Why FAF competitive scene is dying

[Petricpwnz]

It's FAF. Solutions that don't require coding will always be preferable to solutions that require coding at the short term. Until a dev expresses any interest at the ability to introduce some sort of tagging system for news, all we can do is create some sort of biweekly newsletter that provides people with news in broad, clear categories that they can read at their leisure.

The idea is to have a sort of system where people can post their own individual news during the week, however, the weekend would be reserved for an official newsletter/news from FAF itself.

In regards to the replay idea, dunno, I just think it's redundant. Like, all we have for high-level replay analysis is essentially Jagged. Sure, other people have casted a few games (blodir, mephi, thomas) but they're irregular or dead channels now. Personally, I take it to mean that there isn't much of a point to appeal to this market. Especially when it will require like what, a team of 5-6 dudes regularly going thru tons of replays to skim which ones are worth watching?

We have tournaments to get a concentration of high level games to watch and play in, and we have casters that can introduce people to these games in either a casual or analytical style. Are we trying to appeal to the new player that has somehow squeezed out all the juice in jagged/ta/heaven/whatever videos and has no interest/idea to watch streams of high level games and is also too dumb to figure out how to search for 1800+ in the replay vault? It's just an idea that has crossed the spoonfeeding bridge to the point where I'm dumping more effort in a week than I'll see the idea give in output during its entire existence.

Im not saying your solution is bad i just want to make sure the news category thing is something devs wouldve at least heard of. Can't code something you dont even think about.

As for the replay idea i dont think its redundant, and there are already quite a few people going through most of the top 1v1 replays, just ask them to drop a replay id if they find something cool. If you dont want that then can pull a replay automatically after sorting by highest rated recent ladder game with regard for length (ideally weighted for rating and length with the map size in mind). I dont get your appeal argument at all, the point isnt to appeal to a wide audience, the point is to catch attention and sieve out those few tryhards who might actually climb up to 1800+ in this dark day and age of 2020. And yes, ive been on faf for a good amount of time and the average faf coombrain is indeed incapable of filtering replays by 1800+, this isnt helped by the clunky replay search ui of the new official client. Hell i wouldnt be surprised to hear half the faf population doesnt even know replay vault exists. As for whether its worth the effort, a feature like this is a very small side project in comparison to matchmaker or ice or bulky api/client work, if nobody in the client gang wants to do it its fine, but again dont see why its not something they should know is on the table, especially the low effort automatic version as that would skip the work of adding an interface of putting up the replay up there manually.

[ThomasHiatt]

There are quite a lot of people who don't know about replays or how to access them. The client hides replays behind some obscure tab that is just labeled "Vault". There should definitely be a clearly labeled tab for replays and more things guiding players into checking them out.

[FtXCommando]

There are quite a lot of people that do not know about replays, but the overlap between this group and the group that enjoy watching high level analytic casts/streams and want to learn the game will be extremely low. If you automate the functionality for this feature, it just becomes the sort of thing that can be done through a quick skim of 1800+ replays. I really don’t understand what it provides.

If you want to argue about the replay value being more presentable, sure, I can get behind that. But the tools to look for top replays are there, if anything all that would be needed is a quick “top replays” button that immediately parses for 1800+ replays.

[armacham01]

"if anything all that would be needed is a quick “top replays” button that immediately parses for 1800+ replays."

I happen to have made a custom search along those lines. I've been using it for a few days. I paste it in to the search bar:

(featuredMod.technicalName=="*ladder*",name=="*1v1*",name=="*2v2*");playerStats.player.ladder1v1Rating.rating=gt="1800"

The problem is not being able to filter for them. It's that there are only about seven 1800+ ladder matches per week. I think people got burnt out from LotS, because before people started practicing for that, there would be many high-level ladder matches every day.

There is already a section in the replay vault for "Highest Rated" replays of all time, but that is not useful at all. It is stale and a lot of the replays are old. I would propose changing it as follows:

Every time the user starts up the client, it picks a random number "M" between 1 and 12. It will only show high-rated replays from M months before the present. Also, every time the user starts up the client, it picks a random number "X" between 0 and 9. It will only show replays where the final digit of the replay ends in "X." Then, it shows the replays (that are not filtered out) in descending order of the average number of stars they got. This way, if you open the client five times, you will get five different lists of high-quality replays, and none of them will be suuuuuper old. We will avoid having a stale list. Rosh v. Lockmad was not a bad game, but keeping it permanently at the top of the list is NOT delivering value to FAF users.

AND, if this means that it is useful to the community to rate interesting replays, maybe more players will give good rating to games that are worth watching, so the fun games will actually make it into the list. Yes, there will be some spam. Some 500-rated idiots playing Astro Craters will give their games 5 stars because they want attention or something. We could try to correct for this by downrating boring games, or by having a minimum average rating for the game in order to make it into the list of high-quality replays. In fact, that could be another random number. Let's say between 1000 and 1800, is the minimum rating to appear in the list. So again, sometimes when you load it you will only see replays from the highest-rated players, and other times you will get a broader range, but you shouldn't have any replays from very new players, even if they give themselves five stars.

[Petricpwnz]

There are quite a lot of people that do not know about replays, but the overlap between this group and the group that enjoy watching high level analytic casts/streams and want to learn the game will be extremely low. If you automate the functionality for this feature, it just becomes the sort of thing that can be done through a quick skim of 1800+ replays. I really don’t understand what it provides.

If you want to argue about the replay value being more presentable, sure, I can get behind that. But the tools to look for top replays are there, if anything all that would be needed is a quick “top replays” button that immediately parses for 1800+ replays.

Stop assuming every person on faf has automatic knowledge of existence of every caster including analytic ones and fancy watching them. Just not the case. I found faf through a deliberate youtube search that pointed me to TA video of IMBA cup. Had I found it via a friend recommendation I wouldnt even know there is some competitive scene with tourneys or that its any interesting and fun, the entire games page and replay vault are filled with mediocrity. Well actually i'd just turn around at the steam link step but lets ignore that. And nowadays even if you find casts its likely residentsleeper casual teamgames.

Top replays button would surely be nice, if people couldnt figure out replay vault in py client they certainly cant in java. But its just a worse idea, its perhaps good for teaching people that you can search for rating etc, but can just do an easy change of pre-autofilling the search query with filter by 1800+ or whatever in java client. A dedicated slot for 1 replay gives nice focused attention, and without any need to click buttons, if someone is gonna code something this + autofill sounds much better to me.

[Sir Prize]

Ladder pool. Current pool leaves many people not satisfied, any pool will leave many people unsatisfied... or will it? I have already written a nice (salty) post describing the perfect map pool designed to satisfy absolute most people viewtopic.php?f=2&t=15110&start=10#p153945 and I would really like to hear the inherent flaws it has or otherwise I'm puzzled as to why there is no movement, being it even the slowest and slightest of them, done towards this map pool. Even if the chance of someone volunteering to work on it is small, if they have no idea the issue exists no progress can be made. I would suggest bringing this up in the next dev call.

- Ladder already has been changed, likely won't be made any better until we get some work done on it alongside matchmaker implementation.

Has ladder changed since the start of this thread, beyond the size of the pool and increased rotations? I was afk, genuine question. You implied more changes would be made to the pool in an ideal world, so what is the medium/long-term vision for ladder? Why is it dependent on work for "matchmaker implementation"? I assume that's team matchmaker (f*** I want that to happen, I literally made this forum account in 2014 to ask for it - NINJA EDIT: Wooo they just asked for testers!). I have a vague memory that the Zep pool suggested in the OP actually can't just be pasted back in for some reason, like the old code would no longer work... is that right?

Spoiler: show

I'm not a top player so I'm going to put my opinion in a spoiler, don't wanna derail this into another ladder pool thread when people were already having so much fun raging talking about replays.

I am pretty much exclusively a 1v1 player but often prefer to play custom 1v1s or tournaments than play ladder, pretty much entirely because of the post-Zep pools, which in turn makes it less likely I'll ever be relevant for this main discussion - arguably I should just HTFU if I really want to "make it" though. Regardless, FAF's map pool is bigger than most RTS ladders (GPGnet, Starcraft for sure) and rotates MUCH faster than any other RTS I'm aware of, which effectively makes it even bigger. I would search more if I didn't feel like half the challenge was learning or re-learning the map, possibly against someone who already has done so - there's a big difference between your first and third attempt at a map and most of them cycle out after a fortnight. I know that people who do spam ladder get bored with the small, static map pools and mappers want a realistic chance of their map being used, so imo the best compromise is probably the return of the Zep pool as Petric said in the OP and in the thread he linked.

Ladder rewards. This was done with the division system, while it wasn't very relevant for top players it certainly had a positive effect on ladder in general. Checking the statistical data each 3 month "season" and giving out some avatars based on activity is a good start.

- Already exists in a minor fashion for leagues. Divisions are their own project.

What do you mean by "divisions are their own project"? Seems to me like ladder league months are what Petric was talking about and are already implemented. Anyway, I think it would be really positive if the leagues weren't so minor and so hidden from non-forum goers. This: viewtopic.php?f=26&t=18042&start=20#p181598 could easily go in the News section of the client and/or the QAI spam for example.

Honestly, every month should be a ladder league month, or if it's easy to do have it calculated based on the last 30 days on a rolling basis. If the leagues were always running the various leaderboard links and info displayed in chat could more prominently feature league and rank over 1v1 rating (which obviously should still be used for the actual matchmaking and shouldn't be hidden) and more in-client and website news coverage of the results could be added. CnCnet are a similar size and a similar project to FAF, here's how their ladder looks: https://ladder.cncnet.org/ladder - which highlights the best active ladder players and sort of gives a possible demonstration of the idea Petric/others are talking about in highlighting replays too. Might be nice if 1-5 ladder matches that went for more than 10 minutes in the last week with the highest aggregate ratings between the two players is automagically in the News section with a pic of the map that is also a download link?

[FtXCommando]

Has ladder changed since the start of this thread, beyond the size of the pool and increased rotations? I was afk, genuine question.

We've added no repetition code that restricts players from getting one of their last 3 maps again.

We added a queue system which, for now, mainly functions as a way to create more balanced games by creating matches every 3-5 minutes rather than as soon as someone searches that meets a criteria. Later on, this queue system is meant to allow easy transition between ladder and the various matchmakers while not dividing the pool of players. It will do this by giving people the opportunity to decide what they want to queue into (for example, I want to play a 1v1 and a 2v2 but not a 4v4) and will sort out games based on these preferences.

We added code that has the games of new players on ladder weighted to playing against 800-1000 rated players for their first game. As this is the median on ladder, it is the most efficient way to classify a player as above or below average.

You implied more changes would be made to the pool in an ideal world, so what is the medium/long-term vision for ladder? Why is it dependent on work for "matchmaker implementation"?

Because we don't have the quantity of devs that can work on both matchmaker and ladder. I was able to talk to Geo about creating multiple pools (noob/joe/pro) and a probability based system for deciding what pool you get that is determined by rating. This should come out alongside the initial matchmaker release, but anything like, for example veto/choice pool systems will not happen for quite a long time. Something like veto/choice will require a ton of client ui work as well as a ton of client ui debugging and so it will add an immense amount of work. From what I can tell, this is probably one of the most time consuming things on FAF so anything that can be done to minimize it will likely expedite features. This is why we agreed upon just doing a multiple pool system as a way to artificially do a choice pool server-side, the problem is that I'm the middleman generally deciding what is good for each rating group.

Regardless, FAF's map pool is bigger than most RTS ladders (GPGnet, Starcraft for sure) and rotates MUCH faster than any other RTS I'm aware of, which effectively makes it even bigger. I would search more if I didn't feel like half the challenge was learning or re-learning the map, possibly against someone who already has done so - there's a big difference between your first and third attempt at a map and most of them cycle out after a fortnight.

Maps are an immense pillar of strategic variety on FAF. Honestly, even factional differences play second fiddle to the role of maps. I should probably do surveys to gauge feelings about what incremental change people would prefer. I CAN tell you that every map pool consists of a max of 3 new/experimental maps. Everything else tends to be more commonly enjoyed maps and the classic maps. If you were gone for 2 years or so, well you missed a lot. FAF went from probably having 1 new map functional for ladder every two months to having a new map someone wants in ladder every 3-4 days. You probably feel like a lot of the maps that have been made "common" maps over the last year are new maps, but they really aren't anymore.

I'd also like to emphasize that part of the function of ladder is to provide incentive for mappers to create interesting maps. Maps that have had effort put into them will be put into ladder which in turn (ideally) benefits both the player as well as the mapper and creates a better community overall.

What do you mean by "divisions are their own project"? Seems to me like ladder league months are what Petric was talking about and are already implemented. Anyway, I think it would be really positive if the leagues weren't so minor and so hidden from non-forum goers. This: viewtopic.php?f=26&t=18042&start=20#p181598 could easily go in the News section of the client and/or the QAI spam for example.

Divisions in modern FAF terminology refer to a system that would blackbox trueskill. It would eliminate the need for new players to worry about trueskill and equate it with rating and hopefully eliminate a lot of the toxicity revolving around the point winning/point losing based on a single match. Instead, it would operate like many other games where you could lose say 3 games but still be a Silver II or whatever.

The League system is just a way to reward ladder activity, give newer players a goal to work towards to, and ideally a way to notice people that are putting work into improving their gameplay. I'll ask Dog about including league info in a tournament news post that he's writing soon.

Honestly, every month should be a ladder league month, or if it's easy to do have it calculated based on the last 30 days on a rolling basis. If the leagues were always running the various leaderboard links and info displayed in chat could more prominently feature league and rank over 1v1 rating (which obviously should still be used for the actual matchmaking and shouldn't be hidden) and more in-client and website news coverage of the results could be added. CnCnet are a similar size and a similar project to FAF, here's how their ladder looks: https://ladder.cncnet.org/ladder - which highlights the best active ladder players and sort of gives a possible demonstration of the idea Petric/others are talking about in highlighting replays too. Might be nice if 1-5 ladder matches that went for more than 10 minutes in the last week with the highest aggregate ratings between the two players is automagically in the News section with a pic of the map that is also a download link?

Well, a lot of what you're saying comes down to ui work. Which, as I said, is the worst thing to ask for with regards to getting features anytime soon. I fear burnout and monotony coming from a league system that is essentially always running. Kind of just a personal bias because I would hate putting a ton of effort into winning something and then immediately be subject to losing it the literal next second.

[Sir Prize]

Thanks for the thorough response, the division idea sounds... ambitious. I don't quite have it in my head what that would look like when implemented but not having people worry about rating so much would probably be a bonus. Is it something like what they're calling leagues here? https://liquipedia.net/starcraft2/Battle.net_Seasons

I fear burnout and monotony coming from a league system that is essentially always running. Kind of just a personal bias because I would hate putting a ton of effort into winning something and then immediately be subject to losing it the literal next second.

My personal bias is the opposite regarding having to actively defend your status as "the best" at something. Speaking from past experience I would make more effort to keep the very rare tournament avatars I held and had to defend in new tournaments (WWPC mostly) so I imagine if I got to the top of a league I would have the same response, I imagine others would respond similarly.

Interested in what you mean by burnout there too. Once it was setup, I'm not sure how you/devs would get burnt out? If you mean the players, it's pretty standard to have permanently active and meaningful ladder seasons in other RTS communities. Also can't quite get my head around why it's more effort to restart the ladder month on the 1st of every month, rather than every second month, but obviously I don't know your/the relevant dev's process.

Also I'm reading between the lines here, but is the big issue with returning to a Zep pool (with the added no-repeat code and the other useful stuff that has already been implemented) "just" be the fact that it would be a UI nightmare now the official client is not the python one the Zep pool was originally designed in?

[FtXCommando]

It would be a nightmare because it delays significantly more important features for another year or more. I don’t value a choice pool over tmm. The coding reasons for why it would be difficult relate to how the java client was built compared to the python client but I don’t want to get into specifics without possibly getting info wrong.

[Blodir]

Can we hide zock and keyser from leaderboards. They havent played since 2017