[Relic][Dev Diary] CoH-Development
Posts: 1158
Posts: 1382
The point of the posts is that trying to help the coh community is wasted effort. Both because the community are simps and because Relic doesnt care to help you. So I have moved on. It is a cautionary tale to others who are currently wasting their efforts helping Relic. You will be sadly let down at some point. It happens to everyone who works with Relic.
I am currently looking into having my account removed here if possible.
It seems to me like you're looking for an excuse for someone to make that decision for you. Hey man, it's okay to move forward in life. I'm sure all the people you've helped are very thankful for it, and plenty of people enjoy your maps.
If you would rather play guitar and improve your skills with that than play CoH2, make maps, etc. then do it! You don't need other people on this board to give you approval for it. Just leave it behind if that's how you really feel.
EDIT: There are plenty of accounts here with 1,000+ posts that simply stopped posting and never came back. Sad to see for a newcomer like me but what can you do.
Posts: 317
Classic Relic. We dont have 30 minutes to fix the garbage match making on our server but lets waste a week making a video about how cool our stuff is for the lemmings ready to spend money on our broken trash.
Very sad to see this. Just ridiculous. They should be ashamed.
....uMMMM Then why are you here? Go away I dont want negative whiny Haters Ruining my Coh3 experience.
It would be nice if we could all work together Community, Devs, Players etc all be positive and insightful disagree or debate with respect and manners as well as lose or win with dignity to make this the greatest community and game out there period.....LETS DO IT!
Posts: 317
I dont think you had to preface your statements here. Your comments do not seem that harsh to me.
The game is not in a playable state, that is the point. It has not been in a playable state for a long time (Forever?).
Since you guys have trouble with reading comprehension (most do not speak english as your first language), I will continue to try and help you.
* The matchmaking on the server is broken.
* Has been forever.
* Relic could fix it easily in 30 minutes.
* But instead they spend 100's of hours jerking themselves off about Coh3. Because they do not care about YOU, the coh2 player. YOUR experience in coh2 means nothing to them. Just like DOW3, Relic has abandoned the game. But the coh2 community are a bunch of little simps that keep licking Relics balls no matter how many times they shit on your chest.
* I care about you and your playing experience 1000 times more than Relic does.
Was that clear enough for everyone?
Should I rephrase it again?
If you understand what I wrote, please post a simple "YES" here so I can leave this site forever.
BLACK ADMINS: if 10 people post yes here, please delete my account from Coh2.org.
Hmmm If the game is so unplayable why do I and many other players as some have mentioned here have over 3000h hrs????
Posts: 978
Posts: 2148 | Subs: 2
....uMMMM Then why are you here? Go away I dont want negative whiny Haters Ruining my Coh3 experience.
It seems to me like you're looking for an excuse for someone to make that decision for you.
Don, I only come here and poke fun at Relic for you. I could care less if relic ever fixes anything. My point of the 10 people posting was I am here to help you. But hearing the truth bothers people who have embraced the lies (Trumpers, Anti-vaxxers, Flat Earthers etc). If so many players are just annoyed by my posts, I will leave. I am not here to piss players off, just Relic.
Here is a "balanced" game from last night?
Hmmm If the game is so unplayable why do I and many other players as some have mentioned here have over 3000h hrs????
From the picture above I would have to say you have so many hours because you have mental problems that I cant help you with. You may wish to seek professional help. Or like most of us here you find the overall game of Coh2 to be the best RTS out there, and want to turn a blind eye to the fact Relic is screwing you hard by being lazy and incompetent. You want to live in the dream world Neo.
Once again our lord and savior PageP comes thru to show us the light. If only Relic were 1/10th as competent as he is.
Here is a chart showing the length of matches on City 17. As you can see out of approximately one months worth of matches 81 lasted less than 5 minutes. Presumably because the game is still causing bug splats for people. So thats 81x8=648 people who did not have a great time playing Coh2. For more populare maps this number is closer to 140 games or 140x8=1120 screwed over players.
Most of the 4v4 maps have very similar stats, this being the worst. You would think that since there are more players for each match there will be more crashes. However, the 3v3 matches do not have this many 5 minute games.
So we have four options:
- 4v4 players are mouth breathing donkey apes (Duffman™).
- All 4v4 maps suck and crash peoples computers. Even though they are fine when only played as 3v3.
- There is a huge disparity in the quality of 4v4 maps. So when people load into some 4v4 maps they /L out immediately since there are too many bad maps to Veto.
- More people are running CELO and seeing poorly matched the games are and ALT-F4 during load.
Out of those 4 options 2 can easily be fixed by Relic with minimal effort. Will they?
Posts: 1295 | Subs: 1
Posts: 3114 | Subs: 2
- More people are running CELO and seeing poorly matched the games are and ALT-F4 during load.
Out of those 4 options 2 can easily be fixed by Relic with minimal effort. Will they?
Can Relic though?
In October, 4v4 made up about a third of the playerbase, judging by the amount of games on PageP's site when weighting for the player count.
CoH2 also has a peak of about 7,5-8k players during the week and ~10k on weekends. Let's assume about 9k players on average (which is already high), giving us 3k players for 4v4.
Average play time per map is ~25 minutes. With an assumed 5 min search time about a sixth of all players (600) are available for the server for balancing games, spread across Axis and Allies of about 300 each.
The ladders in 4v4 randoms span about 10k ranks, which gives us an average distance of 33 ranks per player. Best case scenario: Players are at least 133 ranks apart within their own side.
Now, factor in we've only compared peak players. Play 2 hours too early or late and you're already down 1k players.
We've assumed all players are in multiplayer and do nothing else than going from game to game. No comp stomps, no custom games, no breaks with the game running in the background, no single player, no arranged teams that might require special conditions.
Every player should find a game within a reasonable time. It makes sense for the algorithm to widen the search range the longer it searches, leading to suboptimal match ups.
And obviously normal fluctuations. What if your bracket is underpopulated, if the players near your elo are by pure chance not searching, players not evenly spaced etc.
I don't want to defend those "top 50 gets matches with rank >5000" or something. The algorithm is has a lot of shitty sides and needs optimization. But even under optimal conditions, I'd assume that a rank difference of at least 400-500 within your own team is to be expected, plus whatever is available for the opposite team. If you play at a slightly suboptimal time, I don't think there is much you can do against rank differences of 1000-ish.
Posts: 2148 | Subs: 2
Can Relic though?
.
.
.
If you play at a slightly suboptimal time, I don't think there is much you can do against rank differences of 1000-ish.
I love your thinking here and I agree totally. Perfect matches are impossible. Thank you for taking the time to help the community.
The main issues with the MM are:
- ELO imbalanced.
- Premade teams needing to smash noobs before they get a rank.
ELO BALANCE
For ELO balance I like to think in terms of percentage of that factions ladder. That is why it is a thing in MakoCELO. In your example you are saying generically we would have a range for each team of approx 1k rank for a faction of 10k players (10% ladder range). What is a good cutoff for range?
I would like to see Relic implement some kind of new cutoffs.
- No match found if the range is too large (30%-40% maybe).
- No match found if all of one team is greater than the other teams best player. I constantly see games where team A is all top 15% and team B is all 50% with one player at 20%.
- No match found if ELO averages differ by more than 20%. We dont want three monkeys with one man vs four cro-magnons (see Aerafields MM image).
There are also some harder to fix issues like having one team all be a single faction against a mixed faction team. This is especially hard in 4v4 if the team is all OKW. But you end up with a team missing some valuable tools to deal with the well equipped enemy. Could maybe reduce the ELO average by 5% for single faction groupings. 2v2 and 3v3 are probably fine as is. 2v2 has less players and the 3v3 maps are usually large enough that it is almost a 1v1.
PREMADE TEAMS
Balancing this would take a little more code to get good. But something needs to be done to push premades to a higher ELO. A brand new team should not be considered as NO RANK.
The best case scenario would be to search each players best ranks in the closest matching game mode. For example a player is rank 15% in 3v3 OKW but 45% in other factions, that player is ranked at 15% to start. Even though the current game mode is 4v4. This will always be a problem since many team players actually have no ranks at all listed. But they have played some games so Relic must have their "hidden" ELO value stored somewhere.
Since premade teams are so much more effective, they should have their ELO pushed a little higher. For example a team of 4 players should be pushed 10% higher in ELO than their calculated value. Maybe go 5% for 3v3 and 0% for 2v2.
Best case scenario would be Relic gives an option for you to select a range of ELO when you start searching games (Duffman™). Since this would require the game EXE to be changed, this is too much to ask for Coh2 but could be implemented in Coh3.
Posts: 3114 | Subs: 2
...
I pretty much agree to all of this.
Different genre, but Rocket League manages to balance arranged teams and randoms pretty well. In such matchups, I can usually see that the enemy is better than I am individually, but obviously fails at quick communication and the general advantage to know the playstyle of your mates. Their big advantage is that there are fewer variables and a match only takes about 6 min with a huge player base to draw from.
For CoH2, I think only Relic can fix this. They need to have a deep look into their data and correlate player elo with victories and VP count at the end of the game. Only Relic has the data to do so, even if there are thousands of variables. They really do not have an easy job.
I would be a bit more careful with percentage of ladder ranks. I assume that there is a larger difference in skill between #1-1000 than #9000-10000. I can't tell what the perfect window is. Does it even make sense? Players at the edges of the ladder might wait even longer for matches. What if there are not enough players? Or the players that are available would fit better into another match than into yours? Shall we screw over a couple of players to get more even matches in general? Otherwise the disparity might propagate for multiple matchups. In the end the players waiting too long will stop playing, which just moves the problem to the next set of players.
Similar with elo selection: Relic would need to male all data public. I don't know where I should set my range to if I don't know how many players are in this range. Should I even care? I actually should not need to with a half decent matchmaker. On top of that, it could create a lot of problems if half the players are not able to be matches because they set their elo ranges too narrow or so.
Well, enough for now.
Posts: 2148 | Subs: 2
I pretty much agree to all of this.
For clarity:
PERCENTAGES
I like to see percentages of rank because each faction has a different amount of players. 4v4 for instance USF usually has about 8k players but OST has about 14k players.
So a rank 7K USF player (87.5%(7/8)) is no where near as good as a 7K OST player (50%(7/14)).
The leaderboard LEVEL is a similar value to percentage but it breaks at rank 200 (LEVEL 16) for low player count modes like 4v4 teams. You are either a level 13 or 16. Level 14 and 15 do not exist.
ENTER AN ELO VALUE
The Duffman™ idea was for new teams. If the game sees you have zero games it can ask you what rank or percentage you would like to start as. If you are an SoE team you may enter a value of 10% (meaning top 10% of this game mode). Or even make it like 4 buttons Noob, Ez, Good, or God Mode. So you can get close to a good game much faster than smashing 5 noob teams before the server decides to give you a higher ranked game.
Of course it could also verify this value against your ranks in other modes so you dont have noobs trying to play against SoE.
Posts: 578
The game client should match populations of premade to premade first, then filter them on ELO within that result to find a match.
It should not then start hoovering randoms in when no suitable premade matches are found, because this then makes the match premade vs random.
It should keep searching instead until another premade team is found.
For bigger game modes, e.g. 4v4 when 2 players are a premade within team1: match with another 2-player premade within opponent's team first, then backfill with randoms.
The reason ELO rank sucks for matchmaking are numerous in coh2:
1. people use drophacks regularly to change their standings
2. lag and distance from the servers reduces ones ELO artificially
3. premades smurf to get easy wins
4. the quality of randoms you get are diverse, yet nearly always terrible because of how broad the ELO search is
5. the game client can crash, people can quit during load etc, nobody wants these errors as any part of their matchmaking ELO experience (I'm not sure if the game writes any ELO data changes in these scenarios but it really shouldn't because the players didn't actually get to play)
6. the ladder system doesn't make sense nor provide any meaningful way to evaluate someone's skill because the quality of data its using to put people at their rank is bad
TLDR, the ELO method is gross in any game where you can get premades vs randoms. These are 2 different game modes and the client as well as the matchmaking service need to treat them as separate cases entirely.
Posts: 964 | Subs: 11
Just like DOW3, Relic has abandoned the game. But the coh2 community are a bunch of little simps that keep licking Relics balls no matter how many times they shit on your chest.
As a pro-CoH2 messenger that is staying vigilant, your job is not to save everyone.
There is none as blind as those who refuse to see (willful ignorance).
Relic/Sega leadership presents no evidence for the CoH2 abandonment. Until Relic/Sega present the evidence the abandonment is not based on reason and logic.
CoH2 is not DoW3. If Sega/Relic wanted to they could easily make more money from CoH2 content. CoH3 remains a high risk investment. If Sega leaders had any logical brains they should have used AoE4 for Relic to prove trust. But fuck no, they take another gamble with CoH3.
So yeah, if AoE4 is the best Relic can do, then CoH3 will be a bumpy ride with shitty post launch support..
Relic is now 3x the size of when they made CoH2. Are Sega fools into thinking Relic profit will go up by running a large studio with average leadership (at best)?
As an investor, seeing a critically thinking impaired Sega/Relic leadership throwing money into a black hole is like;
The gaming community has plenty of informed people who has been trying to wake people up to the major threats that is facing the industry. There is zero excuse for not being informed.
Livestreams
16 | |||||
16 | |||||
2 | |||||
103 | |||||
29 | |||||
15 | |||||
8 | |||||
8 | |||||
1 | |||||
1 |
Ladders Top 10
-
#Steam AliasWL%Streak
- 1.655231.739+15
- 2.842223.791+5
- 3.939410.696+5
- 4.35459.857-1
- 5.599234.719+7
- 6.278108.720+29
- 7.307114.729+3
- 8.645.928+5
- 9.10629.785+7
- 10.527.881+18
Replay highlight
- cblanco ★
- 보드카 중대
- VonManteuffel
- Heartless Jäger
Board Info
aerafield
9 posts in the last week
27 posts in the last month
Welcome our newest member, swimmingpoolsofflori
Most online: 2043 users on 29 Oct 2023, 01:04 AM