Jump to content

Paging Karl Berg...karl Berg, Please Pick Up The White Courtesy Phone...


1911 replies to this topic

#121 Modo44

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bad Company
  • 3,559 posts

Posted 10 April 2014 - 05:16 AM

12:3, not 12:6. A free win/guaranteed loss often has the best players on the losing team scoring "their" kills and/or someone derping on the winning team. An easy loss predictor is provided in the first 30s of a match. Look at the minimap. Are people splitting up? Yes -- GG. No -- there is a chance. Things this obvious are squarely on the matchmaker.

Edited by Modo44, 10 April 2014 - 05:16 AM.


#122 Heffay

    Rum Runner

  • PipPipPipPipPipPipPipPipPipPip
  • The Referee
  • The Referee
  • 6,458 posts
  • LocationPHX

Posted 10 April 2014 - 05:50 AM

View PostModo44, on 10 April 2014 - 05:16 AM, said:

12:3, not 12:6. A free win/guaranteed loss often has the best players on the losing team scoring "their" kills and/or someone derping on the winning team. An easy loss predictor is provided in the first 30s of a match. Look at the minimap. Are people splitting up? Yes -- GG. No -- there is a chance. Things this obvious are squarely on the matchmaker.


That's not an indicator at all. Sorry, but different roles require different strategies.

#123 Modo44

    Member

  • PipPipPipPipPipPipPipPipPip
  • Bad Company
  • 3,559 posts

Posted 10 April 2014 - 05:59 AM

View PostHeffay, on 10 April 2014 - 05:50 AM, said:

That's not an indicator at all. Sorry, but different roles require different strategies.

Yes, it is. I am not talking about a light mech. I am talking about half the team enjoying the scenery for no apparent reason.

Anyway, enough word wrangling. Anyone from PGI with access to player data can simply watch an actually high-Elo player stream. Some problems are really painfully obvious in those matches, and given the system's duality, also affect new/very bad players.

#124 Heffay

    Rum Runner

  • PipPipPipPipPipPipPipPipPipPip
  • The Referee
  • The Referee
  • 6,458 posts
  • LocationPHX

Posted 10 April 2014 - 07:52 AM

Dear Mr Berg,

In all the back end work and upgrades you've been doing, have any tools/APIs been created to allow the player base to mine data? Things like being able to access your in game Friends list so that clients can be built to chat with them via a web client? Or being able to send group invites outside of the game client, that will let people start up the game client and hop right into the group?

I know PGI_Fox was working on web API to do things like this as well, such as access profile info and mech details through third party clients as well. So much could be done by the community if we had better access to some of this information.

#125 Klappspaten

    Member

  • PipPipPipPipPipPipPipPip
  • Bad Company
  • Bad Company
  • 1,211 posts
  • LocationGermany

Posted 10 April 2014 - 08:04 AM

I agree, especially an API systm would make things a lot easier.

#126 Roadbeer

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Elite Founder
  • 8,160 posts
  • LocationWazan, Zion Cluster

Posted 10 April 2014 - 08:37 AM

This thread has moved from interesting to awesomesauce!

If only we'd received communication like this all along, and not just "dictates from on high".

Looking at you Niko :) Raise the bar, don't just slump into "Cicada build of the week" posting ;)

#127 DEN_Ninja

    Member

  • PipPipPipPipPipPipPipPip
  • The Blade
  • The Blade
  • 1,097 posts
  • LocationCrossing, Draconis March

Posted 10 April 2014 - 09:12 AM

View PostRoadbeer, on 10 April 2014 - 08:37 AM, said:

This thread has moved from interesting to awesomesauce!

If only we'd received communication like this all along, and not just "dictates from on high".

Looking at you Niko :) Raise the bar, don't just slump into "Cicada build of the week" posting ;)


Rasc41 said:

Yes, Karl Berg appears to be full of win. My shock has turned to awe. He doesn't have to promise anything. Just the fact that he posts well argued points HERE, not twitter, NGNG or whatever. I really can't fathom why the devs seem at least partially willing to discuss their game everywhere else but the game's *actual* forums.


If you look at the List of Viewers of the forums I see Red names quite often. I am going to make an assumption that they do freaking read the forums but the more militant members ward them away from an intelligent conversation. This case is most extreme in gameplay balance where the bulk of forum posters read and post.

Anyways I am absolutely loving the two way feedback right now.

#128 Shamous13

    Member

  • PipPipPipPipPipPipPip
  • 684 posts
  • LocationKitchener, Ont.

Posted 10 April 2014 - 09:53 AM

I'm loving the thread. keep it up

Karl,
Apparently, there is some controversy about people using VPN to help with Hit registration and altering their ping so they take less damage. I don't know that there's anything conclusive here, but someone took the time to put this together:

http://www.reddit.com/r/OutreachHPG/com ... lords_vpn/

The second video seems the most damning, but I'm not really sure the VPN is doing what they think it's doing. when connecting to your back end.

#129 Scott Decoy Stone

    Member

  • PipPipPipPipPip
  • Survivor
  • Survivor
  • 106 posts
  • Facebook: Link
  • Locationhonolulu, hi

Posted 10 April 2014 - 10:45 AM

solid read so far. i'm really looking forward to more. thanks karl for taking the time to communicate with people. if only others would do the same.

#130 Karl Berg

    Technical Director

  • 497 posts
  • LocationVancouver

Posted 10 April 2014 - 10:56 AM

View PostHeffay, on 10 April 2014 - 07:52 AM, said:

Dear Mr Berg,

In all the back end work and upgrades you've been doing, have any tools/APIs been created to allow the player base to mine data? Things like being able to access your in game Friends list so that clients can be built to chat with them via a web client? Or being able to send group invites outside of the game client, that will let people start up the game client and hop right into the group?

I know PGI_Fox was working on web API to do things like this as well, such as access profile info and mech details through third party clients as well. So much could be done by the community if we had better access to some of this information.


Yes, Fox and I have been working on opening this up much more. We have several changes in the pipeline related to exactly this kind of live info sharing.

View PostShamous13, on 10 April 2014 - 09:53 AM, said:

I'm loving the thread. keep it up

Karl,
Apparently, there is some controversy about people using VPN to help with Hit registration and altering their ping so they take less damage. I don't know that there's anything conclusive here, but someone took the time to put this together:

http://www.reddit.com/r/OutreachHPG/com ... lords_vpn/

The second video seems the most damning, but I'm not really sure the VPN is doing what they think it's doing. when connecting to your back end.


Yup, a few of these users have contacted me directly. We are investigating.

#131 Karl Berg

    Technical Director

  • 497 posts
  • LocationVancouver

Posted 10 April 2014 - 11:06 AM

View PostModo44, on 10 April 2014 - 12:18 AM, said:

Please understand: Your ability to correctly predict a stomp does not make a stomp any more fun for the participants.


Problem subdivision. We need to have a metric for how likely a stomp is going to happen if we are to have any chance of minimizing that metric on the matchmaker. In this case, we set out to prove that Elo was a better metric than random selection, and if it was to gauge how much better it was.

Once we have that metric, there is a separate problem to solve, which is to have the matchmaker use those Elo values to produce better matches. Two pieces to the problem, hence the two pieces of info I've tried to share on the topic. One to show that match outcome prediction using Elo has a strong correlation with actual outcome, and one to show that the matchmaker is actively minimizing that metric.

Can it get better? Certainly. As we fix communication, teamwork, grouping, and other sources of bias, we should see that predicted Elo curve and actual outcome curve converge. We could also trade longer wait queue times for closer Elo matches. We don't want to move forward haphazardly though. We will only make changes with high levels of confidence, because if we mess this up it could ruin the experience for everyone. That means running repeated tests on our production telemetry to show measurable improvement at the very least.

#132 Karl Berg

    Technical Director

  • 497 posts
  • LocationVancouver

Posted 10 April 2014 - 09:14 PM

@Chronojam: That post is still coming, sorry for the delay. Changes were requested, so I spent some time tonight to implement everything that was requested. It's going back for a second review; but hopefully tomorrow it will get approved for posting.

#133 Tekadept

    Member

  • PipPipPipPipPipPipPipPip
  • Knight Errant
  • 1,290 posts
  • LocationPerth, Australia

Posted 10 April 2014 - 09:43 PM

View PostKarl Berg, on 10 April 2014 - 09:14 PM, said:

@Chronojam: That post is still coming, sorry for the delay. Changes were requested, so I spent some time tonight to implement everything that was requested. It's going back for a second review; but hopefully tomorrow it will get approved for posting.

See this is what needs to happen with PGI, any other annoucement that was "due" from PGI we do not hear anything, all we know is that it just hasnt been posted yet and may happen "SOON" we "guess".

You on the other hand, found out there had to be changes that need to be made, and have informed us, gave us a new eta and kept us in the loop. BRAVO, if only the rest of PGI worked like that.

Edited by Tekadept, 10 April 2014 - 09:55 PM.


#134 Chronojam

    Member

  • PipPipPipPipPipPipPipPipPip
  • 2,185 posts

Posted 10 April 2014 - 10:03 PM

Yes, thanks for the update. I'm eager to see the feedback of course, and had a few people ask me just today if it had been posted yet.

#135 slide

    Member

  • PipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 1,768 posts
  • LocationKersbrook South Australia

Posted 11 April 2014 - 12:29 AM

I am late to this party but I would like to Thank Karl for his participation in this conversation.

Additionally I would like to reinforce what Chronojam had to say in his rather long post to which we are all eagerly awaiting a reply.

Much of what CJ has said applies to almost all of he Units have had contact with in the past 2 years or so. I think the thing that annoys people the most and in particular Founders is the seeming inability to deliver on promises or proposals made initially or continually by PGI themselves. The "reasons" may well be valid but to often the delivery of those "reasons" has been poor at best and dismissive at worst.

As is well known, many people do not visit these forums, however one thing I can tell you that happens is that people who do not come here often go to unit forums or elsewhere, where I am sure information is reposted in a manner that only enhances the fall out of the already speculative/disingenuous information. (Chinese whispers anyone).

I'll give you an example, one I am partly responsible for.

We were having a meeting on TS about the same time as Paul posted about the launch module. One of ARMD's founding members logs in to listen. During this time he quizzes me in a private text chat about what is new in MWO. I give him a brief run down of how groups will have to have premium time to run 2-11 and there will be no groups larger than 4 in the public queue, something I know he was specifically waiting for. I gave him the link to read in full. I haven't spoken to him since or seen him online. Now why should PGI care, well the member in question was a founder, had bought several lots of MC, set up administered and paid for the ARMD website and web address out of his own pocket. He had spent literally hundreds of hours helping to set up ARMD as a unit with all the Web work and documentation to go with it. He had every reason you can think of to keep playing, but Pauls post was the straw that broke the camels back.

No community can afford to lose people who have this type of commitment, whether it be a game, a sporting club or a charity.
Yet it is a scenario that I am sure is being repeated every day, in droves. The conversion rate of White Knights to the dark side tells me that it is so. The thing that worries me is that for every poster you see here (light or dark), the ones who still care a bit at some level, how many have gone "pfft", click (uninstall) never to be heard from again.

FWIW ARMD could be around 800-900 strong, as it is, it is around about 50 regularly active with another 50 to 100 awaiting CW.

Again Karl I thankyou. You have done more for PGI's reputation with me in this one thread than has been done by your bosses in 2 years.

#136 Roadbeer

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Elite Founder
  • 8,160 posts
  • LocationWazan, Zion Cluster

Posted 11 April 2014 - 09:03 AM

View Postslide, on 11 April 2014 - 12:29 AM, said:

No community can afford to lose people who have this type of commitment, whether it be a game, a sporting club or a charity.
Yet it is a scenario that I am sure is being repeated every day, in droves. The conversion rate of White Knights to the dark side tells me that it is so. The thing that worries me is that for every poster you see here (light or dark), the ones who still care a bit at some level, how many have gone "pfft", click (uninstall) never to be heard from again.

FWIW ARMD could be around 800-900 strong, as it is, it is around about 50 regularly active with another 50 to 100 awaiting CW.

Again Karl I thankyou. You have done more for PGI's reputation with me in this one thread than has been done by your bosses in 2 years.


Just to reinforce this point.

Posted Image

This is a recent screenshot of an automated tagging process on the http://www.house-marik.net website. We recently revamped our website in October and this reflects logins since that time, with the "inactive" tag being automatically applied after not visiting the site for 90 days. As the website is also one avenue for connecting to our TS server, many people visit the site just to launch their TS client and sync their account with the appropriate permissions (not all, but some).

As you can see, in that time, almost 2/3 of our members are no longer active on our site. Of that, one would suspect a large portion are no longer active in the game as well.

Edited by Roadbeer, 11 April 2014 - 09:03 AM.


#137 Rebas Kradd

    Member

  • PipPipPipPipPipPipPipPipPip
  • 2,969 posts

Posted 11 April 2014 - 09:40 AM

I would like to caution everybody, that perhaps part of the reason you haven't heard the explanations for these "tough issues" is because the answers would not be particularly satisfying.

That said, I'm glad they're coming.

#138 Jman5

    Member

  • PipPipPipPipPipPipPipPipPip
  • Littlest Helper
  • Littlest Helper
  • 4,914 posts

Posted 11 April 2014 - 09:49 AM

Hi Karl, I'm not sure if you're the person to ask this, but I have noticed there seems to be a problem with object or terrain draw distances. Certain hills, boulders, buildings, and objects are either not drawing past a couple hundred meters and sometimes not until you're practically on top of it. These are things with solid mass that mechs can hide behind.

The result is that you get a lot of scenarios where the mech looks visible from 3-400 meters away, but when you shoot at it, you wont get a hit. You move closer and suddenly this great big rock appears out of no where or the hill morphs to show he was in fact behind him.

This causes a lot of people to blame hit registration for what clearly looked like a direct hit. It also gives away people's position when they think they're behind a solid building, rock, or hill. I can see them moving up to crest a hill and line up my shot.

I don't think it has always been like this, but several months back it was made much worse. Adjusting my graphic settings to max help a tiny bit, but the draw distances for these critical terrain features are still way too short. I think it's important to make structures or terrain that block shots visible from further out.

Are you guys aware of this, or can someone look to see if draw distances are being done correctly? I can provide examples if you want.

#139 arghmace

    Member

  • PipPipPipPipPipPipPip
  • Elite Founder
  • Elite Founder
  • 845 posts
  • LocationFinland

Posted 11 April 2014 - 09:54 AM

Object detail is the setting that controls the draw distance for terrain features. I recommend setting it to high since even medium setting leaves many buildings and hills invisible.

#140 Karl Berg

    Technical Director

  • 497 posts
  • LocationVancouver

Posted 11 April 2014 - 10:00 AM

View PostJman5, on 11 April 2014 - 09:49 AM, said:

Are you guys aware of this, or can someone look to see if draw distances are being done correctly? I can provide examples if you want.


Sounds like aggressive performance settings or LOD settings that may have been poorly tuned or taken way to far. If you have some good examples, I would be more than happy to pass them on to the engineer responsible for overseeing this.





2 user(s) are reading this topic

0 members, 2 guests, 0 anonymous users