Forged Alliance Forever

Re: Improving ladder through map selection

2016-07-12T09:38:36+02:00

Hawkei wrote:
Here in lies the problem... The idea that an such map is 'shitty' is highly subjective and based in opinion. I don't think any kind democratic process will work. Because, players don't vote for maps which are good. They vote for maps which they are comfortable playing, and which will inflate their ranking. Your suggestion is based in the flawed premise that we ought to be making players comfortable.

I don't think that we should be seeking to limit the diversity of play style in ladder - and I don't think players being comfortable is an appropriate objective. The whole idea is to push players outside of their comfort level, and adapt to the unknown.

This is why you do a 1-3 map pool and rotate daily, with no map showing up twice within say a week, all problems solved.

Hawkei wrote:
Having said this, there are some maps which are bad. Either because they promote predictable or stale gameplay. Or because there is an over-riding meta-game which when executed precisely will determine victory. Or because they are just too limited in their tactical diversity. Or because they are imbalanced and obviously favour one player over the other.

Congratulations, you just listed 99% of all maps. My hat is off to you sir!

Statistics: Posted by BushMaster — 12 Jul 2016, 09:38

Re: Improving ladder through map selection

2016-07-12T06:57:39+02:00

@KeyBlue: I used your explanation in the OP. Hope you don't mind?

Statistics: Posted by RealityCheck — 12 Jul 2016, 06:57

Re: Improving ladder through map selection

2016-07-12T05:32:36+02:00

biass wrote:
Scenario: a new map gets added to the pool, noone has played the map before, thus, the average rating of everyone drops as everyone has 0 w/l whatever

Yes, that's valid scenario. I think that the average rating should only be calculated using the maps that are already played by the player. The other maps are "unknown". Can't use unknown quantities in calculations. Using 0 instead of the unknown quantity won't be good either.

On the issue of map rating vs map win %. I actually think that keeping the rating system may be easier to implement. Because it is the same system as now, just applied to each map/map group separately. But then, what is easier to implement depends on how things are actually coded currently. Only devs know this.
I also thought that map ratings were easier to reason about, because everybody knows how current rating system works.
Win % probably can work, but what if I win against player who has -200 rating and against player who has +200 rating? Both wins are not equally valuable.

Astrofoo wrote:
My main fear of a system like this however, is that it's very ambitious and a little complex for our little FAF.

Yes. I felt very confused at some point trying to think about all the things that could go wrong. I hope, however, that we can arrive at a set of rules that are simple and clear. The main thing that gives me hope is that the change is maybe not so big - instead of tracking a single rating, track rating for each map/map group. As a result we just give the matchmaker more information, which cannot be a bad thing.

Statistics: Posted by RealityCheck — 12 Jul 2016, 05:32

Re: Improving ladder through map selection

2016-07-12T04:39:07+02:00

or because they look like ass cancer

Statistics: Posted by biass — 12 Jul 2016, 04:39

Re: Improving ladder through map selection

2016-07-12T04:39:26+02:00

LichKing2033 wrote:
You all propose over complicated solutions. Just remove shitty maps from the pool and problem solved.
...As for the map pool... it can be decided by a poll. Example is below (list is from my post on how to encourage map diversity)

Here in lies the problem... The idea that an such map is 'shitty' is highly subjective and based in opinion. I don't think any kind democratic process will work. Because, players don't vote for maps which are good. They vote for maps which they are comfortable playing, and which will inflate their ranking. Your suggestion is based in the flawed premise that we ought to be making players comfortable.

I don't think that we should be seeking to limit the diversity of play style in ladder - and I don't think players being comfortable is an appropriate objective. The whole idea is to push players outside of their comfort level, and adapt to the unknown.

Having said this, there are some maps which are bad. Either because they promote predictable or stale gameplay. Or because there is an over-riding meta-game which when executed precisely will determine victory. Or because they are just too limited in their tactical diversity. Or because they are imbalanced and obviously favour one player over the other.

Statistics: Posted by Hawkei — 12 Jul 2016, 04:37

Re: Improving ladder through map selection

2016-07-12T03:46:12+02:00

had the idea of making a map QA team but idk

Statistics: Posted by biass — 12 Jul 2016, 03:46

Re: Improving ladder through map selection

2016-07-12T03:39:04+02:00

You all propose over complicated solutions. Just remove shitty maps from the pool and problem solved.

Spoiler: show

I'd say that your matchup should be decided by this inequality:
x-300 < x < x+300
Where x is you, the other two are the max and min rating you can be matched up against. Say I me, ~430 ladder. Maximum I can matched up is vs. 730, min 130. Both of those are fun to play. Just today, I played vs. desernator, very interesting games. Since his rating is ~710, I can learn from him because he is not too far up.

As for the map pool... it can be decided by a poll. Example is below (list is from my post on how to encourage map diversity)

Spoiler: show

Statistics: Posted by Lieutenant Lich — 12 Jul 2016, 03:39

Re: Improving ladder through map selection

2016-07-12T02:12:47+02:00

Astrofoo's opinions wrote:
Whoever disagrees with me is wrong.

Go back to the balance forums mate
It's very easy to learn from a beatdown if you know how to teach yourself properly

Scenario: a new map gets added to the pool, noone has played the map before, thus, the average rating of everyone drops as everyone has 0 w/l whatever

who has to take on the shitstorm?

Statistics: Posted by biass — 12 Jul 2016, 02:12

Re: Improving ladder through map selection

2016-07-12T00:01:28+02:00

Don't listen to Mad Mozart. He is confused and doesn't understand the proposal. Also, his statement about bad players only playing other bad players and always sucking is incorrect. In a ladder game there are two players. One will win, one won't. The one who wins may think he doesn't need to change something, so maybe he doesn't improve. However, for every winner there is always at least one loser. The human brain is DESIGNED for us to learn from our mistakes. Therefore it is likely that the loser will learn something from his loss. For Example: "Wow those t2 gunships got me good! I should make flak next time, or maybe try to make some inties..."
This is how ladder and competition and in fact the actual meta game works.

The map rating thing however I actually agree with Mozart on. Use win % instead. Easier. The idea of a very good Theta player playing another very good Theta player while being average on Roanoke and playing against another average player on Roanoke explains this very well. I really like that concept. Let's be real, no-one learns from a massive beatdown. Too much went wrong. We learn infinitely more from close games. Whoever disagrees with me is wrong. Plain and simple. (And no I'm not referring to your first 15 games where you don't even know how to land spam, obviously a big beat down in that situation can help you a little bit..) My main fear of a system like this however, is that it's very ambitious and a little complex for our little FAF.

Statistics: Posted by Astrofoo — 12 Jul 2016, 00:01

Re: Improving ladder through map selection

2016-07-11T23:15:53+02:00

Mad`Mozart wrote:
KeyBlue wrote:
Mad`Mozart wrote:Isn't this how current system works?
500 plays against 300-700
They both suck, according to your standards. And since they both suck they can't learn anything from that so they keep sucking.
So following your logic: it is impossible for a 500 (after 100 games) to become a 2000 or even 1000.

PS: damn i need to get a life.

Started reading from bottom and pretty much stopped after reading this.
Not even going to bother to argue

Which part?
My bad attempt at making a joke about the length of my posts?
Or
The statement about people sucking? (which took your comment to the extreme in an attempt to provoke an explanation, but that failed i guess)

I would like it if you read the rest of the post aswell though. If only for my need to be heard or the opportunity to fix my flawed ideas.
(and the fact that i basically agreed with you in most of it.)

Statistics: Posted by KeyBlue — 11 Jul 2016, 23:15

Re: Improving ladder through map selection

2016-07-11T18:19:38+02:00

Mad`Mozart wrote:
Started reading from bottom and pretty much stopped after reading this.
Not even going to bother to argue

But what is wrong with KeyBlue's words? When two weak players are matched they won't suck equally. One of them will suck less. So the other one will lose and will learn something. It is the same with current system.

Statistics: Posted by RealityCheck — 11 Jul 2016, 18:19

Re: Improving ladder through map selection

2016-07-11T18:06:12+02:00

KeyBlue wrote:
Mad`Mozart wrote:Tell me more pls. When you want to match someone who suck vs other guy who suck as well. They will simply never learn too.

Isn't this how current system works?
500 plays against 300-700
They both suck, according to your standards. And since they both suck they can't learn anything from that so they keep sucking.
So following your logic: it is impossible for a 500 (after 100 games) to become a 2000 or even 1000.

PS: damn i need to get a life.

Started reading from bottom and pretty much stopped after reading this.
Not even going to bother to argue

Statistics: Posted by Mad`Mozart — 11 Jul 2016, 18:06

Re: Improving ladder through map selection

2016-07-11T15:35:18+02:00

Mad`Mozart wrote:
You guys want to create an obviously biased system. I guess you havent played a ladder, where out of 15 selected maps almost every game is cycling between 5 of those and another 2 you get to play once 2 to 6 months (not even exaggerating here). This is where maps you like become maps you hate and this is what will happen if map selection will favor lowest rated player in a match up. Not even saying how this can be abused.

I dont see why do you want to make a competitive environment favoring anyone or anything. It kills the competition the moment someone gets a better start.

That was mostly related to previous posts, now lets talk more about map rating.

First we don't want anything. The goal was a theoretical discussion about RealityChecks idea of a possible improvement, see first post.

But you bring up a good point. Since you are one of the few at the top, you will most likely have to play maps you're not so good on, to balance you out against a lower rated player. And you will barely be able to play your favorite maps, since you perform well on them and can't get matched with anyone on those maps because of your superior map rating.
Which will result in you having less fun, i suppose. (I'm at the middle of the pack, so i didn't consider this situation for myself)

And the competitive environment will be shifted to map rating.
Average rating will become failry meaningless, because of the biased general matchmaking, like mozard *explained*.
But we can have competitive play on map level. Where you strive to be best player of a certain map.
So there is still space for a fair competitive environment inside the suggested system. It is just split up.

I can see that this isn't appealing, since being 5th best on theta isn't as impressive as being 5th best overal and it doesn't say anything about your overall skill. (A category system could make this a little more interesting, but i doubt it.)

Mad`Mozart wrote:
KeyBlue wrote:

Basically, everything you just said about map rating i can put to win % and it will make perfect sense.

If you have 1600 rating, you will usually get matched vs 1400 to 1800 players, right? (i fail to see here how you can get matched 100 games vs newbies) So if you are good on theta, you will win there a lot and your win % will go up and opposite for roanoke. So inside your 1600 ladder rating, you have big win % on theta and low win % on roanoke. Then using your idea, you can get matched on theta vs 1800 players and on roanoke vs 1600. Ofc, the more games you played, the harder it will be for the stat to change so yeah, there's 1 downside to it after all. But ultimately, its way more simple to implement since we can already draw the stat from existing games rather than coding and storing more new stats for a player (which will happen exactly never )

Quite funny how you say that you doubt that you can only play against a noob on a certain map, while first telling me that there are several maps you barely play. As if playing those few games on those barely played maps couldn't be coincidentally against newbs.

I didn't consider that you would use the standard rating together with win percentage. This solves the issues I mentioned about map rating having more info, because this extra information is covered by standard rating.
Your suggestion is actually quite interesting and could replace or be used in conjunction with the map rating idea.

Mad`Mozart wrote:
I can see the point that you want to make ladder matches fair (for a lot of low rated players haha) which is wrong. Its a competition and everyone should play on equal conditions. Dont try to make game easier for you if you suck.

The main idea was to make it easier for people to get matched with bigger rating differences due the certain weaknesses and strengths.
And thus make each game as competitive as possible, which results in more fun games, since the general consensus is that competitive game = fun game.
So no intended cheating. If it is possible with the suggested system, its good that you pointed it out. Now we just need to figure out how to remove that cheating possibility.

Mad`Mozart wrote:
Tell me more pls. When you want to match someone who suck vs other guy who suck as well. They will simply never learn too.

Isn't this how current system works?
500 plays against 300-700
They both suck, according to your standards. And since they both suck they can't learn anything from that so they keep sucking.
So following your logic: it is impossible for a 500 (after 100 games) to become a 2000 or even 1000.

PS: damn i need to get a life.

Statistics: Posted by KeyBlue — 11 Jul 2016, 15:35

Re: Improving ladder through map selection

2016-07-11T14:41:01+02:00

RealityCheck wrote:
The game won't become easier, just more balanced. Everyone is still going to have about 50 % winrate on average.

Tell me more pls. When you want to match someone who suck vs other guy who suck as well. They will simply never learn too.

Statistics: Posted by Mad`Mozart — 11 Jul 2016, 14:41

Re: Improving ladder through map selection

2016-07-11T14:29:29+02:00

Mad`Mozart wrote:
Dont try to make game easier for you if you suck.

The game won't become easier, just more balanced. Everyone is still going to have about 50 % winrate on average.

Statistics: Posted by RealityCheck — 11 Jul 2016, 14:29