Paging Karl Berg...karl Berg, Please Pick Up The White Courtesy Phone...
#1461
Posted 25 August 2014 - 11:50 AM
#1462
Posted 25 August 2014 - 12:41 PM
Karl Berg, on 24 August 2014 - 09:44 PM, said:
Yup, very sorry all, but I've been in crunch for the last several weeks; mostly due to Matchmaker maintenance, CW phase 1, and the new data center provisioning and setup. I've been attempting to keep up with reading all of your replies in the meantime.
I hope you saw my question
#1463
Posted 25 August 2014 - 03:58 PM
p4r4g0n, on 25 August 2014 - 12:07 AM, said:
Glad to hear from you, fully understand the workload constraints and appreciate that you took the time to drop us a note. Be interested to know the reason for the change in the data centre when you have the time or are able to talk about it.
I'm just glad he's not dead.
#1466
Posted 31 August 2014 - 11:57 AM
Karl Berg, on 05 July 2014 - 12:44 PM, said:
This increases non-linearly with group size, and I have much poorer data for 5+ groups due to the reduced set of sample data. But as we continue gathering data on production I will continue to be able to refine these numbers and improve these adjustments.
Sir: I was wondering if these values are holding up?
#1467
Posted 02 September 2014 - 11:17 AM
Since you are the matchmaker guru
There seems to be a lot of complaints about the matchmaker on the forums at the moment. Some people are complaining of streaks of losses and unbalanced matches. I haven't experienced the issue myself but I was wondering if you have any statistics on win/loss streaks for players and whether there may be a subset of the player base that is experiencing some sort of bias from the matchmaker? Perhaps depending on weight class or Elo? ... which results in a streaky win/loss result. Maybe the player is near an Elo bin boundary in match formation?
Also, could a match balance issue be related to Elo bin expansion as the timer for match creation increases? Does the matchmaker form each team in an alternating manner so that when release valves are opened both teams are equally affected? If the matchmaker works on one team at a time and then can't find enough to form the other team and then opens the release valves ... there is the possibility for uneven matches to be formed.
Anyway, I think the player base could benefit from a detailed explanation of the algorithms used in the matchmaker, the nature of the release valves and the points at which they kick in.
I was also wondering if it would be possible to list the difference in Elo between the two teams and the spread of Elo on each team as part of the end game screen? This would not reveal individual player rank but it would let the players know that the MM is working correctly and give them an idea of the player skill balance in the match.
#1468
Posted 02 September 2014 - 03:47 PM
I suspect that the experience of bias is mainly one of perception rather than reality and is based on a specific duration rather than overall. Unless someone is using a third party app to collect match data as they occur, it is not unusual for individuals to feel that the matchmaker has done them in when getting a few losses in a row.
Sadly, MechCollect which was a great tool for monitoring match results no longer works since the data centre transfer or I'd recommend it to anyone who needs to collect data on streaks of losses.
@Karl Berg
On a related note, AFAIK one of the matchmaker's current safety valves allows for broadening of the Elo ranges over time when attempting to create a match particularly evident for players in the extreme ends of the Elo ratings range. This is perfectly understandable since PGI would not wish for such players to have extremely long waiting times.
However, one unforeseen result of this is that it is creating a disincentive for players to form larger groups as there seems to be a distinct direct correlation between larger groups sizes (threshold appears to be around 6) and a much higher possibility of a match up with competitive teams. Such match ups frequently result in a series of 12-0 3 minute matches which frequently leads to quick dissolution of such groups and is reminiscent of the old 12 man queue.
IMO this is not a desirable outcome as it does not allow players to gradually increase their skill levels or level of teamwork. In other words, the transition from grouping for relatively casual play to full blown meta mech, competitive level skills / teamwork is brutal and abrupt.
Following from a discussion with some friends after a recent series of such matches, I wonder if there is any merit in putting in some additional filters on the matchmaker to prevent the formation big high Elo groups unless said groups are prepared to accept unlimited or much longer wait times to find a closely matching team. The benefits of this could be:-
- elimination of the need to use group size modifiers (penalties) which does not actually take into consideration skill / experience;
- reducing the disincentive to lower Elo rated players to form bigger groups;
- creating larger numbers small high Elo groups which in turn would would make it easier for the matchmaker to balance teams; and
- reduced wait times for high Elo players as there are more groups of equivalent Elo ratings for matching.
Average Group Elo : Max group size
2,300 to 2,500 : 2
2,100 to 2,299 : 3
1,800 to 2,099 : 4
The only downside that comes to mind is that high Elo players may object to limitations in the number of friends they could play with at any one time in the public group queue. However, given the amount of time high Elo players already spend with similar rated players in practice, scrimmages, etc this may not really be an issue in most cases.
Edited by p4r4g0n, 02 September 2014 - 03:58 PM.
#1469
Posted 03 September 2014 - 02:18 AM
I have been seeing exactly what you describe in my matchups on a repeated basis. Being in Australia probably limits the choice of match ups even more during our prime time and when ever we form up a 12 man we repeatedly get matched up against known high level players and teams. Consequently we get stomped.
To add to the pain our 10 or 12 man is often matched against smaller groups of known high level players. The consequence of this is that the Elo handicap pitches better players at us as well as giving them a tonnage advantage over our team as well. whilst we have to comply with the rule of 3's the other team is sometimes composed of 5 or 6 assaults 4 madcats and maybe a light or a medium or 2. The release valves are a good idea but when they start throwing high Eo players with a tonnage advantage against more casual groups then disaster is about to happen. I see this on an almost nightly basis so it is not my imagination.
Karl check out my Elo and then tell me if I should ever be matched against House of Lords players, team or not. These might be edge cases but MWO seems to have a lot of edges.
I like a challenge but some match ups are getting ridiculous.
Edited by slide, 03 September 2014 - 02:19 AM.
#1470
Posted 05 September 2014 - 11:44 AM
That was enough to have us end up playing against a small team of the very best the game has to offer.
On the upside it was a privilege seeing Wispsy at work, on the downside it was more than frustrating coping with 2+ elite pilot SDR 5D s. Or not cope, as happened here. And I doubt the Lords guys had more than a moderate bore of time there, the skill difference was that noticeable.
Edited by Hammerhai, 05 September 2014 - 11:46 AM.
#1471
Posted 05 September 2014 - 03:33 PM
Hammerhai, on 05 September 2014 - 11:44 AM, said:
That was enough to have us end up playing against a small team of the very best the game has to offer.
On the upside it was a privilege seeing Wispsy at work, on the downside it was more than frustrating coping with 2+ elite pilot SDR 5D s. Or not cope, as happened here. And I doubt the Lords guys had more than a moderate bore of time there, the skill difference was that noticeable.
That is what happens when you match teams based upon a "team Elo". You get 12 medium-skilled pilots against 4-8 high-skilled and 4-8 low-skilled pilots.
Team 1
1500 score x 12 players = 18,000 Team Elo
Team 2
2700 score x 4 players (10800) + 900 score x 8 players (7200) = 18,000 Team Elo
#1472
Posted 05 September 2014 - 04:26 PM
Cimarb, on 05 September 2014 - 03:33 PM, said:
Team 1
1500 score x 12 players = 18,000 Team Elo
Team 2
2700 score x 4 players (10800) + 900 score x 8 players (7200) = 18,000 Team Elo
One game out of 10 billion, this maybe happens.
#1473
Posted 05 September 2014 - 04:32 PM
I don't believe you …
#1474
Posted 05 September 2014 - 06:19 PM
Heffay, on 05 September 2014 - 04:26 PM, said:
One game out of 10 billion, this maybe happens.
Not so I am afraid. I got one small group of Lords players 3 games in a row, mixed with different groups, and then a separate groups of Lords player the next 2 games. This can happen multiple times a week it is not a rare occurrence at all.
It is either because they cannot get 12 man games in their Elo bracket or they are deliberately gaming the system, knowing they will get mixed in with low Elo players so they can pad their stats. From the comments in game I would suggest the latter. Why they would need to is beyond me but with some people I guess there is no limit to their ego's.
#1475
Posted 05 September 2014 - 08:02 PM
slide, on 05 September 2014 - 06:19 PM, said:
Not so I am afraid. I got one small group of Lords players 3 games in a row, mixed with different groups, and then a separate groups of Lords player the next 2 games. This can happen multiple times a week it is not a rare occurrence at all.
It is either because they cannot get 12 man games in their Elo bracket or they are deliberately gaming the system, knowing they will get mixed in with low Elo players so they can pad their stats. From the comments in game I would suggest the latter. Why they would need to is beyond me but with some people I guess there is no limit to their ego's.
As long as they are not deliberately creating groups to game the group average Elo, I have no problems with them splitting into smaller groups to drop in public queue. They do still need to raise C-BIlls and XP somehow right?
Protip: Add them to your friend list and only launch when they are deployed
#1476
Posted 05 September 2014 - 09:58 PM
Don't be get me wrong this is not aimed at high Elo players, I was just pointing out that it happens way more often that Mister Ever Optimistic suggests.
Fundamentally it is a problem with match maker or more likely a low player base in the group queue. There is an extreme gulf between casual players who want to group up and professionals. Elo probably can't tell the difference. The sad thing is, it drives the casuals back to the pug queue, thus reducing the numbers of those who play which makes it harder to find matches which means more match ups against better players. It's a vicious circle.
#1477
Posted 06 September 2014 - 12:17 AM
For Karl;
Technical question about a single item in the game,
- TAG -
Is there a specific reason that the described infrared beam is actually visible in normal vision for all to see?
- Is there some coding issues that do not allow for it's function to only be seen in thermal?
Currently it rather destroys the stealth of spotters if the beam can just be followed back to the 'shooter'.Is it 'on the list' of items to look at in the future? Designing a beam that is invisible to all unless in thermal? Possibly adding a filtered view mode to those that are using it so they can see their TAG.
Thanks,
9erRed
#1478
Posted 06 September 2014 - 09:15 AM
How much have you been working on the new PGI game.
Thanks
#1479
Posted 06 September 2014 - 10:17 AM
Some questions on the Matchmaker;
1. How does the matchmaker prioritize filling teams? Does it grab the single largest bundle fitting 4x3 in the same Elo range to fill one team or does it just start with oldest waiting?
2. Are maps dedicated servers that cycle repeatedly or is it virtual servers on a rotation? I'm noticing that with some careful timing I can consistently hit the same map. People do this on accident sometimes and I'm not sure if it's just luck or a timing issue.
You're a boss, Karl. Thank you again for the investment of time and effort into keeping this thread going. You do your thing, best of luck to you guys with the changes with the data center.
#1480
Posted 06 September 2014 - 10:32 AM
I'm going to start working backwards. Probably easiest this way. There has been a lot of discussion and questions over the last couple weeks.
MischiefSC, on 06 September 2014 - 10:17 AM, said:
The matchmaker 'seeds' with an oldest player or group. Once it's decided on it's seed, it then proceeds to iteratively add groups / players to the smallest team until the game is full and ready for launch. The criteria for how it adds players is complex and deserves a fairly in-depth post; but beyond the initial seed, there is no consideration of age.
MischiefSC, on 06 September 2014 - 10:17 AM, said:
Depends. For all public games, the selection of map is chosen randomly by the server when it first boots. Long before any players connect to it. When the game ends, that entire dedicated server process also ends, and a new process is immediately started up to take its place. Obviously for private matches, the map can be overridden by the players in their lobby, so we inform the dedicated server to throw away it's random map selection to use the player lobby selection instead.
3 user(s) are reading this topic
0 members, 3 guests, 0 anonymous users