44 replies to this topic

[Fragnot]

I played some games on my T4 stock mech only account and was surprised to see Jay Z on the opfor one game. I recognized the name since I had recently watched the world finals playoff. He killed half our team by himself lol.

[Cabusha3]

It's less the lack of match maker and more the PSR system that's pretty effortless to level through. Takes only a couple of months for casual play to take you from Tier 5 to Tier 3, even with abysmal performance. So at tier 2 you'll still regularly group with people that really shouldn't be there yet.

[HauptmanT]

Just before I hit t3 (mid to high t4) i had a game with Baradul (Molten Metal Youtuber) maybe 2 weeks ago.

Within 2 days of hitting t3, I had M4J35T1C and Cuddles Time.

All 3 are t1. Guaranteed (as I've seen their screens!) and I should not have seen Baradul when I did. But I did.

In short, can confirm. MM broken.

[Ratpoison]

You know, these topics come up all the time, as it's easy to criticize, but not once has there been one that presented a skill matchmaking system that doesn't die screaming in the face of 12v12 matches. Matchmaking doesn't run on magic.

[Jun Watarase]

Tristan Winter, on 22 December 2016 - 12:49 PM, said:

I have no idea why it's like this. It's the Christmas holidays, the number of people playing MWO is probably a lot higher than usual. Yet the quality of matchmaking is going straight into the crapper.

Anecdotal evidence in the spoiler below, but anyone who has been playing in the solo queue these past few days or keeping an eye on the forum know what I'm talking about already.

Just got out of a match with my just-for-fun gauss rifle Shadowcat. Even with only 25 rounds of ammunition, I got the top score. I wasn't even playing very well, I was almost standing still in open terrain, just shooting at people who were new to the game and didn't know how to control their mech.

Anyone who has played with me or have checked my stats know that I'm not a particularly good player, even for Tier 2.

I know it's an established fact that a lot of "weekend warriors" arep laying during holidays and such, but that doesn't explain why MM has just given up.

Spoiler

Your sample size of 1 game isnt large enough.

Edited by Jun Watarase, 22 December 2016 - 05:33 PM.

[General Solo]

Could be a lot of T1,2 and 3 peeps playing CW and thus removed from the Solo que
Robbin Peter to pay Paul sort a thing
Robbin might be a bit of a strong word to use, but I think you get the idea

[Tarl Cabot]

And the seeded player could have been a tier 4 and you were the odd man out. No where does PGI provide examples of how gates open up, dependent on the seeded player. All they have posted are generalizations.

Edited by Tarl Cabot, 22 December 2016 - 06:41 PM.

[Tarogato]

If you cross out the names in the screenshots, I can't look at the match and tell you if it was balanced or not. So it's kinda useless.

[Snazzy Dragon]

Tarogato, on 22 December 2016 - 06:23 PM, said:

If you cross out the names in the screenshots, I can't look at the match and tell you if it was balanced or not. So it's kinda useless.

Forum terms of services and guidelines specifically mention not showing names in screenshots and name dropping, as far as I know.

Edited by Snazzy Dragon, 22 December 2016 - 06:32 PM.

[Tarogato]

Snazzy Dragon, on 22 December 2016 - 06:25 PM, said:

Forum terms of services and guidelines specifically mention not showing names in screenshots and name dropping, as far as I know.

The only lines in the CoC that seem to apply to screenshots that I can find:

• Disruptive or abusive behavior that negatively affects other player's experiences or the service
  
• Naming and shaming the alleged misconduct of another individual, including but not limited to accusations of cheating or exploiting.

Also, this post:

Primus of Comstar said:

Q: What about scoreboards? May I post a screenshot of my the MechWarrior Online end-of-match screen?
A: Yes, so long as it does not serve the purpose of humiliating or defaming another person. If you feel as though a post containing a scoreboard screenshot has been posted in an attempt to target, inflame or otherwise 'make fun' of a specific player or group of players, please use the forum Report function and our moderators will use their judgement accordingly.

Q: Does this mean that I do not need to obfuscate (blur/black/cover) the names on my screenshots if I'm simply looking to show off my score?
A: Correct, as long as you're posting in the spirit of friendly competition, there shouldn't be an issue.

So if you are not making disparaging comments about people in a screenshot (such as calling them scrubs or whatever), and you're not using them as an example of bad players (shaming them), or going "look how bad this player(s) is", then I don't see anything wrong with sharing score screens and saying "matchmaker didn't balance this properly".


Of course, then again, I did get a moderated once for making a thread to celebrate really good players, and then naming specific players and why I respect them. But that was a misunderstanding and was intended as parody, so lol. xD

Edited by Tarogato, 22 December 2016 - 06:56 PM.

[Fox With A Shotgun]

I think the problem is more that PSR is too heavily influenced by damage.

Consider this issue; a light that has to successfully alpha 20-30 times to score 500+ damage, vs an assault that has to alpha maybe 8-10 times. The light has to survive for much, much longer to contribute the same amount to the battle, all while having far lower survivability. A similar gradient happens for all mech weights from lower to higher; a smaller mech has to survive for longer to do as much damage, and thus to increase their PSR by the same amount.

What actually happens, therefore, is that a mediocre heavy or assault pilot gets pushed up tiers before his or her situational awareness or piloting skills is actually at an acceptable level for that level of play. Only through the merit of being able to dish out a metric crapton of damage because they can carry heavy ballistics, heatsinks etc. Because of this, a lesser-skilled heavy mech is actually pushed up faster than a more skilled light mech pilot; IIRC, Tarogato posted data that supports this (a top 5% light is actually scored about as well as a top 50% heavy, or some such).

Considering that heavies and assaults are actually the most populated mech classes, it should come to no surprise that the higher-tier assaults and heavies end up with a very wide range of actual player skill. This is why sometimes the MM doesn't look like it's working at all.

[Tarogato]

Fox With A Shotgun, on 22 December 2016 - 07:40 PM, said:

I think the problem is more that PSR is too heavily influenced by damage.

Consider this issue; a light that has to successfully alpha 20-30 times to score 500+ damage, vs an assault that has to alpha maybe 8-10 times. The light has to survive for much, much longer to contribute the same amount to the battle, all while having far lower survivability. A similar gradient happens for all mech weights from lower to higher; a smaller mech has to survive for longer to do as much damage, and thus to increase their PSR by the same amount.

What actually happens, therefore, is that a mediocre heavy or assault pilot gets pushed up tiers before his or her situational awareness or piloting skills is actually at an acceptable level for that level of play. Only through the merit of being able to dish out a metric crapton of damage because they can carry heavy ballistics, heatsinks etc. Because of this, a lesser-skilled heavy mech is actually pushed up faster than a more skilled light mech pilot; IIRC, Tarogato posted data that supports this (a top 5% light is actually scored about as well as a top 50% heavy, or some such).

Considering that heavies and assaults are actually the most populated mech classes, it should come to no surprise that the higher-tier assaults and heavies end up with a very wide range of actual player skill. This is why sometimes the MM doesn't look like it's working at all.

Match score disparity between weight classes in the PSR system is not a big deal. The actual problem is with PSR inflation - the tendency for all players above a certain threshold to drift toward and eventually reach Tier 1. Thus, there are really only three tiers.

Tier 5 - where the weakest players are who can't meet the threshold
Tier 1 - where everybody else gets to eventually
Tier 2, 3, 4 - where people are who are above the threshold, but still in the process of gradually climbing to Tier 1


This is due to the fact that PSR is not a zero-sum system. In other words, there are more ways for your PSR to go up than there are for your PSR to go down. It is intentionally imbalanced with the assumption (Paul's idea, I believe) that as you play more games, the better you are. Which is an atrociously fallacious abomination of logic.

[Fox With A Shotgun]

Tarogato, on 22 December 2016 - 08:15 PM, said:

Match score disparity between weight classes in the PSR system is not a big deal. The actual problem is with PSR inflation - the tendency for all players above a certain threshold to drift toward and eventually reach Tier 1. Thus, there are really only three tiers.

Tier 5 - where the weakest players are who can't meet the threshold
Tier 1 - where everybody else gets to eventually
Tier 2, 3, 4 - where people are who are above the threshold, but still in the process of gradually climbing to Tier 1


This is due to the fact that PSR is not a zero-sum system. In other words, there are more ways for your PSR to go up than there are for your PSR to go down. It is intentionally imbalanced with the assumption (Paul's idea, I believe) that as you play more games, the better you are. Which is an atrociously fallacious abomination of logic.

Yup. Skill ceiling and all that.

I'm still not sure why they discarded ELO, tbh. Sure, it's less accurate than it would be in smaller team games ala MOBA games, but it'd still be more accurate than the PSR 'Exp Bar'.

[Ratpoison]

Fox With A Shotgun, on 22 December 2016 - 08:20 PM, said:

Yup. Skill ceiling and all that.

I'm still not sure why they discarded ELO, tbh. Sure, it's less accurate than it would be in smaller team games ala MOBA games, but it'd still be more accurate than the PSR 'Exp Bar'.

ELO dies spectacularly in the face of 12v12. The more players on a team, the higher the margin of error, exponentially increasing the number of matches required to make an accurate judgment of player skill to the point where nearly no one is ranked properly at any given time, defeating the purpose.

Other games like League of Legends and CounterStrike: Global Offensive spent YEARS trying to overcome the problems of using an ELO system for 5 vs 5 games just to make them servicible, with HUGE playerbases to draw data from(and you still have to put in a ton of hours to get an accurate rank). Trying to do this with 12 people per team on a much smaller game is laughably impossible by comparison. This is why PGI went with a different formula that's more weighted on time spent; the simpler system leads to a higher number of well matched games than a more complicated ELO formula that struggles to function at all.

[Tristan Winter]

Ratpoison, on 22 December 2016 - 05:30 PM, said:

You know, these topics come up all the time, as it's easy to criticize, but not once has there been one that presented a skill matchmaking system that doesn't die screaming in the face of 12v12 matches. Matchmaking doesn't run on magic.

That's fine, but I think PSR has gradually decreased and it seems to happen a lot faster than player dropoff, from what little we know about how many active players there are (via counting Leaderboard stats). I remember actually being blown away by the PSR system when it first started. Then it deteriorated, but it was still better than the Elo system for a long time. But lately, I feel like it's not doing its job at all.

Cabusha3, on 22 December 2016 - 04:20 PM, said:

It's less the lack of match maker and more the PSR system that's pretty effortless to level through. Takes only a couple of months for casual play to take you from Tier 5 to Tier 3, even with abysmal performance. So at tier 2 you'll still regularly group with people that really shouldn't be there yet.

I don't think it only takes a couple of months anymore, I think it has slowed down considerably from when it started. I see several people on the forum who have stayed Tier 3 and Tier 4 for a looong time, for example. And I notice that my own progress on this account from T2 to T1 is now taking quite a long time.

My whole problem with Elo was that I had been playing for 2 years and I was getting matched up with people who had been playing for 2 weeks, even 2 days. That problem was gone when they introduced PSR, at least for me. I almost never see a trial mech anymore. Of course, in this match, there appear to have been a few trial mechs, since one team has 3 Stalker Champion mechs.

I don't expect PSR to fix the problem of divergent skill, even as much as it's trying to do that indirectly. I only expect it to shield the really fresh players from the really experienced players, because MWO has a steep learning curve. That's where it's failing, in my opinion.

Jun Watarase, on 22 December 2016 - 05:32 PM, said:

Your sample size of 1 game isnt large enough.

That criticism isn't valid in this case, because I'm not using the OP as proof to say that this is the norm. I'm adressing the people who have already experienced that MM is worse than it used to be. Consider it a case study, if you want to pretend the MWO forums is a journal for peer-reviewed papers.

I referred to it as anecdotal evidence in the OP.

Tarogato, on 22 December 2016 - 06:23 PM, said:

If you cross out the names in the screenshots, I can't look at the match and tell you if it was balanced or not. So it's kinda useless.

I don't really care that it's balanced or not. It could be that both teams had two Tier 2 players, two tier 4 players and eight tier 5 players. I don't think Tier 2 and Tier 5 players should be mixed at all, except if it happens during really weird hours when no one's playing. But that wasn't the case here. Indeed, it's fairly rare for me to wait any length of time to get into a match, really. Tends to happen a lot faster than other PVP games, by orders of magnitude, quite often.

[Ratpoison]

Tristan Winter, on 22 December 2016 - 10:10 PM, said:

That's fine, but I think PSR has gradually decreased and it seems to happen a lot faster than player dropoff, from what little we know about how many active players there are (via counting Leaderboard stats). I remember actually being blown away by the PSR system when it first started. Then it deteriorated, but it was still better than the Elo system for a long time. But lately, I feel like it's not doing its job at all.

Maybe, maybe not, I don't see much use in speculating things you can't know and drawing conclusions from it.

[Tristan Winter]

Ratpoison, on 22 December 2016 - 10:19 PM, said:

Maybe, maybe not, I don't see much use in speculating things you can't know and drawing conclusions from it.

That's fair enough, but that's a fair amount of what goes on in the MWO forums. Speculating about player population, speculating about PGI's long term business plan, speculating about MW5:Mercenaries, speculating about PGI's next big project for MWO in 2017, speculating about why PGI has done things in the past. We can't know the answers to any of this, anymore than I can know why Natalie Portman seemed to deliver a poor performance in the Star Wars movies when she's otherwise a very capable actor. But I can always speculate and draw my own conclusions, for fun.

[Ratpoison]

Tristan Winter, on 22 December 2016 - 10:27 PM, said:

That's fair enough, but that's a fair amount of what goes on in the MWO forums. Speculating about player population, speculating about PGI's long term business plan, speculating about MW5:Mercenaries, speculating about PGI's next big project for MWO in 2017, speculating about why PGI has done things in the past. We can't know the answers to any of this, anymore than I can know why Natalie Portman seemed to deliver a poor performance in the Star Wars movies when she's otherwise a very capable actor. But I can always speculate and draw my own conclusions, for fun.

Speculating facts can be constructive, speculating non-facts is just making noise. Your experience might be accurate, or it might just be bad luck or outside factors shading your perspective. Suggestions based on non-facts are exactly why the forums are largely ignored in the first place.

[Tristan Winter]

Ratpoison, on 22 December 2016 - 10:44 PM, said:

Suggestions based on non-facts are exactly why the forums are largely ignored in the first place.

That's just speculation.

[Ratpoison]

Tristan Winter, on 22 December 2016 - 10:45 PM, said:

That's just speculation.

Actually it's just game development common sense. Anonymous open recommendations is a recipe for huge piles of bad ideas. It's true of virtually all game forums, if you didn't ever notice.