), the team of higher ranked players crushed the team of lower ranked players.

All this is showing that the ranking system is a better than nothing guide to avoiding non-stacked games (even if it's not perfect), and also demonstrably not pretty useless (if most low GQ games are won by the high ranked team, and most high GQ games played by someone have them winning some and losing some, I'd say that's useful not useless). Most people don't like games being stacked against them, and I'd also consider it a bad new player experience.

@arma473 said in Noob matchmaker queue:

@ftxcommando We need the rating system because people don't want to play stacked games. They really, really don't. If they didn't care about rating, they would just ignore it. Anyone is free to host a lobby where teams aren't balanced based on rating. Nobody does that and just about nobody would join, if someone did.

We already don’t rate all the games I just mentioned so I fail to see the point to any of this. There is no real way for FAF to rank ffa unless we homebrew our own MMR system.

@ftxcommando
FFA literally used to affect rating on FAF, with the winner generally getting a whole lot and the losers each losing some... The wiki even still mentions a little detail about FFA's working with the true skill system... That code should still exist (and not need to be created from scratch).... If you think the amount that FFA's used to affect rating was too extreme, I'd imagine it would take less than an hour to make FFA's give a scaled-down effect on rating (ie: divide the quantity of rating change by the number of players in the game or by a more static value, etc).

PS: I'm not arguing for or against rating FFA's at this time.

pfp credit to gieb

@ftxcommando said in Noob matchmaker queue:

There is no real way for FAF to rank ffa unless we homebrew our own MMR system.

Wasn't TrueSkill made with FFA in mind and even works faster there?

@ftxcommando Just because a person plays an unrated match doesn't mean they want to get rid of global rating. From what I have seen, the vast majority of unrated matches (with humans vs. humans) are balanced based on people's rating. Even for games involving AI, like survival, the rating is used by the host as information for what the settings should be.

"Rating removal is predicated on changing the intended use of custom games on FAF and is part of a larger scale motion to shift cultural attitudes. If you make custom games an area for fun script maps and general casual things like ffa or sim mods, it no longer necessitates a rating system."

Why do we need to "shift cultural attitudes" away from people playing dual gap or astro? Let the people play.

Just my two cents as a relatively new (joined 9 months ago) now 1.6k global player:

A large part of what makes our favorite game so fun and rewarding is its complexity. Sadly, this complexity is also one of the biggest hurdles for new players. I mean, even playing sim city is hard!

This complexity is I think what many people misattribute, imo, to "APM". It's not that you need a lot of clicks per second, like in starcraft, it's that you need like two correct clicks per second.

One thing that immensely helped me when starting out was to just copy the build orders and general game plan of better players so that I at least knew what to for the first 10-20 minutes, even if I didn't yet know why.

Not only was I playing semi-competent and fun games fast, it also allowed me to learn the ropes much faster compared to me just "winging it".

This was only possible though because I could choose the same map, and preferably slot, in a custom game.

In general I agree with the slow move from custom games to the matchmaker queue.
It's almost certainly the right direction for the game and its players, even low rated ones, if the "lowest common denominator" game mode becomes a chill 4v4 matchmaking queue.

The thing I'm worried about is the people who, like me, want to reduce the game's complexity via map choice.

If the only way to play ranked multiplayer were to become a 4v4 queue with 6+ maps and randomly assigned slots tomorrow, I would stick around, because by now I have actually learned the fundamentals of the game well enough to adapt.

If that would have been the only way to play 9 months ago, I am skeptical that I wouldn't have just been completely overwhelmed for 5 games and then left, never to come back.

Yeah, I know FFA was rated. That was terrible. Why are you telling me this?

TrueSkill was developed for Halo 3 multiplayer where an FFA functions like several dozen permutations of 1v1s (players heal up to full hp in like 10 seconds in Halo). You also respawn meaning that the FFA last place is the dude that kills the least amount of people possible over a definite set interval. This is not how an RTS works. Especially one so honed in on the importance of scale like FA.

TrueSkill is not going to work in an rts FFA and you will need to entirely restructure how it reviews games to actually make it even remotely feasible for FFA.

And yes, I literally mean several 1v1s. If you’re 2k and lose to 7 300s walking into your base to com bomb you then as far as TrueSkill is concerned, you just lost 7 1v1s to 7 different 300s. Assuming that you are the first one to die.

@arma473 said in Noob matchmaker queue:

@ftxcommando Just because a person plays an unrated match doesn't mean they want to get rid of global rating. From what I have seen, the vast majority of unrated matches (with humans vs. humans) are balanced based on people's rating. Even for games involving AI, like survival, the rating is used by the host as information for what the settings should be.

"Rating removal is predicated on changing the intended use of custom games on FAF and is part of a larger scale motion to shift cultural attitudes. If you make custom games an area for fun script maps and general casual things like ffa or sim mods, it no longer necessitates a rating system."

Why do we need to "shift cultural attitudes" away from people playing dual gap or astro? Let the people play.

The first issue isn’t really an issue as it could be solved by some form of showing a player’s ratings in other game modes. If they only play unranked custom games and are 0 then they would be 0 in both of these situations so it’s irrelevant.

There is more to FAF cultural attitude than just playing dual gap or astro. What about the absurdly anal balance farming where people refuse to launch sub 92.279% games. What about the inability to “take a risk” and play with someone that has high sigma. What about the variety of expectations that every subcommunity has on what X rating should play like which creates massive aversion to any transitions between said communities.

I see, so Halo FFA is more like a lot of 1vs1 in a row.
I don't think ranking FFA would be useful for beginners anyway, it's more of a novelty game mode in RTS games.

I think that the problem with global rating is that you can not rely on it. Last seton's game I played with ratings all over the place (from 500 to 1700) my air player was 1100, where as the other air player was 1000. I had to take down the first three T3 strategic bombers myself as beach player.

As Tagada mentioned - you have to balance depending on what you know from people. The rating is at times irrelevant. Whether or not it should be removed or hidden is another story - but I personally don't use it to gauge my allies / opponents.

And that is why I don't want to see the ladder maps and / or rating changed. It is one of the most reliable tools to determine whether someone understands the game. Soon there will be more queues (3v3 / 4v4, woop!). A casual queue that has been suggested for alternative game modes like phantom or a survival is already being worked on content-wise by biass, madmax and myself.

And to those that say they have to wait hours on hours in a lobby: you're approach is odd. You can find people by looking at who played map X or Y by checking the replays. Then socialize on discord by just sending them a PM and check if they're up for a game. If they're not, discuss if / when they would be. Maybe you'll meet more people through them. It takes a few minutes, but that way you get to know the community more and you can study / work / do exercises until you're at the agreed time. It is not perfect, but a whole lot better than waiting in a lobby all day.

A work of art is never finished, merely abandoned

Jip, I was saying if you out up a type of game that's proposed in the casual, it will not fill during the time I'm available. Several other game types will fill up

What I'm getting from this discussion is that what the community ultimately wants is not just MORE players, but more QUALITY players. This is a funny game, in the sense that I regularly see rather highly rated players completely beating themselves up in the chat over how bad they are. There's a unique level of acknowledgement of how difficult this game is, but also how enjoyable it is, even when most of us aren't that good at it. And so everyone wants everyone to get better, so that we can all get more from this game. So that's a good place to start.

So my own views on the issue, as someone who's sorta new to FAF (tried playing it over the years but got completely humiliated and ran crying back to vanilla SupCom skirmish vs CPU where I could satisfy my addiction to the pause button in private).

First off, the attempts to educate players need to improve. It's easy enough to find tutorial content, but it's hard to find the RIGHT tutorial for a given skill level and skill deficiency. This is a common problem that exists also in music education: the best guitarist in town is not necessarily the best person to teach you how to play guitar. I find the tutorials in the FAF application to be useless to someone of my skill level. "Build a Land Factory, 2 pgens, mex, pgen, mex, pgen, mex, mex" or whatever it is. OK, simple enough, you're scaling your power production to meet your needs. Then it gives you this list of units to build from your land factory that a new player would have no chance of memorizing or understanding the logic behind. It's just way too much for a "first tutorial" and it feels to me like it was created by a player with very advanced skills (best guitarist in town) that just assumes that everybody understands very basic things.

And so these discussions about a "noobie ladder" and the Stoneage mod and other attempts to make better matchmaking are valid, but I think that if what the higher-ranked members of the community want is more skilled players to go up against, they really need to invest more time and energy into education, and I think the first thing they need to do is find someone who's good at making tutorial content, because that person is rarely going to be one of the better players (those who can't, teach, right?).

Beyond that, I think the Stoneage mod is a good concept that could be expanded on. It immediately made me think of promotion and relegation in soccer. Maybe there needs to be ladders for every tech level, and maybe you need to play yourself into a promotion before you can play ladder games at higher tech levels. Maybe the current 1v1 rankings can be used to initially slot players into different "leagues", but after that it should just be about who you beat and who beat you. If player A is demolishing everybody in the Tech 2 league and player B is getting walked on by everybody in the Tech 3 league, it doesn't matter what their rankings are: player A deserves the opportunity to go up, and player B needs to go down to make room. Something to think about.

The problem is you cannot divide the game like that. If you lock the game to pure t1 stage you have essentially turned uef v aeon into a complete autoloss situation. If you lock t3 stage how in the world do half the factions deal with uef t2 stage once they really get going with pillar/shield/arty?

You simplify the game but it’s to the detriment of actually learning the game, instead you just learn what faction has an advantage and accept the fact you have been restricted from half your tools to respond to it. Imagine getting a navy map and you have Aeon frigates against Cybran frigates AND you can’t even use torps. The game is just boned.

We have a couple guides written out to set people on the right path:
Blackheart’s guide is intended for you to get in the right mindset on what you need to do to see improvement

https://forum.faforever.com/topic/1222/how-to-improve-forever-6-laws

My guide here is meant to teach you the macro relationships in the game which allow you to scale up properly. This is often all you need to focus on to get to around 800-1000.

https://docs.google.com/document/d/13S4nBDfcBK4WmFtykXGKNmvIPe9L2nbiriISpHNgE4U/edit?usp=sharing

arma here wrote a guide focused on more of the gameplay nook and cranny elements that can help you slowly improve from a rating level to another rating level once you fix the errors.
https://forum.faforever.com/topic/766/ladder-1v1-beginner-intermediate-and-advanced-topics-by-arma473

Generally there are two more guides I want to write. I want to write a guide on engie managements to explain how you can decide where to send engies on a map and I want to write a guide on tempo in FAF to explain how to zoom out and gauge the game situation to then decide on a strategy.

There were also plans to write guides that explain how to expect to play maps that go into the <500 pool on ladder but it kind of lost steam.

What about having these guides linked to in the Tutorials section of the client? (in addition to the current build order tutorials and link to heaven videos)?

Don’t have a problem with it but would need to talk to some devs about it.

I guess that's my point - there's no lack of tutorial content out there, but if FAF is to be the central nexus (pun not intended) for this game, then it needs to do a better job of sorting out what the best tutorials are and what order they should be consumed in. Just like how a university doesn't give you any information that you couldn't theoretically find elsewhere, but it structures it in a way that increases the chance of actually being able to retain that information, because you've already learned the things you need to be able to understand what you're learning.

And again, I think because of how difficult this game is, you really need to find somebody who is particularly qualified in the field of education. I will take a look at your tutorials and I'm sure there's a lot to learn, but my experience has been that I usually retain about 5% of a given tutorial because 95% of it is all way over my head, and then I'll learn a bit more and can maybe take away another 5% on rewatch. That way of learning can be incredibly hard to keep track of and can potentially tire out new players.

On the subject of faction differences at different tech levels - I just assumed that the attempts to balance factions were being done with respect to tech levels. So you're saying factions aren't balanced until we're able to build experimentals?

Factions have advantages at certain spheres at certain tech levels. Cybran has the unquestionably best t1 frig but considered to be the worst cruiser and the 2nd worst destroyer. Aeon has the worst frigate but the best destro.

If every faction had parity at tech levels there would be no significant pressure on players to tech up and progress the game since why would you want to risk the 2k mass investment into new tech if it doesn’t improve your situation much at all. Enemy could just react to your tech up and you would be at the exact same situation as before except now you lost map control for your earlier tech up and likely have a harder time pushing in to recover it.