504 replies to this topic

[Sandpit]

Deathlike, on 09 March 2014 - 01:43 PM, said:

I experience roflstomps on many levels (for and against), so I'm well versed in this.

I'm actually unsure how it will create bigger problems for lone wolves and new players... outside of the hairbrained Elo adjustment done after collecting the Cadet bonus. If you're suggesting that new players will not get "edumacated properly" at the lower Elo tiers (because, they will all be playing together), I guess you could have a point there. There are still very few tools at your immediate disposal (outside of downloading TS3 and join the "hordes" on Comstar or NGNG) that facilitate this easily. Although, if you check up on "theplan", PGI "appears" to agree with that assessment, but I sincerely don't know how they'll actually accomplish this (I mean, tutorials finally became srs business now?)

I'll just leave this thread with a deep sigh.

No No,

What I'm saying is that due to a smaller population the MM will have no choice but to expand its parameters and pull from further out on the Elo spectrum. If you only have 10 players to choose from your options are much more limited as opposed to ahving 100 players to choose from.

We know new players will drop in the solo queue at first because that's simply the easiest way to jump into a game. Hit launch, shoot, rinse, repeat. That means you'll have new players playing alongside a smaller population. They'll ahve no choice but to play against what I mentioned above. As more players gravitate to the group queue, less players will be available in the solo queue.

The only way I can see that not happening is give solo players an option to choose both queues when hitting launch. So solo players would be able to click "all" and drop in with the first match they find regardless of queues. That presents a completely different list of problems though.

That's why I think a separate queue would hurt the new player experience though

[Roadbeer]

Deathlike, on 09 March 2014 - 01:55 PM, said:

Don't be using up my playbook! I was gonna save that for the "Launch Module" release...

©
Suck it.

Deathlike, on 09 March 2014 - 01:55 PM, said:

6-man queues are DOA by 2016.

Is that a serious proposal of something they're planning on doing?
I have to admit I'm somewhat out of the loop on "hare-brained" schemes this week.

Edited by Roadbeer, 09 March 2014 - 02:04 PM.

[Deathlike]

Sandpit, on 09 March 2014 - 01:55 PM, said:

No No,

What I'm saying is that due to a smaller population the MM will have no choice but to expand its parameters and pull from further out on the Elo spectrum. If you only have 10 players to choose from your options are much more limited as opposed to ahving 100 players to choose from.

We know new players will drop in the solo queue at first because that's simply the easiest way to jump into a game. Hit launch, shoot, rinse, repeat. That means you'll have new players playing alongside a smaller population. They'll ahve no choice but to play against what I mentioned above. As more players gravitate to the group queue, less players will be available in the solo queue.

The only way I can see that not happening is give solo players an option to choose both queues when hitting launch. So solo players would be able to click "all" and drop in with the first match they find regardless of queues. That presents a completely different list of problems though.

That's why I think a separate queue would hurt the new player experience though

Now I'm kinda confused.

I'm only going by my "imagination" with respect to how Paul put those pictures out.

Solo/Group -> Sorted by Elo bucket -> 1 Premade + Non-Premade Filler -> Drop

If you're saying it's simply a refinement of what we have... then we have a problem...

Solo/Group -> Add Group Restriction Through Client Interface -> Use Same Elo formula (in MM v4+) -> 1 Premade + Non-Premade Filler -> Drop

What you're essentially saying is that we will get terribad Elo filler players, probably in the least desirable weight class brackets.

...

...

...

/FACEPALM

[Roadbeer]

Deathlike, on 09 March 2014 - 02:06 PM, said:

If you're saying it's simply a refinement of what we have... then we have a problem...

Solo/Group -> Add Group Restriction Through Client Interface -> Use Same Elo formula (in MM v4+) -> 1 Premade + Non-Premade Filler -> Drop

Yeah, I'm still somewhat confused by this, but the description I quoted is the gist of what I'm getting from it.

Premade [average of 4 players Elo = Elo, 4 of the heaviest slots taken (because honestly, who's going to trust a PUG with an assault slot)] + 8 solos with Elo in the same 'bucket" [there were 3 tiers of buckets] (filling in 2 of the heavy and the rest of the medium/light slots, because, lets be honest, nobody cares if a hunchback derps its way into Mt. Doom) = Team

[Sandpit]

Deathlike, on 09 March 2014 - 02:06 PM, said:

Now I'm kinda confused.

I'm only going by my "imagination" with respect to how Paul put those pictures out.

Solo/Group -> Sorted by Elo bucket -> 1 Premade + Non-Premade Filler -> Drop

If you're saying it's simply a refinement of what we have... then we have a problem...

Solo/Group -> Add Group Restriction Through Client Interface -> Use Same Elo formula (in MM v4+) -> 1 Premade + Non-Premade Filler -> Drop

What you're essentially saying is that we will get terribad Elo filler players, probably in the least desirable weight class brackets.

...

...

...

/FACEPALM

100 players
10 players
which gives more variety for an MM to pull players from and match up more evenly?

so
if players leave the solo queue to join the group queue you have less players in the solo queue
new players are in solo queue
ggclose dabgs are in solo queue
less players in the queue means new players have less players to be matched up against.

[Deathlike]

Roadbeer, on 09 March 2014 - 02:12 PM, said:

Yeah, I'm still somewhat confused by this, but the description I quoted is the gist of what I'm getting from it.

Premade [average of 4 players Elo = Elo, 4 of the heaviest slots taken (because honestly, who's going to trust a PUG with an assault slot)] + 8 solos with Elo in the same 'bucket" [there were 3 tiers of buckets] (filling in 2 of the heavy and the rest of the medium/light slots, because, lets be honest, nobody cares if a hunchback derps its way into Mt. Doom) = Team

Yes, I remember this. I think they already effectively do this in the current MM, so the work there is already done (averaging the Elo of the premade group - which is kinda easy to fudge/mess with). Essentially, if your premade has the super-meta mechs, you're actually pushing more ggclose results. Effectively you have to ask the question "is our premade better than the opposing premade" to determine whether or not you win. That is bad... on all levels.

This is "less effective" as the premade size is smaller however. As the premade size gets smaller, the queue becomes closer to a literal solo PUG queue. I'm not sure how teamplay will get better because of that.

Sandpit, on 09 March 2014 - 02:14 PM, said:

100 players
10 players
which gives more variety for an MM to pull players from and match up more evenly?

so
if players leave the solo queue to join the group queue you have less players in the solo queue
new players are in solo queue
ggclose dabgs are in solo queue
less players in the queue means new players have less players to be matched up against.

Well, the queue assumes that the ratio of grouped players to PUGs has a rather low (or high?) ratio (at this point, I can't tell). Essentially, there needs to be at least 8 solo PUGs for every 4-man premade (9 for 3-man, 10 for 2-man). Someone would have to do the math for me, but if and only if PGI's metrics are correct in the first place, then this is not a problem.

Since we're trying to want more groups, the entire idea by PGI would implode on itself... but since the tools for playing together are not there, PGI's "metrics" will actually win the day (and not for the rest of us - it becomes a self-fulfilling prophecy).

Edited by Deathlike, 09 March 2014 - 02:26 PM.

[Roadbeer]

Afterthought...

I don't know if this was covered in the 70 pages of the Launch Module feedback, but has it occurred to anyone how incredibly easy it is going to be to game the MM now?

Just a cursory look at my mech stats in my profile and I can pretty much 'guesstimate which of the 3 "buckets" of the Elo spectrum I'm going to fall into.
So now, instead of smaller groups trying to sync drop who have ~20% chance of syncing. All you need to do is build one group of 4 then have everyone else around that can't quite build a 12 man take one of the unused weight classes and also guesstimate their Elo "bucket", and all try to sync.

I'm betting we'll get a lot closer to 6+ players on a team than we have a shot at doing now.

Edited by Roadbeer, 09 March 2014 - 02:27 PM.

[Triordinant]

Roadbeer, on 09 March 2014 - 02:23 PM, said:

Afterthought...

I don't know if this was covered in the 70 pages of the Launch Module feedback, but has it occurred to anyone how incredibly easy it is going to be to game the MM now?

Just a cursory look at my mech stats in my profile and I can pretty much 'guesstimate which of the 3 "buckets" of the Elo spectrum I'm going to fall into.
So now, instead of smaller groups trying to sync drop who have ~20% chance of syncing. All you need to do is build one group of 4 then have everyone else around that can't quite build a 12 man take one of the unused weight classes and also guesstimate their Elo "bucket", and all try to sync.

I'm betting we'll get a lot closer to 6+ players on a team than we have a shot at doing now.

If this is true, then it's another reason to have a premade-free, PUG-only queue.

[Sandpit]

Deathlike, on 09 March 2014 - 02:22 PM, said:

Yes, I remember this. I think they already effectively do this in the current MM, so the work there is already done (averaging the Elo of the premade group - which is kinda easy to fudge/mess with). Essentially, if your premade has the super-meta mechs, you're actually pushing more ggclose results. Effectively you have to ask the question "is our premade better than the opposing premade" to determine whether or not you win. That is bad... on all levels.

This is "less effective" as the premade size is smaller however. As the premade size gets smaller, the queue becomes closer to a literal solo PUG queue. I'm not sure how teamplay will get better because of that.




Well, the queue assumes that the ratio of grouped players to PUGs has a rather low (or high?) ratio (at this point, I can't tell). Essentially, there needs to be at least 8 solo PUGs for every 4-man premade (9 for 3-man, 10 for 2-man). Someone would have to do the math for me, but if and only if PGI's metrics are correct in the first place, then this is not a problem.

Since we're trying to want more groups, the entire idea by PGI would implode on itself... but since the tools for playing together are not there, PGI's "metrics" will actually win the day (and not for the rest of us - it becomes a self-fulfilling prophecy).

Just about everything I've said is in regards to separating the queues into solo and group. I'd love it. New players wouldn't though. I would put new players in a situation where they're facing higher Elos more often and more of the ggclose crowd the enjoy stomping on them. Those new players will either gravitate (if they stick around long enough to learn about it) to the group queue or they'll just quit.

Roadbeer, on 09 March 2014 - 02:23 PM, said:

Afterthought...

I don't know if this was covered in the 70 pages of the Launch Module feedback, but has it occurred to anyone how incredibly easy it is going to be to game the MM now?

Just a cursory look at my mech stats in my profile and I can pretty much 'guesstimate which of the 3 "buckets" of the Elo spectrum I'm going to fall into.
So now, instead of smaller groups trying to sync drop who have ~20% chance of syncing. All you need to do is build one group of 4 then have everyone else around that can't quite build a 12 man take one of the unused weight classes and also guesstimate their Elo "bucket", and all try to sync.

I'm betting we'll get a lot closer to 6+ players on a team than we have a shot at doing now.

I've thought about it. Just another reason the new launch module isn't going to fix anything I guess. The launch system is just not going to work the way they think it is.

[Deathlike]

Roadbeer, on 09 March 2014 - 02:23 PM, said:

Afterthought...

I don't know if this was covered in the 70 pages of the Launch Module feedback, but has it occurred to anyone how incredibly easy it is going to be to game the MM now?

Sure, and lots more ggclose to be had.

Quote

Just a cursory look at my mech stats in my profile and I can pretty much 'guesstimate which of the 3 "buckets" of the Elo spectrum I'm going to fall into.
So now, instead of smaller groups trying to sync drop who have ~20% chance of syncing. All you need to do is build one group of 4 then have everyone else around that can't quite build a 12 man take one of the unused weight classes and also guesstimate their Elo "bucket", and all try to sync.

I'm betting we'll get a lot closer to 6+ players on a team than we have a shot at doing now.

Well, solo sync dropping isn't usually tested, but it kinda-already works in the tourney environment (where winning is everything, usually generating high Elo). It would have to be tested, but it wouldn't be farfetched (yet harder to accomplish).

It may be easier to simply create 2 "equal sized" groups with the same weight classes used, and dropping together. It will guarantee people opposing each other, but it would require less effort though. It would only be sustainable as both sides win/lose at the same rate (50-50) because Elo will eventually shift you to a slightly better/worse average.

[Sandpit]

Triordinant, on 09 March 2014 - 02:29 PM, said:

If this is true, then it's another reason to have a premade-free, PUG-only queue.

go for it I say. I hope they do, then I can play how I want as can everyone else. It's just going to make things worse for that solo queue though. It will make a much more difficult NPE overall but at least those of us that group up will have better games

Deathlike, on 09 March 2014 - 02:31 PM, said:

Well, solo sync dropping isn't usually tested,

We do it every week in Marik. We call it marik monday madness and all join a TS channel and sync drop as solos. A lot of times we wind up in the same game. We may not be on the same team but we wind up in the same game. With CW we CAN'T drop on the other team

[Deathlike]

Sandpit, on 09 March 2014 - 02:34 PM, said:

We do it every week in Marik. We call it marik monday madness and all join a TS channel and sync drop as solos. A lot of times we wind up in the same game. We may not be on the same team but we wind up in the same game. With CW we CAN'T drop on the other team

I did actually state the exception to the rule... as I see more JHunch (Jman5) than say I do with mudhutwarrior (never seen him in any match).

So, I guess we can rig the MM some more, or can we have that awesomesauce 6-man premade queue?

[Roadbeer]

Deathlike, on 09 March 2014 - 02:31 PM, said:

Sure, and lots more ggclose to be had.

Exactly

Deathlike, on 09 March 2014 - 02:31 PM, said:

It may be easier to simply create 2 "equal sized" groups with the same weight classes used, and dropping together. It will guarantee people opposing each other, but it would require less effort though.

Which is fine if you are playing against your friends for bragging rights, but for playing WITH your friends, solo syncing with imposed weight classes AND only 3 Elo buckets... ggclose ensues.

Deathlike, on 09 March 2014 - 02:31 PM, said:

It would only be sustainable as both sides win/lose at the same rate (50-50) because Elo will eventually shift you to a slightly better/worse average.

Which would be a problem with several Elo buckets, but with only 3... it's kind of like the level restrictions in MMOs that prevent Plvling, and you're not in your friends bracket for only a short amount of time, then once you've reached the next 'bucket' it's business as usual.

Deathlike, on 09 March 2014 - 02:38 PM, said:

or can we have that awesomesauce 6-man premade queue?

Not sure if joking....

[Deathlike]

Roadbeer, on 09 March 2014 - 02:40 PM, said:

Which is fine if you are playing against your friends for bragging rights, but for playing WITH your friends, solo syncing with imposed weight classes AND only 3 Elo buckets... ggclose ensues.

You know, I'd still like to know my own Elo despite it being this ninja number.

Anyways, I don't think people mind fighting each other (it's hard to get a working sync drop as it is), but yes it is preferable to play together as much as possible.

Quote

Which would be a problem with several Elo buckets, but with only 3... it's kind of like the level restrictions in MMOs that prevent Plvling, and you're not in your friends bracket for only a short amount of time, then once you've reached the next 'bucket' it's business as usual.

Well, I doubt those Elo buckets are going to be fixed (they actually need to be floating values, adjusted by the state of the players waiting on the queues)... since averaging isn't done beyond the premade construction in the proposed MM.

You're probably right on that though... unless you are on the "edge" of the Elo boundaries... which normally is fixable by being a premade and winning often.

Roadbeer, on 09 March 2014 - 02:42 PM, said:

Not sure if joking....

I'm serious in wanting the queue, but not serious in that PGI isn't even thinking about it... ever.

No joy.

Edited by Deathlike, 09 March 2014 - 02:47 PM.

[Novakaine]

Roadbeer, on 09 March 2014 - 01:52 PM, said:

Its all part of the lead up to Sept. 2015 "Launch 2.0", after we muck around with "Launch 1.5", around Q4,2014

Don't ya mean Sept. 2017?

[Sandpit]

Basically, some of us, after reviewing the released data and "new" launch module, are trying to raise awareness and inform others (hopefully PGI) that the decision will not fix things like ggclose and roflstomps and was based on a poor interpretation of the data itself.

This is an example of PGI looking at the data, making a decision that completely contradicts the plan they've had for a year+, posting the data to show why they made the decision, and people who know how to break down and interpret the data a little better than whoever did it for PGI looked at it and went "Uhm, you're wrong and here's why"

It's not about attacking PGI, the data, or other players. It's about trying to explain WHY the decision they made was bad and how the "new" launch module is NOT going to solve the roflstomp or ggclose. There are several examples of why it's not going to help.

PGI's whole basis for this decision is based on improving new player retention and experience and mitigating roflstomps. Their decision simply isn't going to do that. The launch module is going to result in the exact same kind of patterns they do now, the exception being they can no longer blame premades for it and the total tonnages used. By the time (based on previous track records) PGI sees and acknowledges this and begins to actually fix it, I don't know that it won't have hurt the community past the point of this being a really successful game as opposed to a mediocre size population that people play while taking a break from other games they enjoy more. The organized groups will have moved on because there's not much playing if you can't participate in the campaign portion (CW) and with it a lot of money.

None of us who enjoy the game want that to happen.

[Artgathan]

Another point to add - I strongly believe the 3/3/3/3 matchmaking system will actually lead to more stomps (in the favor of premade teams). This belief is based on the following premise:

Assault > Heavy > Medium > Light

Essentially, if we assume that the 3 Assaults on a team represent ~35% of their firepower (while only representing 25% of the team), it becomes much easier for a pre-made lance to control a significant portion of their team's firepower. In the current system premades are powerful, but due to the fact that drop composition is essentially random (in terms of the mechs that appear in the match) the amount of firepower that the pre-made controls is variable. (IE: If you drop in a premade of 4 assaults now, and end up on a team of 8 assaults, the premade "controls" less of the team's firepower than they will if they drop 3 Assaults + 1 Heavy in the 3/3/3/3 system).

Furthermore the 3/3/3/3 system is really going to punish the "snowflake" players (the people who enjoy piloting Trebuchets, Dragons, Locusts...). Consider that currently if you player one of the mechs it's not a huge deal since (as above) the random team compositions will balance out. However, bringing a Locust in a 3/3/3/3 game means that you're giving up a Jenner while allowing your enemy to have 3 Jenners.

I really think this launch module will be bad for the game.

[Supersmacky]

Something I thought about earlier, and may have already been brought up, is modifying the 3/3/3/3 model to allow a little more flexibility. My that I mean min-max, assault/heavy/medium/light. So, it would be something like 2-4/2-4/2-4/2-4.

In other words, each side would have to have at least 2 mechs of a given class, but could have as many as 4. This would accomplish PGI's stated goals and would give players/4-mans more options. I believe it will also reduce the queue times when waiting for matches. Yes, you might face a team that has 4 assaults and 4 heavies, but you also might face a team that has 4 mediums and 4 lights. The idea does not limit players to having to work out exact weights.

Again, this may have been mentioned previously, but I believe it is a much better option that the planned 3/3/3/3 PGI has in mind. I believe their vision is driven by a solo-only model where you will just have a mass of individuals waiting for a drop and not and that within that group all of them have equally chosen to any given class of mech. Even if you take group drops out of the picture this is an erroneous view. At the least they will have to force people into mechs they might not really be interested in running or increase wait times terribly. It is a losing proposition to create a game and then put in artificial restrictions to make the players play the game a certain way. There should at least be the illusion of choice.

Just a thought...

[Agent 0 Fortune]

Roadbeer, on 09 March 2014 - 02:23 PM, said:

Afterthought...

I don't know if this was covered in the 70 pages of the Launch Module feedback, but has it occurred to anyone how incredibly easy it is going to be to game the MM now?

Just a cursory look at my mech stats in my profile and I can pretty much 'guesstimate which of the 3 "buckets" of the Elo spectrum I'm going to fall into.

So now, instead of smaller groups trying to sync drop who have ~20% chance of syncing. All you need to do is build one group of 4 then have everyone else around that can't quite build a 12 man take one of the unused weight classes and also guesstimate their Elo "bucket", and all try to sync.

I'm betting we'll get a lot closer to 6+ players on a team than we have a shot at doing now.

The exceptions being

• You will face opponents of more similar skill level (hard Elo boundaries)
  
• You can’t do it in 6+ assaults
  
• You sync dropping will eventually move you to the highest Elo bracket, away from the newer and more casual players.

I don’t want to put words in your mouth, but I get the feeling you would prefer to have a queue with unlimited groups size and no Elo boundaries, and force PUGs to fillout the rest of your team (likely PUGs who are significantly less skill so that the matchmaker can level out the Elo rating of your team). Which is the exact issue this Launch Module seeks to remedy.

[Sandpit]

Agent 0 Fortune, on 10 March 2014 - 08:42 AM, said:

The exceptions being

• You will face opponents of more similar skill level (hard Elo boundaries)
  
• You can’t do it in 6+ assaults
  
• You sync dropping will eventually move you to the highest Elo bracket, away from the newer and more casual players.

1) That's the same system we have now. If MM doesn't find an opponent in that range it expands until it does

2) The weight class doesn't exclude individual skill. I can take any weight class and be competitive.

3) Sync dropping is probably the single biggest myth/boogeyman I've seen in this game. Most of the time when we sync drop (i'd guess about 80-90% but I can start tracking it) it's a 2man and 3man not two 4man groups. The reason we're synch dropping is because we can't drop with that 5th guy so we break up our 4man to create to groups so our buddies aren't having to drop solo. Not to mention some of us are now completely against dropping solo because that would just show "more people drop solo now" which is completely false in this regard

You have continuously just tried to represent the ones not agreeing with the stats as people who want nothing more than to prey on other players and that's flat out false and untrue. The only one who seems to have an agenda in this regard isn't me or road....