125 replies to this topic

[Zoid]

Redshift2k5, on 22 September 2014 - 03:28 PM, said:

Yeah, actually, it is hard- balance team aggregate elo, balance team size cluster while retaining weight class balance (most of the time two 6s aren't gonna have exactly the same value of weight classes to fit together in 3333), all at the same time without making you wait more than 5 minutes.

Go write a better sorting algorithm and get back to us.

Not that hard if you make it so that the larger group is at a disadvantage in weight and ELO. It's not at all unreasonable to give the four groups of three 4 or 5 assault 'mechs when going against a full group of 12 with 3 assaults. I'm sure lots of people jump into the group queue with more than one heavy, medium or assault, it shouldn't be that hard to set up.

[EgoSlayer]

IceSerpent, on 22 September 2014 - 03:21 PM, said:

...and it remains broken. The size of a "bracket" is large enough to make match outcome luck-based, which means that rating doesn't accurately reflect player skill, which in turn makes the whole setup fairly useless.

It's only broken in the group queue. Solo queue is fine. It's the variety and disparity in groups that is throwing it off.

Russ Bullock, on 22 September 2014 - 12:16 PM, said:

Honestly any groups in the solo queue really screws it up. The solo Queue right now is actually pretty amazing with an average team elo difference of around 40. Exact weight class matching 95% of the time etc.

It is the group queue that is frustrating players - especially groups on the smaller side.

[IceSerpent]

Gyrok, on 22 September 2014 - 03:20 PM, said:

1. Coding what? Java script?

C, C++, Cache Objectscript

Quote

2. I would have made different design decisions, but what they have now, while a bit buggy, is not malfunctioning...

System that doesn't fulfill its primary purpose doesn't qualify as "a bit buggy, is not malfunctioning".

Quote

3. If you agree it is more complicated than you anticipate, then WTF?

"Programming is hard" is not a valid excuse in my book - if you develop something, the result has to meet basic project requirements. A bridge has to be able to support traffic, a car has to run, MM has to create fair matches.

Quote

4. I realize that sacrifices are made for the sake of more of one thing and less of another. I pulled the 20-30 minutes figure out of the air, but considering the PTS times were often 15 minutes under some circumstances, I think you can imagine it might not be too far off.

I don't want to take a SWAG a.k.a. imagine things. PGI has statistics, they should be able to get a fairly accurate number. Then we can talk about whether sacrifice is worth it or not.

Quote

As for making everyone happy, if you really are a coder, then, you and I both know that is a pipe dream that sales people talk about while they stand around and sing kumbayah at a convention. Especially in the world of video games. You constantly piss people off, you are just making educated guesses about who is going to be the least detrimental to piss off.

In my experience giving customers more options always makes them happy. When you have a group of folks who want it painted black and another group who want it painted red, throw in a color selector and both groups will be your best friends forever

Quote

As for this matchmaker crap...I can tell you this much, if they screw with the ability to select game modes, or the group queue size capabilities right now...it would literally kill a boat load of units. My entire active population would up and quit overnight if they did this...literally...our clan forums are full of comments like..."dealbreaker", "they gave us great groups, and now they kill them...done", "Deuces, call me when they unf*ck this sh!t"

I agree about game modes. Don't feel quite that strongly about it myself, but I have a bad premonition that involves people fake DCing when they end up in a mode they dislike. Not sure what you mean by "group queue size capabilities" though.

[Zoid]

IceSerpent, on 22 September 2014 - 03:36 PM, said:

I agree about game modes. Don't feel quite that strongly about it myself, but I have a bad premonition that involves people fake DCing when they end up in a mode they dislike. Not sure what you mean by "group queue size capabilities" though.

If you're going to quit over not being able to select "deathmatch" vs "deathmatch but you don't have to chase the lights at the end", good riddance.

[Koniving]

Einaescherin, on 21 September 2014 - 12:08 PM, said:

3*4 player grps vs 12 player grp
or
10player grp + 2 player grp vs 4 *3player grps

I don't like that either, but...
What's hilarious is 12 players versus 2 + 2 + 2 + 2 + 2 + 2.

But on a serious note:
12 player group? Forced into 3 lights, 3 mediums, 3 heavies, 3 assaults.
10 player group? Probably will be 3 assaults, 3 heavies and 3 (of either mediums or lights with 1 left overor 2 and 2).

That 4*3 group? 3 assaults, 3 assaults, 3 assaults, 3 assaults.
That 3*4 group? 3 assaults, 1 heavy. 3 assaults, 1 heavy. 3 assaults, 1 heavy.
That 2*6 group? 2 assaults, 2 assaults, 2 assaults, 2 assaults, 2 assaults and 2 assaults.


Unfortunately the matchmaker tries to nitpick tiny groups of smaller or equal mechs to prevent that.

[Threat Doc]

It's unfortunate that only two people in this thread actually absorbed what I wrote, and then commented on it properly. For those who have, you understood what I said; for the rest of you... just, stop.

I'm out.

And, and the two were Joseph Mallan and Ice Serpent; thanks, guys, for taking the time to read.

Edited by Kay Wolf, 22 September 2014 - 03:46 PM.

[Willard Phule]

I have no idea how the group queue works....I'm in the solo queue exclusively.

As far as the solo queue works, as soon as you monkeys realize that PGI's version of "increased difficulty" lies in the fact that the bettter you are, the more newbs you have to carry...the better off you'll be.

Face the facts. In the solo queue, you will NEVER....let me reiterate this....NEVER....be playing against anyone in your Elo bracket.

The better you are, the more bads you're expected to carry. That's how they make it tougher for you. It's not " the better you are, the more talented people you have to go up against."

Just accept it. You, the people in the solo queue are NOT the "target audience"....you're not the one paying their bills. It's all about the organized groups.

We the unwashed masses need to bend over and just accept what is.

[Project_Mercy]

Willard Phule, on 22 September 2014 - 04:49 PM, said:

The better you are, the more bads you're expected to carry. That's how they make it tougher for you. It's not " the better you are, the more talented people you have to go up against."

And the bads don't want to play with you either, But somehow the game is CONVINCED that because I started, I should be equal to the average person who's been playing for 2 years, and it's completely irrelevant how terrible I do in the match, because someone WILL eventually carry me every few games, and if I lose when I'm expected to lose (which is pretty much always) my ELO doesn't even go down.

So i start every game, and I get to hit tab and see 11 other people who are now at a massive disadvantage because I'm in this game, and there's nothing I can do about it. It's like I have to apologize at the start of every game because I know they're probably going to lose. Which, btw, is a great feeling to have.

I could seriously just spin in circles and blow out the back of every friendly mech I see, and baring someone reporting me, it would have almost no effect on what games I got into. Apparently I need to find the RIGHT matches to lose in for a few months(times four apparenly), before I can somehow figure out how to tank myself so hard that I can find people of equivalent skill. The irony is, at the point where I finally do that, I MAY have actualy figured out the game enough to actually play some, and then I'll be nowhere near people of similar skill.

Russ says that solo queue is fine, I can't imagine how this is fine. It looks more like something where new people were basically laughed at or ignored as expendable.

Even Guns of Icarus, who is like 2 guys in a closet somewhere, has a better new-player experience.

Edited by Wraeththix Constantine, 22 September 2014 - 05:59 PM.

[Wolfways]

Wraeththix Constantine, on 22 September 2014 - 05:21 PM, said:

Russ says that solo queue is fine,

If fine means that I'm always on the losing team and have to play for hours just to get one win then it's fine.

I play mainly for the battles, and wins are a bonus, but recently the amount of losses is just getting so annoying I'm getting bored with playing. I've just been playing for over four hours and still don't have the 2xXP on my KFX from yesterday! There aren't enough hours in the day to get the daily 2xXP.

As far as I can tell the MM always either puts me on the low ELO team, or puts me in the higher ELO team but puts better players in the other team.

EDIT: 1st match after making this post was a win. I know what to do now. Every time my team loses, post about it

Edited by Wolfways, 22 September 2014 - 06:10 PM.

[Cmdr Rad]

I like a lot about what Ice Serpent has wrote. Here's a question:

Let's say Russ decided to toss this to the community. What statistical data sets would the community need -specifically-, and how large of a sample size? Some context would be provided for the sample (or each, if multiple). For expediency's sake, let's say they'd release the data mostly raw (Except removing names), and leave it up to some people in the community with some know-how to format the data into something the whole community could utilize.

Some "armchair designers" are content to sit in their seat and spout their theories; others are on the edge of their seat, and would happily get out of their armchair if given the tools to prove their theories.

[IceSerpent]

Cmdr Rad, on 22 September 2014 - 08:27 PM, said:

I like a lot about what Ice Serpent has wrote. Here's a question:

Let's say Russ decided to toss this to the community. What statistical data sets would the community need -specifically-, and how large of a sample size? Some context would be provided for the sample (or each, if multiple). For expediency's sake, let's say they'd release the data mostly raw (Except removing names), and leave it up to some people in the community with some know-how to format the data into something the whole community could utilize.

Some "armchair designers" are content to sit in their seat and spout their theories; others are on the edge of their seat, and would happily get out of their armchair if given the tools to prove their theories.

You mean for approximating ballpark figures for the wait times? We would probably need timestamped log of group sizes, weight class of their mechs, and their Elo ratings for 24 hour period on a workday (making an assumption here that one workday is not much different from another and that more people play on a weekend). Obviously, getting that for a full week would be better, but it's 7 times more data and I am guessing we're talking tens of thousands of records per day at least.

[Cmdr Rad]

IceSerpent, on 22 September 2014 - 09:13 PM, said:

You mean for approximating ballpark figures for the wait times?

Basically, yeah. PGI has their own stats they've been looking at, and they've taken that data, drawn trends, and interpreted that data for their current model to come up with their own solutions. Perhaps if the community looked at the data, they'd draw different trends and conclusions, and then from there better theories and proposals could be created, and perhaps implemented in future revisions of the Matchmaking system.

In general, as we're moving forward as Russ keeps saying, I think more data in the hands of the community would be a good thing. I'm an EVE player of course (I know -just- enough to be dangerous with a spreadsheet), and I'm constantly amazed by all the statistical analysis that goes on by people in that community with the vast arrays of data, both current and historical, that they can harvest from the APIs, composing their own graphs and finding their own trends to support various theories on the game.

I don't think datasets like that would lead to the community coming up with solutions overnight, but I do think that given enough time, that enough people would go through the data to perhaps come up with something useful. PGI's cost? Just taking a bit of time to dump some data for the community to look at with fresh eyes.

[Alex Warden]

Bront, on 21 September 2014 - 12:30 PM, said:

Sometimes? Yes.

If after 3 minutes (estimate on my part) there's only 1 10 man team in queue, the matchmaker starts to do it's best.

I've been in a 4 man vs a 12 man (and won). It happens on occasion. Usually, the group sizes are similar.

yea happens... but speaking of matchmaker fail: i had a matchyesterday where the enemy had FOUR Direwhales, and our team had not a single 100 ton mech... afaik the heaviest we had was 2 warhawks... but hey, we had 4 lights (no, no spiders ) ...

[Phashe]

OP (and all forum gripes?): Listen to NGNG podcasts for updates before rants. This was a big topic in the 3.5 HOUR town hall that Russ did a few weeks back.

[Joseph Mallan]

Willard Phule, on 22 September 2014 - 04:49 PM, said:

I have no idea how the group queue works....I'm in the solo queue exclusively.

As far as the solo queue works, as soon as you monkeys realize that PGI's version of "increased difficulty" lies in the fact that the bettter you are, the more newbs you have to carry...the better off you'll be.

Face the facts. In the solo queue, you will NEVER....let me reiterate this....NEVER....be playing against anyone in your Elo bracket.

The better you are, the more bads you're expected to carry. That's how they make it tougher for you. It's not " the better you are, the more talented people you have to go up against."

Just accept it. You, the people in the solo queue are NOT the "target audience"....you're not the one paying their bills. It's all about the organized groups.

We the unwashed masses need to bend over and just accept what is.

By this logic, wouldn't those 'Bads' be expected to carry the 'Worse'. I mean if each of us is expected to carry those worse than us, Someone MUST be carrying you every drop as well William.

And FYI it is just as bad in the group que if you are the less organized team, so Don't think you are alone sir.

Edited by Joseph Mallan, 23 September 2014 - 01:48 AM.

[Joseph Mallan]

IceSerpent, on 22 September 2014 - 09:51 AM, said:

Joe, Elo and BV are two separate beasts. In TT there's no metric for player skill at all, "pilot" is essentially a part of the mech similar to tank crew in WoT. There's no real difference (aside from fluff) between better pilot and better targeting computer. Both are just doodads that change mech stats.

Regarding BV schemes being exploitable, they are no more exploitable than current 3/3/3/3 scheme - you can "exploit" former by using the most effective config that fits into chosen BV limit just like you can "exploit" latter by using the most effective config for a chosen weight class. With BV it's much easier to counter those exploits though, because it only requires changing BV of the element that makes that particular config so effective. With weight class you don't have the luxury of "fine tuning" JJ-capable mechs or PPC boats, which results in things like ghost heat.

Chosing the most effective config is not exploiting though. Taking off 2xERPPCs To have 12x MGs with no ammo to lower your BV so you can have a 1/1 P/G Is exploiting the system.

[Willard Phule]

Wraeththix Constantine, on 22 September 2014 - 05:21 PM, said:

And the bads don't want to play with you either, But somehow the game is CONVINCED that because I started, I should be equal to the average person who's been playing for 2 years, and it's completely irrelevant how terrible I do in the match, because someone WILL eventually carry me every few games, and if I lose when I'm expected to lose (which is pretty much always) my ELO doesn't even go down.

So i start every game, and I get to hit tab and see 11 other people who are now at a massive disadvantage because I'm in this game, and there's nothing I can do about it. It's like I have to apologize at the start of every game because I know they're probably going to lose. Which, btw, is a great feeling to have.

I could seriously just spin in circles and blow out the back of every friendly mech I see, and baring someone reporting me, it would have almost no effect on what games I got into. Apparently I need to find the RIGHT matches to lose in for a few months(times four apparenly), before I can somehow figure out how to tank myself so hard that I can find people of equivalent skill. The irony is, at the point where I finally do that, I MAY have actualy figured out the game enough to actually play some, and then I'll be nowhere near people of similar skill.

Russ says that solo queue is fine, I can't imagine how this is fine. It looks more like something where new people were basically laughed at or ignored as expendable.

Even Guns of Icarus, who is like 2 guys in a closet somewhere, has a better new-player experience.

Actually.......

I played a little experiment last weekend because I was so tired of being put on the "carry me" team and found something that actually works.

The trick is to do less than 100 damage. Doesn't matter how you do it.....you can load a single small laser or just follow someone around and never fire a weapon. The point is to do as little damage as possible.

Once you've done that for 3 or 4 matches (depending on server population), it puts you on the team predicted to win because you obviously need to be carried.

It's pretty hilarious.

Joseph Mallan, on 23 September 2014 - 01:46 AM, said:

By this logic, wouldn't those 'Bads' be expected to carry the 'Worse'. I mean if each of us is expected to carry those worse than us, Someone MUST be carrying you every drop as well William.

And FYI it is just as bad in the group que if you are the less organized team, so Don't think you are alone sir.

I get what you're saying, but try to see it from where I'm at.

We were told, back when the matchmaker was re-written, that New Players (and I mean still in their first 25 matches) were going to be given an Elo score that was low enough to keep them out of general population. That is, in fact, not the case. It may be due to server population at any given time but I can't tell you the number of times recently that we've had people ask in general chat how to move forward, lock targets, etc. Sure, they may be trolls, but they were playing like they sounded.

And, I'll be honest, even if a PGI developer told me that it was completely impossible for new players to enter general population, I'd have a hard time believing them based on their track record of honesty....so, there you go.

Oh, I'm sure I have to be carried sometimes....again, based on server population. If I, a pretty standard mid-ranked guy, happen to be one of the "high Elo" guys on at any given point, it's obvious I'm going to be the one doing the carrying. If I'm one of the lower ones, I'm going to be put on a team as somone to be carried. Thing is, I tend to play when the population is low (I never play in NA prime time....I've always got something going on of an evening). It is what it is.

I stopped caring a while back. In fact, the only reason I'm actually playing right now is because Pathfinder Online's Alpha Test server is down at the moment (let me tell you, their definition of "minimally viable product" is WAY different than PGI's). I'll join a match, see what mechs are out there....watch them move forward and just do my thing. When the wave of derp starts, I just go with it and make fun of everyone....at least I can get some enjoyment from that.

As for the group queue, I figured the biggest problem would have been the small groups being put in with the larger ones. I mean, since the MM averages Elo scores....larger groups would have an overall higher average as opposed to smaller ones. Hence, a 10 man would get put with a low scoring 2 man and be expected to carry them.

I mean...tell me if my math is wrong here...something like this:
10 man team (1x1000Elo, 8x1500Elo, 1x2000Elo= Average of 1500Elo) needs a 2 man to finish out the team. The only 2 man on at the time consists of 2 players with 800Elos. The MM isn't really going to be concerned with the overall average since that's the only 2 man available.

Edited by Willard Phule, 23 September 2014 - 04:06 AM.

[IceSerpent]

Joseph Mallan, on 23 September 2014 - 01:52 AM, said:

Chosing the most effective config is not exploiting though. Taking off 2xERPPCs To have 12x MGs with no ammo to lower your BV so you can have a 1/1 P/G Is exploiting the system.

Joe, what would be the point of doing something like that in MWO? You certainly can lower your BV by dropping weapons, heatsinks, etc., but what advantage do you gain fom that? Let's say a fully pimped Atlas has BV of 2000, I change weapons to a single small laser, drop engine to 200, and end up with BV of 500 (for example)...I don't see any apparent benefits from going through all this trouble.

[Gen Joe]

I really think that the Matchmaker is doing good job, but with the current "settings" he is killing the fun for small (2-4) and not well organized groups of friends smashing them against a 12-er Clan lance, well organized, sitting in Teamspeak and playing all together. At the end he splits this small groups and forces them to play solo, because of this frustrating games. So happen to me and my friends. We are not really bad players, but in 70-80% of the games we don't have the chance to play a nice (and a bit more fair) game. As result we all started to play solo and more and more other games.

I personally love Mechwarrior and wanted to buy the "Clan Wave II"-edition, but my motivation to do this is sunken with every game vs. 12-er lance and every game that I now "could not" play with my friends. So here my suggestions how to handle the situation:

1. We need 3 queues: solo, (2-4) lance and company. Although the company-queue my have long wait times, it isn't an option to satisfy those players for the price of frustrating/losing many others! Speaking of being satisfied: I don't understand how it can be satisfying for players to roll with 12-er lance over any combination of single lances. For me it is a shameful challenge, that would not satisfy me.

2. Because not all like this (hopefully in the future) general matchmaker behavior and to help to keep the wait times in the queues low, all players have the permanent, but fast changeable, settings option to signalize the matchmaker that they also want to be taken into account in other queues in following manner:

- solo players can choose to also be taken into account in the lance-queue, and (separate option) in the company-queue
- the same options of a lance creator will be taken for the whole lance, so he can sign in the lance to be taken into account in the company-queue
- the company leader should have the option to start with a non-full lance, that must not be filled from matchmaker, to show how strong they are

3. In company-queue the limits can maybe be dropped, because a well organized team can win with many tactics and mech compositions. So no need here to limit anything.

I think such settings could prevent that unorganized lances, only playing for a bit fun after work and to meet their friends in a shared game, also can have their fun and stay motivated in the game, while waiting times can be reduced in the queues.

The already proposed suggestions as fair distribution of ECM and Assaults are also welcome, as just a common start position of all players in solo matches as well, because that would help random players to play together.

Edited by Gen Joe, 25 September 2014 - 09:35 AM.

[IceSerpent]

Gen Joe, on 25 September 2014 - 04:22 AM, said:

I really think that the Matchmaker is doing good job, but with the current "settings" he is killing the fun for small (2-4) and not well organized groups of friends smashing them against a 12-er Clan lance, well organized, sitting in Teamspeak and playing all together.

Mathing a bunch of 2-man groups against a full 12-man doesn't qualify as "Matchmaker is doing good job"...

Quote

1. We need 3 queues: solo, (2-4) lance and company. Although the company-queue my have long wait times, it isn't an option to satisfy those players for the price of frustrating many others!

Won't work - how would you match a group of 5 in your suggested setup?

Quote

2. Because not all like this (hopefully in the future) general matchmaker behavior and to help to keep the wait times in the queues low, all players have the permanent, but fast changeable, settings option to signalize the matchmaker that they also want to be taken into account in other queues in following manner:

- solo players can choose to also be taken into account in the lance-queue, and (separate option) in the company-queue
- the same options of a lance creator will be taken for the whole lance, so he can sign in the lance to be taken into account in the company-queue
- the company leader should have the option to start with a non-full lance, that must not be filled from matchmaker, to show how strong they are ;-)

Allowing solo players a choice to enter group queue is IMHO a very good idea and has been asked for before. As a side benefit, it also allows 11-man group to play. PGI didn't like it though, so it probably won't happen.

Quote

3. In company-queue the limits should be dropped, because a well organized team can win with many tactics and mech compositions. So no need here to limit anything.

We had a 12-man only queue (assuming that's what you mean by company-queue), it didn't work.

The problem really boils down to MM having a tendency to match a very large group against a bunch of small groups. If that gets fixed we can theoretically even put everybody back into a single queue, as problem of solo players being "farmed" wouldn't exist any longer.