536 replies to this topic

[Almond Brown]

Craig Steele, on 29 April 2014 - 10:12 PM, said:

So given the rest of your post has pretty sound grammar, I am going to assume you have correctly used "an" and not "a" meaning the missing word starts with a vowel.

So that's an A, E, I, O or U.....

Hangman?

Pick any Vowel and we will fix you up.

[tayhimself]

Is there a tl;dr for the thread? I couldn't even get through the first essay... err post

[Solahma]

Page 16 sums things up pretty well.

[Jun Watarase]

tayhimself, on 01 May 2014 - 10:13 AM, said:

Is there a tl;dr for the thread? I couldn't even get through the first essay... err post

Uh, the picture wasnt conise enough?

[AlmightyAeng]

Jun Watarase, on 30 April 2014 - 02:34 AM, said:

So what this basically boils down to is a small group of players ruining the fun for everyone else because winning is very very important to them. Nobody gets to enjoy a close game when the match is stacked in the favor of one side off the bat because they pulled off a 8 man sync drop full of meta builds and the matchmaker decides to hey, lets put multiple newbies as their opponents because i cant find anyone else thanks to the sync drop.

Just a quick question from back on page 16...how is a "small group of players" ruining it for "everyone" by dropping 8-man sync drops?

Last time I checked, 8 people out of 24 is a full 33%...is that a small group? We're told by PGI that only 16% of the playerbase drops in groups, and of those, not all of them drop in 4-mans...

So, is it more likely that PGI is wrong about the number of groups playing, that Jun is consistently dropping with sync drops every game (and so is everyone else) or Jun is just bad and doesn't want to face the fact that losing is his own fault?

Edited by Ghost Badger, 01 May 2014 - 11:52 AM.

[Hou]

Comic's pretty good. Haven't played in a while. Fired up the NoGutsNoGalaxy twitch stream and decided to check in on here today to see if the game had changed much in the way of being able to hop in solo for a couple matches between stuff going on at home/the kids going to bed. Looks like it hasn't really. Looks like I'll stay signed off. Happy stompy stompy to those of you who have stayed!

Edited by Hou, 01 May 2014 - 12:05 PM.

[Athomahawk]

Bagheera, on 28 April 2014 - 09:34 AM, said:

From the guy who spent most of closed beta pretending to be a disconnected Atlas sitting on base, letting his team die, and hoping he can take out the damaged opponents. Lol.

Dont forget writing those original "guides" where he essentially reposted info straight from sarna.

OH WELL

[CCC Dober]

Guys, I've looked up the 'news' again where the 84% figure came from, but see for yourself:

http://mwomercs.com/...28#entry3185728

I don't know if one needs to have studied statistics to guess what went wrong here, but I will try to explain nevertheless.

Spoiler

First of all, take a look at the diagram inside the spoiler. It looks like each active player contributes to total launches. Now, what about grouped up players? Does a group launch mean that every group member is accounted for individually or only the group itself? How are disconnects treated in this regard? None of that is evident, but the differences can have a significant impact on the numbers we see here. To give an example: imagine a group of 4 players launches, but only one launch is registered in total. Whereas if each member would drop solo, a total of 4 launches would be registered. Do you see how such an easy mistake could lower the total group launches?

Secondly, each player can contribute to each category listed there (solo, 2-, 3-, 4-man groups). That is neither mentioned nor taken into account. Why is that important you ask? Some players play more often than others, so they contribute more to the number of launches in a or more categories. Also, not everybody pugs or groups up exclusively. Furthermore, not every group can build up or maintain the maximum number of players over the course of a session. By looking at total launches alone, you would never know if a minority or a majority of players has an impact on a diagram category.

All of the above hints at significant overlap in the data presented to us. To draw any meaningful conclusions, it really needs to be cleaned up. PGI first needs to make sure they have accounted for all players in groups correctly. That means every player counts as a launch, not just the group itself. Secondly, they need to link each launch to a player ID so they can feed a normal distribution and see how people fit into the bigger picture, how regularly they form groups, how big they are etc. Launches alone don't tell the whole story.

So, what do you think? Should PGI take another look at their data?

Edited by CCC Dober, 01 May 2014 - 01:54 PM.

[Jun Watarase]

Hou, on 01 May 2014 - 12:03 PM, said:

Comic's pretty good. Haven't played in a while. Fired up the NoGutsNoGalaxy twitch stream and decided to check in on here today to see if the game had changed much in the way of being able to hop in solo for a couple matches between stuff going on at home/the kids going to bed. Looks like it hasn't really. Looks like I'll stay signed off. Happy stompy stompy to those of you who have stayed!

You might want to check back after they fix the matchmaker, before the hotfix, matches were much more even than they are now. Not sure what it will be like when everything is fixed though.

[Jun Watarase]

Quote

So, what do you think? Should PGI take another look at their data?

But that would involve spending time on it...which we know they dont want to do.

Im guessing they did the simple thing and took total numbers of matches played over a time period and got X solo players for each match and Y group players for each match.

So if they had 1000 matches played, they got 840 solo players and the other 160 launched in some kind of group for that 1000 matches.

That doesnt help much because with 24 players per match, and at 16% players being premades, statistically wise on average about 4 players per match are going to be in a group of some kind. Which anyone can tell you.

[Craig Steele]

CCC Dober, on 01 May 2014 - 01:53 PM, said:

Guys, I've looked up the 'news' again where the 84% figure came from, but see for yourself:

http://mwomercs.com/...28#entry3185728

I don't know if one needs to have studied statistics to guess what went wrong here, but I will try to explain nevertheless.

Spoiler

First of all, take a look at the diagram inside the spoiler. It looks like each active player contributes to total launches. Now, what about grouped up players? Does a group launch mean that every group member is accounted for individually or only the group itself? How are disconnects treated in this regard? None of that is evident, but the differences can have a significant impact on the numbers we see here. To give an example: imagine a group of 4 players launches, but only one launch is registered in total. Whereas if each member would drop solo, a total of 4 launches would be registered. Do you see how such an easy mistake could lower the total group launches?

Secondly, each player can contribute to each category listed there (solo, 2-, 3-, 4-man groups). That is neither mentioned nor taken into account. Why is that important you ask? Some players play more often than others, so they contribute more to the number of launches in a or more categories. Also, not everybody pugs or groups up exclusively. Furthermore, not every group can build up or maintain the maximum number of players over the course of a session. By looking at total launches alone, you would never know if a minority or a majority of players has an impact on a diagram category.

All of the above hints at significant overlap in the data presented to us. To draw any meaningful conclusions, it really needs to be cleaned up. PGI first needs to make sure they have accounted for all players in groups correctly. That means every player counts as a launch, not just the group itself. Secondly, they need to link each launch to a player ID so they can feed a normal distribution and see how people fit into the bigger picture, how regularly they form groups, how big they are etc. Launches alone don't tell the whole story.

So, what do you think? Should PGI take another look at their data?

For me, the key part of this Source Post is under "Group Size".

That quite clearly portrays the percentages as a measure of activity, not demographic. And that was the first impression I got when the "stats" were first raised in Podcast 103.

Headlines are headlines, but the context is the most important thing for me anyway.

[CCC Dober]

Craig Steele, on 01 May 2014 - 05:10 PM, said:

For me, the key part of this Source Post is under "Group Size".

That quite clearly portrays the percentages as a measure of activity, not demographic. And that was the first impression I got when the "stats" were first raised in Podcast 103.

Headlines are headlines, but the context is the most important thing for me anyway.

The context I see here is that the diagram was used to 'discuss' groups or rather how PGI was surprised by the results they got. Do I think it was the wrong kind of diagram to base the discussion on? Oh you betcha.

Total launches to me are like high water marks: nice to know, but not sufficient to aid in this discussion. Plus, it's easy to mess these numbers up, even by accident. So, how about taking the data to answer more interesting questions instead? How many groups participate in a game on each side. How many members do the groups have in such games? How often do group imbalances occur? How do they influence the results? Wouldn't that be much more interesting, especially in the light of sync drops or 12-0 stomps?

PGI seems to have some data (maybe not all of it), but the way it's used and analyzed is far from optimal. Optimal as in answering some questions regarding the positive and negative impact of groups on gameplay i.e. sync drops due to artificial size limits.

In short: a proper statistical analysis could help to solve problems, if somebody at PGI started to ask the right questions. The hints are all there: sync drops, 12-X stomps, people getting fed up over it and leaving. All of that leaves traces in the data and looking for these patterns is the first step to begin and formulate a solution. I don't see these steps being taken at all, quite the opposite.

[Craig Steele]

CCC Dober, on 02 May 2014 - 04:15 AM, said:

The context I see here is that the diagram was used to 'discuss' groups or rather how PGI was surprised by the results they got. Do I think it was the wrong kind of diagram to base the discussion on? Oh you betcha.

Total launches to me are like high water marks: nice to know, but not sufficient to aid in this discussion. Plus, it's easy to mess these numbers up, even by accident. So, how about taking the data to answer more interesting questions instead? How many groups participate in a game on each side. How many members do the groups have in such games? How often do group imbalances occur? How do they influence the results? Wouldn't that be much more interesting, especially in the light of sync drops or 12-0 stomps?

PGI seems to have some data (maybe not all of it), but the way it's used and analyzed is far from optimal. Optimal as in answering some questions regarding the positive and negative impact of groups on gameplay i.e. sync drops due to artificial size limits.

In short: a proper statistical analysis could help to solve problems, if somebody at PGI started to ask the right questions. The hints are all there: sync drops, 12-X stomps, people getting fed up over it and leaving. All of that leaves traces in the data and looking for these patterns is the first step to begin and formulate a solution. I don't see these steps being taken at all, quite the opposite.

Well sure, but how many companies are in the habit of asking the general public to analyse their corporate knowledge and provide direction on how the business is to be run?

I mean does what you're asking for even sound remotely commercial?

We are consumers, not shareholders, not management. Customers. Our decision is whether to buy or not buy the product. I am sure PGI are very happy to receive feedback from their customers and consider their offering in that light.

If anyone thinks they can do a better job, they can always start up in opposition I suppose. But as consumers it's not our perogative to manage the company is it?

[CCC Dober]

Well, you have managed to answer your own question. Sort of.

Unfair matches (due to flawed group implementation) lead to unsatisfied players and reduced income by extension. If PGI does not to tackle this problem correctly, because their statistical analysis is misleading, then there is a very real danger of losing touch with the community. Therefore it is vital that PGI improves their understanding of data and statistics, especially if they are used to base decisions on.

Is it worth the effort? Financially speaking, I would say so. The alternative for PGI is to kiss goodbye more revenue each time Paul tinkers around with parameters/numbers. Just add up the loss of revenue during and after each LRMgeddon. I'm sure it's more than enough to employ somebody, who is well equipped to deal statistics and maths in general.

[Greyboots]

This is easy to summarise.

1) Newbies don't like getting roflstomped.
2) The MM is happy to build teams to get roflstomped because if it doesn't drops take too long and people complain.

You can find this out through any number of websites so most who research a game before they install it, instead of installing it on a whim, won't install it in the first place. However, PGI think they're doing good because their player retention is doing ok. So the summary here is pretty much "Get used to it because it's not going to change any time in the near future".

Still, I'll add something here. The most common comment I get when I talk to people about MWO? "Yeah, I played that for a little while".

The people I know just aren't going to keep playing a game that doesn't work right. The matchmaker is no good at creating balance, weapons don't hit like they should, the Mechlab is now an unadulterated mess (although it DOES give you more control and info than the old one), you get stuck in endless "connecting" loops, the memory leak prevents the game from shutting down occasionally, it still crashes, they can't get balance right, many mechs are entirely redundant but you STILL have to buy 3 to train the ONE that works, and so on.

If I wasn't a hardcore Battletech/Mechwarrior fan? I probably wouldn't be putting up with it either. the game is fun but it's not that much fun.

Edited by Greyboots, 02 May 2014 - 06:22 AM.

[Craig Steele]

CCC Dober, on 02 May 2014 - 06:02 AM, said:

Well, you have managed to answer your own question. Sort of.

Unfair matches (due to flawed group implementation) lead to unsatisfied players and reduced income by extension. If PGI does not to tackle this problem correctly, because their statistical analysis is misleading, then there is a very real danger of losing touch with the community. Therefore it is vital that PGI improves their understanding of data and statistics, especially if they are used to base decisions on.

Is it worth the effort? Financially speaking, I would say so. The alternative for PGI is to kiss goodbye more revenue each time Paul tinkers around with parameters/numbers. Just add up the loss of revenue during and after each LRMgeddon. I'm sure it's more than enough to employ somebody, who is well equipped to deal statistics and maths in general.

OK, but lets be honest.

You're assuming that your view represents the majority of customer and ergo, the company is failing. The offering is not meeting your needs, therefore no one else is having their needs met.

Now you might be right, but its also possible that PGI are picking up 20 new high paying customers for every one they lose taking thir offering in the direction they are. They might be over the moon with how the company is performing and they are smart enough to know that no product has yet got 100% market penetration. Even Coke has not got 100% and that's about as close as anyone has ever got.

We'll never know until the lights get turned off or we see a headline in 2 years time saying they're gazillionaires.

I have seen hundreds, probably thousands of posts predicting the doom of PGI. Every patch, every balance adjustment, every communication has someone screaming out that it's the end of the world. And so far none of them have been right. You might be, IDK.

But what I do know is that it is unreasonable for us as consumers / customers to assume we have any call over the companies direction.

Anyone who wants to run PGI though can probably send in their resume, they are hiring after all.

[Joseph Mallan]

Jun Watarase, on 01 May 2014 - 03:13 PM, said:

That doesnt help much because with 24 players per match, and at 16% players being premades, statistically wise on average about 4 players per match are going to be in a group of some kind. Which anyone can tell you.

Statistical average from PGIs numbers is a 3 man team not 4 man. So on average 3 players are ruining a match and eating yer babies!

[Craig Steele]

Joseph Mallan, on 02 May 2014 - 06:32 AM, said:

Statistical average from PGIs numbers is a 3 man team not 4 man. So on average 3 players are ruining a match and eating yer babies!

Lies, Lies and statistics?

Edited by Craig Steele, 02 May 2014 - 06:45 AM.

[Joseph Mallan]

Craig Steele, on 02 May 2014 - 06:44 AM, said:

Lies, Lies and statistics?

And I can give you the numbers to prove it!

[CCC Dober]

@Craig
My perspective is that of a fan of the franchise. Even so, I'm also a Founder and have an interest to see MWO succeed and deliver on the promises. Now, if my experience and education allows me to point out the obvious (for me at least) and make a suggestion or two ... why not? At worst I get ignored, which is not surprising given PGI's history. At best I can give them an idea how to work the kinks out.

It's a mutually beneficial relationship as far as I'm concerned. I've got money to spend, they have a product to deliver. If they want more of that, it's up to them to deliver as promised. It's really that easy. Admittedly, group balancing was not specifically mentioned in the deal, but these days you take that for granted and it shouldn't even be a problem in the first place.