268 replies to this topic

[Blade Pride]

Matthew Craig, on 23 September 2013 - 09:15 AM, said:

Match making thresholds have been returned to pre-patch levels for now, we've learned a lot from this attempt to tune them. Thanks for everyone's feedback we'll be discussing a new set of changes internally aimed at improving the match maker further.



Tournament is a completely separate build on separate hardware so no chance of involvement with live.

Kattspya, on 23 September 2013 - 11:26 AM, said:

That might be making me sad.

After the latest patch I did not play until the following saturday and I found that my wait times were doubled or quadrupled. But the matches were the some of the best I've had since Elo was introduced. I think there might have been ten or so matches that could have gone either way up to the last few minutes and sometimes did. It was very nice.

^^ This.

After the patch wait times for me were 30-80 seconds longer but the matches were some of the best I have played. Yesterday was the worst I have ever seen, well back in January may have been a little worst but that was per-ELO. Any idea when you may try to re-tighten the ELO again?

[Matthew Craig]

BIix, on 23 September 2013 - 05:55 PM, said:

With all due respect, wasn't this what the Beta was for?

Yes it would have been better to do this during Beta, but even now systems like the match maker are very hard to simulate perfectly internally (though we continue to work on that) and even public test doesn't have the volume of players to show us the true picture so there is an element of doing our best due diligence and then seeing what happens on live. We waited until we did as mentioned for two reasons 12 vs. 12 needed to be in place and we needed to have enough telemetry to be comfortable we could measure the impact of the changes.

Our goal is to use the data gathered to feed into internal stress test scripts that can emulate live, and also to add even more telemetry to live, as much as possible we want to ensure that once a change goes live it is solid and doesn't need to change but this is sometimes easier said than done, we continue to work towards that goal though.

Blade Pride, on 24 September 2013 - 06:50 AM, said:

^^ This. After the patch wait times for me were 30-80 seconds longer but the matches were some of the best I have played. Yesterday was the worst I have ever seen, well back in January may have been a little worst but that was per-ELO. Any idea when you may try to re-tighten the ELO again?

Yeah sorry for those that were getting better matches, I think it's looking most likely that the next round of changes will line up with the release of UI 2.0 as we need to ensure all changes align with the new UI.

[GreyZ7]

Too bad it had to be rolled back, the wait times were long but I did think the matches were much better and more importantly it was nice seeing the same pilots in your ELO bracket more frequently :/

thanks though for quickly addressing the issue

Edited by kai hohiro, 24 September 2013 - 08:02 AM.

[Mister Blastman]

Even though things have been set back, I can't find a drop when trying to pug alone in my daily mid-day drops (I get up to four). I sat there for 7 minutes, launching twice. Nothing.

I suppose I was spared being teamed with pug hell, at least, in theory.

[Blade Pride]

Matthew Craig, on 24 September 2013 - 07:41 AM, said:

[/size]

Yeah sorry for those that were getting better matches, I think it's looking most likely that the next round of changes will line up with the release of UI 2.0 as we need to ensure all changes align with the new UI.

Awesome, thanks for the update.

[kaffeangst]

I tried to play this afternoon in a 4-man with myself, Villz, Ryan Steel and Wispsy. We searched for an hour and failed every time. Fix match-maker please.

[StainlessSR]

Matthew Craig, on 24 September 2013 - 07:41 AM, said:

[/size]
***SNIP***
Yeah sorry for those that were getting better matches, I think it's looking most likely that the next round of changes will line up with the release of UI 2.0 as we need to ensure all changes align with the new UI.

I am sorry to hear this. Over the last few days I have tracked how the winning team has fared.
20 games winning team had 0-4 loses
14 games winning team had 5-8 loses
5 games winning team had 9-11 loses

the last bracket is where games should be, and where the changes you made on patch put the vast majority of them at. Having to wait till NOV is going to be bad as the constant stomps are going to drive any new player away as they are not going to like how badley they get stomped (it looks like one side is all pro's and are just being fed fodder to inflate their score). I would love to have this game get bigger but it will require severe changes to the matchmaker to both allow high placed players to have a grouping and low placed players to have a grouping with middle players filling in the rest. ie. the top 1/3 get's it own matchmaker which starts filling in with them and goes to the middle 1/3rd to finish filling teams. the bottom 1/3 get's it own also which starts with mechs in the bottom 1/3 and finishes by filling in with the middle 1/3.

Until such a time as the server pop get so large that the two matchmakers are unable to take all available from the middle third this should work and each segment should not have too long of a wait. Once the pop get large enough to leave the middle 1/3 a long time waiting for a match then just start a third matchmaker thread that uses them to create a match. This is most likely an oversimplification of what needs to be done to implement it but it appears to be the most logical way to go about it, as it would keep the new players together with some salting of better players while the top 1% get players who are much closer to their skillset to match their skills with. It should also remove new players being smashed by high skilled players and bring the games closer to being evenly matched.

Edited by StainlessSR, 24 September 2013 - 04:03 PM.

[Chavette]

kaffeangst, on 24 September 2013 - 01:23 PM, said:

I tried to play this afternoon in a 4-man with myself, Villz, Ryan Steel and Wispsy. We searched for an hour and failed every time. Fix match-maker please.

Same, I was on too but trying to do it solo. Then it got fixed all of a sudden, and I'm getting all kind of matches.

[MavRCK]

Jman5, on 21 September 2013 - 10:10 AM, said:

Unfortunately it's a problem that many games with matchmakers have. Elo works great for 99%, but struggles at the extremes. Especially when a lot of the top players tend to split their team between 12 man and pugs.

/Looks around furtively

Now I'm not telling you to purposefully lose games to drop your Elo. All I'm saying is that if you happen to lose a lot of games where you were evenly matched or even supposed to win, it would allow you to find more games.

I'm just saying...

Edit: Alternatively and I'm just speaking out loud here, if someone on a 4 man were to create an alternate account it would average your team's Elo down significantly. But we're just talking hypothetically here right?

Or one could always buy lots of mediocre mechs and level them pugging by yourself...

But then you'll scream with frustration at the idiocy you'll see... and take twice as long to level those mechs.... Earn 1/3rd the cbills....

What the hell was I thinking!?!

kaffeangst, on 24 September 2013 - 01:23 PM, said:

I tried to play this afternoon in a 4-man with myself, Villz, Ryan Steel and Wispsy. We searched for an hour and failed every time. Fix match-maker please.

Stacked team bro.

Edited by MavRCK, 24 September 2013 - 06:00 PM.

[Kattspya]

Matthew Craig, on 24 September 2013 - 07:41 AM, said:

Yeah sorry for those that were getting better matches, I think it's looking most likely that the next round of changes will line up with the release of UI 2.0 as we need to ensure all changes align with the new UI.

Eh, that sounds like more than a UI. Why would the back end need its changes to be lined up with the new UI? That makes no sense.

[cooleagle]

This is f***ing ridiculous. Just had a game with 6 trial mechs on my team. The other team had ZERO. Needless to say we got absolutely stomped. I did 400 odd damage, 2 others 200 odd, rest under 100.

Tell me how this s**t is balanced?

Unfortunately, I can't get screenshot to work or I would be posting it.

Edited by cooleagle, 25 September 2013 - 12:20 AM.

[Inkarnus]

yeah noticed alot of badgames the last days and alot of onesided stomps vs me
should i feel now honored
still it gives me a bad feeling to play more till this get fixed
.....

[Matthew Craig]

If some of the suspected high Elo players/players that were having very long wait times are willing to get on teamspeak, I'd be happy to try tuning again with their feedback. So feel free to PM me if you think that's you and your willing to help, if I get enough response we can have another go at tuning up to the point we start to break things for high Elo players.

Otherwise we'll just have to wait until we have further telemetry added in.

[FREDtheDEAD]

I suspect one problem with matchmaker balance at the moment is the large number of champion trial Atlases present. Highest weight in the game but probably very low ELO pilots. A flood of champion assault trial mech players introduces difficult dynamics for the matchmaker to cope with.

[Imperius]

Me and my group feel ELO is giving us bad players or new to balance us out with a group of average players. Does the matchmaker do that?

[Kattspya]

Matthew Craig, on 24 September 2013 - 07:41 AM, said:

Yeah sorry for those that were getting better matches, I think it's looking most likely that the next round of changes will line up with the release of UI 2.0 as we need to ensure all changes align with the new UI.

I hope feature creep turned UI2 to into everything but the 3D engine and that it is sufficiently impressive if/when it arrives.

Otherwise I can think of no explanation but I'm not very knowledgeable when it comes to these things.

[Kattspya]

StainlessSR, on 24 September 2013 - 04:00 PM, said:

I am sorry to hear this. Over the last few days I have tracked how the winning team has fared.
20 games winning team had 0-4 loses
14 games winning team had 5-8 loses
5 games winning team had 9-11 loses

Are those true stats? As in did you accurately record every match?

If so it is quite the achievement compared to earlier attempts. Almost one half of matches had the winners losing half or more of their mechs. The drop times might have gone from high to unreasonable for some but the matches might have been very good.

EDIT: clarification

Edited by Kattspya, 25 September 2013 - 09:23 AM.

[ohtochooseaname]

I really enjoyed the much closer games when things were tighter. Now, most of the games I play late at night are complete blowouts, with a few close games scattered here and there. However, a close game is much more exhausting than a blowout game, so having all very close games may not actually be a good thing. Also, the closer a game is, the more cbills/xp you get from sympathetic rewards like savior kills. Personally, I think things could be tightened a little further, but I do enjoy the game in the state it is in now.

[Chronojam]

Matthew Craig, on 25 September 2013 - 08:25 AM, said:

If some of the suspected high Elo players/players that were having very long wait times are willing to get on teamspeak, I'd be happy to try tuning again with their feedback. So feel free to PM me if you think that's you and your willing to help, if I get enough response we can have another go at tuning up to the point we start to break things for high Elo players.

Otherwise we'll just have to wait until we have further telemetry added in.

This is the right attitude, but it's a shame to have it so late. Would have had plenty of absurd Elo players for you months ago who are gone now Maybe they'll be back once you get this pinned down.

[StainlessSR]

Kattspya, on 25 September 2013 - 09:20 AM, said:

Are those true stats? As in did you accurately record every match?

If so it is quite the achievement compared to earlier attempts. Almost one half of matches had the winners losing half or more of their mechs. The drop times might have gone from high to unreasonable for some but the matches might have been very good.

EDIT: clarification

Yes these were all the games I played, it may seem to be a small amount but you have to remember no matter when I got killed I had to sit out the whole round to get the end results. The results include both groupings up to 4 players and solo drops. I read a post where the player used a much smaller variable set so my next few days off I may retry this with stats for solo and group separated as they may be skewing the results.

In my personal opinion I prefer the games with 9-11loses on winning team as it makes for a more enjoyable game for myself (even if I am just shoulder surfing)

EDIT clarification

Edited by StainlessSR, 25 September 2013 - 11:16 AM.