76 replies to this topic

[Elizander]

MechaGnome, on 10 April 2021 - 05:37 PM, said:

Math=/=Reality..

No wonder the game industry is like it is..

Math is helpful, but designing an entire game around engagement statistics like Blizzard did for their investor reports made their entire bubble burst in a spectacular manner. After heavy losses, they completely caved and are doing every single ignored request their players made since the last expansion launched.

Now they are desperately cross-promoting their games to make one player count as multiple active users (+1 each extra Blizzard game a player plays) to make up for it in their next report.

Edited by Elizander, 03 September 2021 - 08:29 PM.

[w0qj]

I personally feel that MWO can do game cross-promotion too!

Give MW5 players a few free Special mechs when they play MWO also!
Give MWO players a few free Special mechs when they play MW5 also!

Heck, if they Give MW5 players a few free Special mechs, I might buy it just for that!

(Caveat: MWO Special mechs if one also buys the top tier MW5 DLC Expansion Packs)

Elizander, on 03 September 2021 - 08:27 PM, said:

Now they are desperately cross-promoting their games to make one player count as multiple active users (+1 each extra Blizzard game a player plays) to make up for it in their next report.

Edited by w0qj, 03 September 2021 - 09:00 PM.

[Nightbird]

Elizander, on 03 September 2021 - 08:27 PM, said:

Math is helpful, but designing an entire game around engagement statistics like Blizzard did for their investor reports made their entire bubble burst in a spectacular manner. After heavy losses, they completely caved and are doing every single ignored request their players made since the last expansion launched.

Now they are desperately cross-promoting their games to make one player count as multiple active users (+1 each extra Blizzard game a player plays) to make up for it in their next report.

Absolutely, game design should use focus groups based around your target audience. Only simple things like match making, logistics (server infrastructure), and marketing should use statistics to make sure you optimize the user experience. It is completely opaque to your audience.

It's like how the Battle Rank in COD and Fortnite is just an XP bar, and has nothing to do with match making. For match making, they use a Bayesian model which is the best of both worlds. It uses a long term average to determine your skill but your performance that day will weigh more allowing for quick adjustment if you're having a good or bad day. Your first match of the day actually gets a minus adjustment to allow you to warm up. It's all very complicated and not explained to the player, but it works.

Edited by Nightbird, 03 September 2021 - 08:57 PM.

[MrMadguy]

My WLR has always floated around 1, but I was extremely unhappy prior to current PSR system. So your theory is wrong. Current PSR is the best, because when I perform badly - I always know, that every loss and stomp I suffer from will eventually bring me back to Tier 4, where quality of game is much higher. The biggest offender for me - is when I'm used as cannon fodder to pad Tier 1 matches. My skill is so low in comparison to theirs, that I can't affect results of matches and therefore my WLR depends fully on their performance, not mine. If we would have WLR MM, I would stuck in such situation forever.

[Nightbird]

MrMadguy, on 03 September 2021 - 09:48 PM, said:

My WLR has always floated around 1, but I was extremely unhappy prior to current PSR system. So your theory is wrong. Current PSR is the best, because when I perform badly - I always know, that every loss and stomp I suffer from will eventually bring me back to Tier 4, where quality of game is much higher. The biggest offender for me - is when I'm used as cannon fodder to pad Tier 1 matches. My skill is so low in comparison to theirs, that I can't affect results of matches and therefore my WLR depends fully on their performance, not mine. If we would have WLR MM, I would stuck in such situation forever.

The fact you feel it is working is only because your WLR is around 1. As the graph shows, the MM stops working when you are your WLR is low or high.

[Capt Deadpool]

Nightbird, on 03 September 2021 - 08:53 PM, said:

Absolutely, game design should use focus groups based around your target audience. Only simple things like match making, logistics (server infrastructure), and marketing should use statistics to make sure you optimize the user experience. It is completely opaque to your audience.

It's like how the Battle Rank in COD and Fortnite is just an XP bar, and has nothing to do with match making. For match making, they use a Bayesian model which is the best of both worlds. It uses a long term average to determine your skill but your performance that day will weigh more allowing for quick adjustment if you're having a good or bad day. Your first match of the day actually gets a minus adjustment to allow you to warm up. It's all very complicated and not explained to the player, but it works.

It's almost like effective MM systems have already been pioneered and we don't need to reinvent the wheel or anything! XD

[Elizander]

MrMadguy, on 03 September 2021 - 09:48 PM, said:

My WLR has always floated around 1, but I was extremely unhappy prior to current PSR system. So your theory is wrong. Current PSR is the best, because when I perform badly - I always know, that every loss and stomp I suffer from will eventually bring me back to Tier 4, where quality of game is much higher. The biggest offender for me - is when I'm used as cannon fodder to pad Tier 1 matches. My skill is so low in comparison to theirs, that I can't affect results of matches and therefore my WLR depends fully on their performance, not mine. If we would have WLR MM, I would stuck in such situation forever.

This is a valid opinion and there are many people here that will tell you that the MM is supposed to place you in the tier where your playstyle works best. There is absolutely nothing with enjoying your time in T5/T4 and not liking T3+.

Nightbird, on 03 September 2021 - 08:53 PM, said:

It's like how the Battle Rank in COD and Fortnite is just an XP bar, and has nothing to do with match making. For match making, they use a Bayesian model which is the best of both worlds. It uses a long term average to determine your skill but your performance that day will weigh more allowing for quick adjustment if you're having a good or bad day. Your first match of the day actually gets a minus adjustment to allow you to warm up. It's all very complicated and not explained to the player, but it works.

I really wish our MM at least took into account which mechs I suck at and which I am good at and not just assuming my skill level is flat across the board over all the 100+ mechs I own.

If I wanna derp around in my Ultra AC20 Urbie that moves at 54kph then put me down where I belong.

[Nightbird]

Elizander, on 04 September 2021 - 12:16 AM, said:

This is a valid opinion and there are many people here that will tell you that the MM is supposed to place you in the tier where your playstyle works best. There is absolutely nothing with enjoying your time in T5/T4 and not liking T3+.

It's not a valid opinion because a WLR match maker would not put him against the best players in the game (unless the queue is very short, same as it does today). An WLR match maker does nothing more than place people more accurately than PSR, therefore reducing stomps.

Edited by Nightbird, 07 September 2021 - 10:09 AM.

[Paul Meyers DEST]

Nightbird, on 03 September 2021 - 08:53 PM, said:

It's like how the Battle Rank in COD and Fortnite is just an XP bar, and has nothing to do with match making. For match making, they use a Bayesian model which is the best of both worlds. It uses a long term average to determine your skill but your performance that day will weigh more allowing for quick adjustment if you're having a good or bad day. Your first match of the day actually gets a minus adjustment to allow you to warm up. It's all very complicated and not explained to the player, but it works.

That Bayesian Model is pretty much what i suggested some hours ago in ASHcord. A recent W/L of 3-5 games count in with a historic W/L ratio, with the most recent weighting more than the historic.
That would recognize:
- playing with a strong group (as i sometimes do)
- playing with a weaker group (as i sometimes do)
- playing solo in my favorite mech
- playing in a weak or skilling an unskilled mech
- switching between all above
- and your mentioned having a good or bad day


I am pretty sure that you can ad even more reasons.

Edited by Paul Meyers DEST, 03 November 2021 - 12:16 PM.

[Nightbird]

The Bayesian Model is far more difficult to implement right. The WLR "frequentist" method is like shooting at a target at 10 meters, even if you've never shot a gun you can bullseye easily. The Bayesian is shooting at a target a mile away, you can not only miss the target completely but also break your nose doing it. I don't trust PGI or the community to do Bayesian.

Edited by Nightbird, 03 November 2021 - 02:02 PM.

[Paul Meyers DEST]

So you agree that my idea is really decent?

[Nightbird]

Paul Meyers DEST, on 03 November 2021 - 02:35 PM, said:

So you agree that my idea is really decent?

Results is everything, I think the idea works and said as much, but I think if you try to implement it the result will be a disaster.

[Kotis77]

Can we just go with soft tier reset that other FPS shooter does? At least those handful shooters that i have played uses soft tier resets every season. Meaning you will be dropped 2-3 tiers down every season. Season lasts about 2-3 months. it seems to be working at least on those games if you ask me, but im no expert. Its not perfect but nothing is.

Im not sure why, but are those AAA studios lazy or they just know more than us do? Is perfect matchmaker just too difficult to do, or is just easier to drop those from tier 1 who doesnt belong there?

Could that just work in MWO too? Or am i missing something here?

Edited by Kotis77, 06 November 2021 - 08:48 AM.

[Nesutizale]

What would dropping people a few tiers do for the game? Beside player population beeing so low that, except for the extrems, you see nearly every tier in your match allready. So dropping the top tiers down...wouldn't that just mix skill levels to a point where its not neseccary to have tiers at all? Kinda confused by this idea.

As for createing the perfect MM. That might not be that problematic. I think its more that how the skill level is destributed.
You have a lot of people at the middleground and very little at the extrems. Now when one of the extrems enters a match he could be so over the top or so bad that he must be balanced out but you might not have someone good enough in the pool. So the MM drops in the next best player. At that point you have thrown balance out of the window.

I would say MWO is at a point where, even if a perfect MM is implemented, the playerbase is to thin that even that system would be able to make perfect matches. Except for extreme long waiting times to made a thing again.
That is where a mathematical perfect system and reality clash

Edited by Nesutizale, 06 November 2021 - 09:03 AM.

[Kotis77]

Nesutizale, on 06 November 2021 - 09:01 AM, said:

What would dropping people a few tiers do for the game? Beside player population beeing so low that, except for the extrems, you see nearly every tier in your match allready. So dropping the top tiers down...wouldn't that just mix skill levels to a point where its not neseccary to have tiers at all? Kinda confused by this idea.

To improve soup que quality.Real tier 1 player rises to tier1 in days. Fake tier 1 rises to tier 1 in many months. I gotta belive that in primetime there is enough players to make one good lobby with tier 1 players? I might be wrong too about this. But that i know that there is so many guys in tier 1 that doesnt belong there. And i think that is messing game quality. Groups in QP is messing up quality big time, but is PGI willing to go back to soloque?

Dont these fake tier1 guys wanna good games too?

Just trying to find ways to improve game quality here man...

Edit* sounds elitist, but my POV is from tier1 matches. Havent really played other tiers, But should help other tiers too. Last 6-7 years have played QP mostly solo, but last 2-3 months have been so garbage in QP, so i dont wanna drop solo anymore. Too much 12-0 and 12-1 matches, cant affect match outcome anymore by yourself.

Edited by Kotis77, 06 November 2021 - 01:12 PM.

[Kotis77]

Nesutizale, on 06 November 2021 - 09:01 AM, said:

I would say MWO is at a point where, even if a perfect MM is implemented, the playerbase is to thin that even that system would be able to make perfect matches. Except for extreme long waiting times to made a thing again.
That is where a mathematical perfect system and reality clash

Is playerbase worse than year ago? I seem to remeber more quality matches year ago...

[caravann]

You know what is the common?
Skills and skill points. The competition is irrelevant until both sides has a common ground.

Defenses shouldn't be a skill. You can not skill the armor plate to make it better by training the armor plate.

But that's neat you do in mwo. Then it is about playerbase and groups. Some even play as a distraction and do not care about the performance.

But there's the thing about maintenance; those who stay are boring. The main part about a live game is that people tossing money on the screen and feel good about sharing their tossing money on the screen experience. They have a common ground that those who stay do not share. And why should they stay? They have already explored all content, all that remains is the tossing the money on the screen mini game. Next metagaming coming through. If it really was about friends playing a game it wouldn't be the case that the popularity of the game is based on the meta-gamers being fed with content which we call catering the masses.