268 replies to this topic

[Woozle]

ohtochooseaname, on 19 September 2013 - 09:54 AM, said:

...when I had much more difficulty finding a match, I was able to get into some matches by grouping with a fairly average player to drop the combined ELO down.

Sweet! As a thoroughly average player, I hereby proffer my services to High Elo players as a "foot in the door." With me being lodged firmly in the meaty middle section of the Elo bell-curve, by pairing with me your wait times will vanish! I won't even charge for this valuable service! ;-)

[RF Greywolf]

Woozle, on 19 September 2013 - 10:07 AM, said:

Sweet! As a thoroughly average player, I hereby proffer my services to High Elo players as a "foot in the door." With me being lodged firmly in the meaty middle section of the Elo bell-curve, by pairing with me your wait times will vanish! I won't even charge for this valuable service! ;-)

WOOT! Us mid elo average players now have a reason to live. When I started playing last night I had some long wait times but it got better as the night went on. The tweaking is helping, just a matter of getting the right balance.

[Matthew Craig]

Thresholds have been loosened a bit more please continue to provide your feedback especially players that were having issues with long waits.

[retakrew7]

So here is how I am affected by the matchmaking:

Right now it seems more balanced, I believe my elo is not ultra high but definitley above average with all mech weights.

There was an improvement with the latest adjustments after one patch impacted it negatively.

When I am launching alone I nearly never have problems finding a match, the matches tend to be good also.

However, most problems occurred when I was playing in a group of 2 or 3 with either Bruro, Redbull102, Fulko or MonPax, then our team contained a lot of new players in trial mechs.

When we are launching in a full group of 4 I get the best matches with both teams playing well.

My suggestion would be as follows:

3 basic elo ranges:

- from 0 to sligthly below average
- slightly above average to infinite
- one middle group which uses average and "leftover" players from the other 2 groups

first matchmake the low elo and high elo groups, then matchmake the leftovers and the average players

that way you make sure that the groups with low population have higher priority and surely find a match, the average elo players shouldn't have a problem finding matches at all

[Vernon Jettlund]

I have an idea.

I know, famous last words.

So, if the ELO is a bell curve, and the issue is the numbers across the curve (defines the curve), the MM could attempt to build teams from ELOs from a given range of the curve. This would require awareness on the MM's part of the player's position on the curve, which I don't think exists yet - hence why the top and bottom sides are having trouble.

Take the top 20% of the ELO curve, and match those people, and match the bottom 20% within that 20%. The next regions in are say 10% wide, then 5%, then 2% and 1% slices, creating a much more flat player distribution within each ELO region. There would need to be a logarithmic sliding function to adjust your position within the regions as a second layer, favoring the center over the edges in searching for additional players.

Also, building teams after all 24 have connected is important, as only then can you really balance. Otherwise statistically the odds of that last man being who you are looking for balance are really small. This enables a more diverse ELO range to balance correctly across a match.

[Doc Andrews]

Matthew Craig, on 19 September 2013 - 08:00 AM, said:

Not quite as easy as it sounds the dedicated server expects a certain number of players and there is various code that respects those limits they don't fluctuate as easily as you'd expect.

If a match is found that could be launched as a match under 12 on 12, a number of virtual players are stubbed. The launch happens. Upon detecting the launch, the stubbed players 'disconnect'.

We're already accidentally launching 12 on 10. It's trivial to do it intentionally. If you have any questions, pop me a PM. I have quals at both Namco and PartyGaming. You can't afford me, but I love this game, so I don't mind dropping tips.

[retakrew7]

At the moment it seems too loose our 4 lance totally destroyed the enemy teams.

EDIT:

Still having the 3 person group issue.

Edited by retakrew7, 19 September 2013 - 12:51 PM.

[veri745]

Matthew Craig, on 18 September 2013 - 10:39 AM, said:

Sorry to say we simply can't leave certain players with such long times to find a match so we are slowly returning to settings closer to pre patch for now. As mentioned we'll be using all the data gathered to find better ways to get you those match qualities back potentially even better without degrading it for users at either end of the Elo spectrum.

Don't you know the average ELO and the distribution of players? If the matchmaker starts bucketing players together based on ELO, it seems like you could just use the known distribution of ELO to adjust the range based on the players in that bucket.

i.e. if you have players with near average ELO, use a tighter limit. For players with more extreme ELO ranks, widen the limit.

This should give players with more common ELO rankings good matches without sacrificing average queue time for high- or low-ELO players.

*edit* for bonus points. Use the distribution data for currently logged-in players instead of the total population.

Edited by veri745, 19 September 2013 - 12:58 PM.

[TC Living Tired]

I am curious if it would be possible to implement a system like this:

A kind of "ELO of the Day"

At login, players are put into a set baseline queue logic, where they are playing either above or below their "actual" ELO rating(Meaning if their overall performance over time ELO is higher or lower than the baseline queue, with the possibility of the effective ELO being adjusted somewhat higher or lower from that baseline toward their overall performance over time ELO, so if they are above the baseline to a significant degree overall, it would shift them a certain amount above the baseline, and a player below the baseline would accordingly be shifted a certain amount lower than the baseline to start them off closer to their probable skill level) . Depending on how they perform during these matches, their effective ELO is shifted to a higher rating if they do well or lower rating if they do poorly.

Among other things, this should provide a buffer by which a player having an "off day" will not be forced against players who are beating the daylights out of them. I imagine it will also help alleviate the problem where a player is leveling or figuring out a mech/build they may not be familiar enough with to play at their best skill level being similarly mismatched by their performance capability.

Most importantly to me, this should provide more even matching, as the queue logic will be based off of recent performance, not off of "the best they've ever been able to do." Giving skilled players a chance to warm up before being thrust into a feeding frenzy or to explore new mechs/builds without being a detriment to their team, and giving new or less skilled players the opportunity to face a greater variety of opponents and learn from them and that experience.

The idea is not to break these users into separate queues, but rather how they are handled within the same queue, using (perhaps as a voluntary selection to use this matching type over the "normal" current system in place, where the player has the choice to attempt finding matches with the standard matchmaking logic or can opt for the "Daily Performance" logic) their recent performance characteristics to better fit them into matches composed of similar performance players.

Just an idea I thought might be worth considering.

[Matthew Craig]

veri745, on 19 September 2013 - 12:54 PM, said:

Don't you know the average ELO and the distribution of players? If the matchmaker starts bucketing players together based on ELO, it seems like you could just use the known distribution of ELO to adjust the range based on the players in that bucket.

i.e. if you have players with near average ELO, use a tighter limit. For players with more extreme ELO ranks, widen the limit.

This should give players with more common ELO rankings good matches without sacrificing average queue time for high- or low-ELO players.

*edit* for bonus points. Use the distribution data for currently logged-in players instead of the total population.

Yes this is a valid approach and will likely be tabled internally.

[Throe]

Given that the match maker is actually making use of the fact that sometimes the queue fails to launch, I'd say make it much shorter. Like 60 seconds or less.

Then, everyone hits launch again, and the match maker can make a second attempt at arranging the players who have queued that much quicker. It doesn't help people who would prefer to queue and then go make a sandwich, in order to come back after the game has started just in time to play, but it does allow people who are sitting there waiting patiently actually get into a match sooner. Right?

[Rippthrough]

Personally, I would prefer for the extreme ends of the spectrum, rather than broadening the search massively and still ending up with unbalanced games, if the matchmaker simply balanced the players on either side and launched after 3 minutes, even if it means a 6 vs 6 or similar.

I would rather a close match between 6 skilled opponents than a steamroll on a 12 vs 12 because it grabbed 4 lower elo guys 3 minutes in.

If it is using the re-queues to rearrange players more optimally, rather than have it sitting waiting for a match for 5 minutes, can you not just force it to auto-refesh the queue every 60 seconds - that way you get the benefits of the above post but without the hassle to the end user.

Edited by Rippthrough, 19 September 2013 - 01:47 PM.

[Vulix]

While it takes longer to match me now (as it should for better MM), I have never failed to find a match (neither have my teammates). Games also feel more balanced to me now.

Edited by Vulix, 19 September 2013 - 01:50 PM.

[Monky]

Matthew Craig, on 18 September 2013 - 07:52 AM, said:

We monitored the data throughout the day yesterday and made adjustments to maintain a reasonable average wait time, that said we suspect it is right that very high or very low elo players are probably suffering longer than usual wait times. Reason being that Elo distribution looks like a bell curve so naturally players in the middle of the curve have more players they can match against.

We're looking over the data today and will continue to make adjustments that try to maintain as much of the restriction while maintaining our goals for wait times.

Please feel free to post your feedback on how long you are willing to wait to find a good match, and potential failed to find a match. A quick note on that, the failed to find a match can be a good thing for the match maker e.g. say 48 players trickle into the matchmaker the buckets fill up in the order the players join, now there might be two good matches between them but both buckets have only part of the players they need (like playing a game of cards where you are holding each others cards). When the failed to find a match triggers everyone hits launch again and now the buckets can grab the right players. So bear in mind that it does introduce a dynamic that helps the match maker, that simply extending the timer, or altering ranges does not.

Interested to hear everyone's thoughts as we continue to tune.

Perhaps change it from first come first serve to dynamic matching based on players in queue - maybe a 30 second hold before launching with a specific group to ensure a good match can be achieved.

Personally, a match is normally about 5 minutes long. I am content to hit launch and browse a website while I wait, I do this for other game like warthunder and suspect many others due given the behavior of people I play alongside (aka; they are doing the same thing I am, without prompting). 3-4 minutes is probably the maximum you can push, but it doesn't have to be instant either - a 30 second wait allowing the server to pick and choose even if it's right off the bat could do wonders for round balance.

Another possibility, and this is really throwing caution to the wind - is perhaps one size fits all matchmaking is the wrong approach. An accessible setting on the player's end that says 'I'm willing to wait longer for a better match' and 'I just want a match now' that the server can work with, and is also adjustable on the fly by the player to meet their immediate wants, could really be a better overall possibility.

Edited by Monky, 19 September 2013 - 01:59 PM.

[Throe]

Rippthrough, on 19 September 2013 - 01:46 PM, said:

can you not just force it to auto-refesh the queue every 60 seconds - that way you get the benefits of the above post but without the hassle to the end user.

I posted the 'above post' and I approve this message.

[Matthew Craig]

Yes it's something that will be discussed if we can do that behind the scenes. Something that was noted internally is that if you were able to continue to make changes in mechlab to other Mechs (not the one dropping) while you wait that would potentially help ease the strain of longer wait times?

[Monky]

Matthew Craig, on 19 September 2013 - 02:46 PM, said:

Yes it's something that will be discussed if we can do that behind the scenes. Something that was noted internally is that if you were able to continue to make changes in mechlab to other Mechs (not the one dropping) while you wait that would potentially help ease the strain of longer wait times?

This would be great if it could coincide with being able to save mech loadouts, as adjusting a 'loaded' mech (ie; one queued in the matchmaker) could cause headaches both for players and you guys if something went wrong. Simply charge 100k cbills for a loadout profile and 5 MC if you don't want to pay cbills, and it also becomes a free currency sink and another microtransaction for you guys.

Edited by Monky, 19 September 2013 - 02:56 PM.

[bobthebomb]

what a nice discussion !!

2 things you could try :

1) value elo modifier , You could modify the elo rating of a player by the price of his mech. So during the low pop time, high elo could try to use cheaper mech. If the price of a mech and his equipment is not a good representation of his power you could try fixed modifier ex : medium mech =elo*0.80, assault = elo*1.2.

Would solve the high elo problem, high elo player are usually smart and could decide when to play a stock hunchie
Low elo player may not be able to play highly customised mech (or don't even own one)

2) Rookie/star pool.
Used in a lot of team game when the game experience is very important ( mainly in table top game with random team generation)

-If 1 high ELO player (define high for your player base) fail to find a match X times. put him in the STAR pool with an ELO of 1.
When 2 players are in the STAR POOL, add them to the next game with 1 star per team.

Do the same for the low elo rating. with the title : ROOKIE

With this modification you can narrow the matchmaking and still include very high/low elo player.

During very very very low population time, you could even try 2 stars and 2 rookies per games (and 8 regulars)

[Throe]

Monky, on 19 September 2013 - 02:55 PM, said:

Simply charge 100k cbills for a loadout profile and 5 MC if you don't want to pay cbills, and it also becomes a free currency sink and another microtransaction for you guys.

While I understand the developers and production company deserve to, and indeed must be, paid, I still wish they would just can all this microtransaction baloney. At the very least, if they're going to continue with it, I would subscribe if there was a subscription model available which allowed me to play the whole game without having to think about the "free vs. pay" metagame. Even if it's a flat $15/month and there is no option for savings beyond that, I'd rather just pay it, and have my account otherwise locked if I stop subscribing, than deal with the existing convoluted system. Everything is just so, artificially limited for the advancement of the bottom line.

Matthew Craig, on 19 September 2013 - 02:46 PM, said:

Something that was noted internally is that if you were able to continue to make changes in mechlab to other Mechs (not the one dropping) while you wait that would potentially help ease the strain of longer wait times?

I sincerely hope this comes, with or without UI 2.0. I'd take it as a foregone conclusion, and it's frustrating to have to make myself 'un-ready' just to go look at it. My only solution currently takes me to another domain entirely, at Smurfy's 'Mechlab, which is less than desireable, no matter if you're player or developer. Keeping players in the game(or at least on the official website) makes them more likely to buy in. If players are forced to leave the game/website to find what they want, they're more likely to spend their money on the things advertised on those sites, rather than on the game.

[Monky]

Throet, on 19 September 2013 - 03:16 PM, said:

While I understand the developers and production company deserve to, and indeed must be, paid, I still wish they would just can all this microtransaction baloney. At the very least, if they're going to continue with it, I would subscribe if there was a subscription model available which allowed me to play the whole game without having to think about the "free vs. pay" metagame.

More options for the consumer is always better, but F2P games like MWO currently tend to work on the transaction basis, and at the very least having both free and non free options works for all players in a technical sense.