1911 replies to this topic

[Cimarb]

Karl Berg, on 30 June 2014 - 05:57 PM, said:

Heh, obviously two different things being discussed there.

Regarding those grouping stats that you were inquiring about:

I've been asked about those stats several times in this thread even, and I don't know what time period they were pulled from or how exactly they were gathered. They are certainly out of date right now; and they are about to be completely obsoleted this Wednesday. Your best bet is to ask Paul or Bryan for updated numbers once we've gathered a few weeks of data on production with the new grouping systems.

And as to what Bryan was talking about:

We're still in the middle of ongoing investigations as to the measurable impacts of grouping on win outcomes in our game. We literally have millions and millions of data points to examine; so we can produce some highly statistically significant results. Because of this, I have some really neat results regarding the impact of premades on solo games and premades of other sizes. This is all in addition to all of the various predictive tests we've been running on different algorithms for Elo convergence using production data. It's fascinating stuff, we've had several extremely late nights examining, hypothesizing about, running experiments on, and generally discussing this data as a group, and we have some really cool results that I'm hoping we can share at some point soon. So that's likely why Bryan was a bit confused about the topic of discussion.

Note that these investigations are still ongoing. We will likely be making some fundamental changes to the way we track and measure skill ratings in the game; but those changes will be made only with high confidence, after all our investigations are completed.

Well, to be totally honest, I knew it was slightly different and I was hoping to just evolve the question to get the answer I wanted. I fail at doing that quite horribly, so it didn't work

I know you have been asked the question A LOT, but that is because we have never received an answer! If we get an answer, you won't have to field the question again

So............... is a 4-man group considered a single "drop", or is it 4 "drops"?

[Kageru Ikazuchi]

Karl Berg, on 30 June 2014 - 05:57 PM, said:

We're still in the middle of ongoing investigations as to the measurable impacts of grouping on win outcomes in our game. We literally have millions and millions of data points to examine; so we can produce some highly statistically significant results. Because of this, I have some really neat results regarding the impact of premades on solo games and premades of other sizes. This is all in addition to all of the various predictive tests we've been running on different algorithms for Elo convergence using production data. It's fascinating stuff, we've had several extremely late nights examining, hypothesizing about, running experiments on, and generally discussing this data as a group, and we have some really cool results that I'm hoping we can share at some point soon.

Nice ... I really would like to know more. Statistical trend analysis is a hobby of mine ... yes, I know, it's a masochistic hobby, but trying to find meaningful patterns in a large amount of data is actually interesting to me.

[Kageru Ikazuchi]

Karl Berg, on 30 June 2014 - 05:57 PM, said:

I've been asked about those stats several times in this thread even, and I don't know what time period they were pulled from or how exactly they were gathered.

Just to be clear ... this command chair post ... was the source of the "84%" outrage, and the problem is one of unclear communication. While it appeared to me that the message and conclusion that was reached was "84% of the players for that period of time dropped solo", but that's not what the post clearly said ...

What it said was ...

Quote

Let’s start with some surprising facts pulled from the game server’s metrics:
- Out of all matches launched, 84% are solo launches.
- 16% are group launches. (We omitted our 12-mans because they will be dealt with separately)
- Of that 16%, 8% are 2-man groups, 4% are 3-man groups and 4% are 4-man groups.

Which has been interpreted by some to mean that 84% of the "launch button presses" are solo, and 16% are group "launch button presses", which comes up with a different figure for the player population ... 84*1 + 8*2 + 4*3 + 4*4 = 128, or 84 of 128, or 65.625% of the player base at that particular time dropped solo, or 34.375% of the player base was grouped.

Honestly, I think it was a communication error, and not an intentional effort to disenfranchise players who prefer to play grouped ... and without the raw data, I don't really trust simple pie charts anyway. The stats themselves really are irrelevant (even moreso in a couple of days).

Cimarb, on 30 June 2014 - 06:33 PM, said:

I know you have been asked the question A LOT, but that is because we have never received an answer! If we get an answer, you won't have to field the question again

I really wish the issue would just go away. It's very quickly becoming obsolete, because of Karl's efforts to integrate larger groups into the game. It should be obvious that PGI is not "against" players who prefer to play grouped (whether we represent 16% or 34% of the player base), but I guess the only way to clear the air once and for all would be to ask them:

"What did you really mean? and, why didn't you just say that (whatever "that" is) more clearly?"

[Shlkt]

Quote

We're still in the middle of ongoing investigations as to the measurable impacts of grouping on win outcomes in our game.

Sounds like a difficult analysis. You'd have to differentiate between players who group up often (i.e. their Elo has had time to adjust to their grouping behavior) and those who group up infrequently or with players of inconsistent skill levels (i.e. not always in a high-level group, sometimes dropping with noobs, etc...). I suppose you could make some very general statements about the average impact of group size... but as soon as you apply it to matchmaking then it could very well cause player grouping behavior and/or player performance to change, invalidating your previous analysis. Have fun with that

EDIT: on second thought, probably no more difficult than accounting for the various crazy builds that people run. Heavy mech != heavy mech. Nevermind.

Edited by Shlkt, 01 July 2014 - 05:02 AM.

[SgtKinCaiD]

@Karl

Thanks for your time and response, but i got another question :

Karl Berg, on 30 June 2014 - 01:50 PM, said:

It will see much tighter Elo restrictions brought into place. How restrictive depends on how constrained we will be able to set these values to on production and still maintain our goals for worst case user wait times.

So, from what i understand from your response, the ELO bucket are not re-introduced with the 4*3. Now i would like to know if the new MM will still use the rule of balancing a high ELO and a low ELO player against 2 medium ELO players ?
Matthew Craig post for reference : http://mwomercs.com/...ost__p__2633230

Bonus question : did you tune the ELO scoring formula for the new MM so that it's not based on Win/Loss only ?

[Cimarb]

Kageru Ikazuchi, on 01 July 2014 - 12:38 AM, said:

Just to be clear ... this command chair post ... was the source of the "84%" outrage, and the problem is one of unclear communication. While it appeared to me that the message and conclusion that was reached was "84% of the players for that period of time dropped solo", but that's not what the post clearly said ...

What it said was ...

Which has been interpreted by some to mean that 84% of the "launch button presses" are solo, and 16% are group "launch button presses", which comes up with a different figure for the player population ... 84*1 + 8*2 + 4*3 + 4*4 = 128, or 84 of 128, or 65.625% of the player base at that particular time dropped solo, or 34.375% of the player base was grouped.

Honestly, I think it was a communication error, and not an intentional effort to disenfranchise players who prefer to play grouped ... and without the raw data, I don't really trust simple pie charts anyway. The stats themselves really are irrelevant (even moreso in a couple of days).

I really wish the issue would just go away. It's very quickly becoming obsolete, because of Karl's efforts to integrate larger groups into the game. It should be obvious that PGI is not "against" players who prefer to play grouped (whether we represent 16% or 34% of the player base), but I guess the only way to clear the air once and for all would be to ask them:

"What did you really mean? and, why didn't you just say that (whatever "that" is) more clearly?"

Very well put. I don't agree that it is becoming obsolete, as I am interested in how many people ARE dropping solo compared to groups, but any information that I can get to clarify an issue that has been argued about is very worthwhile to me in general. PGI is doing immensely better lately, but they still like to be vague, which always leads to debate about what was meant, when it should be a clear message instead.

SgtKinCaiD, on 01 July 2014 - 05:49 AM, said:

@Karl
Bonus question : did you tune the ELO scoring formula for the new MM so that it's not based on Win/Loss only ?

Oh 100 times this. I have never been a fan of judging my solo drop performance upon all of the other solo people I am dropping with.

[Imperius]

As promised finally got my computer built and MWO installed on it. I was very pleased with my results.

4790k 4.7GHz (OC'ed)
Asus Z97 Deluxe motherboard
G.Skill 16 GB 2400MHz RAM
Crucial MX100 512GB SSD (500MB R/W per sec)
EVGA GTX 680 2GB GDDR5
Corsair Vengeance C70 case
Corsair K70 Mechanical Keyboard
Razer Ouroboros gaming mouse

MWO ran 80-100FPS so I added sys_MaxFPS=60 anything over 60 is wasted energy and FPS

If anyone would like pictures just ask and I'll post

[SnagaDance]

And how much would such a sweet machine set me back?

[Heffay]

Imperius, on 02 July 2014 - 06:20 AM, said:

As promised finally got my computer built and MWO installed on it. I was very pleased with my results.

4790k 4.7GHz (OC'ed)
Asus Z97 Deluxe motherboard
G.Skill 16 GB 2400MHz RAM
Crucial MX100 512GB SSD (500MB R/W per sec)
EVGA GTX 680 2GB GDDR5
Corsair Vengeance C70 case
Corsair K70 Mechanical Keyboard
Razer Ouroboros gaming mouse

MWO ran 80-100FPS so I added sys_MaxFPS=60 anything over 60 is wasted energy and FPS

If anyone would like pictures just ask and I'll post

What resolution are you running the game at? And what chipset is that motherboard at? 1155?

Edited by Heffay, 02 July 2014 - 06:39 AM.

[Cimarb]

Imperius, on 02 July 2014 - 06:20 AM, said:

As promised finally got my computer built and MWO installed on it. I was very pleased with my results.

4790k 4.7GHz (OC'ed)
Asus Z97 Deluxe motherboard
G.Skill 16 GB 2400MHz RAM
Crucial MX100 512GB SSD (500MB R/W per sec)
EVGA GTX 680 2GB GDDR5
Corsair Vengeance C70 case
Corsair K70 Mechanical Keyboard
Razer Ouroboros gaming mouse

MWO ran 80-100FPS so I added sys_MaxFPS=60 anything over 60 is wasted energy and FPS

If anyone would like pictures just ask and I'll post

SnagaDance, on 02 July 2014 - 06:26 AM, said:

And how much would such a sweet machine set me back?

Based upon his list, it was roughly $1512, plus the video card (which was about $500 two years ago, but is out of stock most places currently so hard to get a good price on it) and things he didn't list like monitor, power supply, network card, headset, etc.

So, if you are buying it all new, it would likely be $2k-3k depending on the quality of that other stuff.

BTW, Imp, how does that mouse work? Looks like a half-shifted Transformer from the pictures I saw...seems like it wouldn't be very ergonomical, lol

[Cimarb]

Heffay, on 02 July 2014 - 06:38 AM, said:

What resolution are you running the game at? And what chipset is that motherboard at? 1155?

That board is 1150 so it can run the newer Haswell processors. 1155 is the old one. (Never understood the lower number being better...)

[Goose]

It's the actual pin count of the socket; Why it's going down as they get more uptight about the voltage as the CPU downclocks …

[Imperius]

SnagaDance, on 02 July 2014 - 06:26 AM, said:

And how much would such a sweet machine set me back?
About 2k

Heffay, on 02 July 2014 - 06:38 AM, said:

What resolution are you running the game at? And what chipset is that motherboard at? 1155?

1080p not willing to jump on 4k till supported and cheaper.

Cimarb, on 02 July 2014 - 06:40 AM, said:

Based upon his list, it was roughly $1512, plus the video card (which was about $500 two years ago, but is out of stock most places currently so hard to get a good price on it) and things he didn't list like monitor, power supply, network card, headset, etc.

So, if you are buying it all new, it would likely be $2k-3k depending on the quality of that other stuff.

BTW, Imp, how does that mouse work? Looks like a half-shifted Transformer from the pictures I saw...seems like it wouldn't be very ergonomical, lol

Love the mouse comes with replacement sides. Works flawlessly, I can custom set up DPI levels via a slider. Perfect for sniping and switching to fast brawl play.

[Cimarb]

Imperius, on 02 July 2014 - 08:36 AM, said:

Love the mouse comes with replacement sides. Works flawlessly, I can custom set up DPI levels via a slider. Perfect for sniping and switching to fast brawl play.

That would be awesome if you could have it toggle DPI setting when you zoom in... Not sure if that would be as feasible as doing it manually when you need it, but...

[Karl Berg]

Li Song, on 21 June 2014 - 09:20 AM, said:

All the weapon range modules that are currently in game affect weapon heat. So I interpret your answer as that these modules do not affect base heat but rather is a "bonus" that is added on top after any ghost heat penalty is calculated?

Hi Li Song,

Unfortunately none of the gameplay team in the office right now are familiar with the specific implementation of ghost heat; however Brian Buckton ran some quick tests, and it appears that the added heat from the weapon range modules gets scaled up by the heat scale system.

Tests were run using latest production build on a HBK with 9 medium lasers. Without the range module, an alpha would generate 70% heat standing still on Forest. With the range module, a full alpha generated 73% heat instead.

[Alaskan Nobody]

That sounds like a headache
(not that it hasn't before)

[Hobo Dan]

Karl Berg, on 30 June 2014 - 01:50 PM, said:

Hey Hobo Dan, sorry for the delay on this response. Unfortunately this bug is apparently still open and under investigation. The bug is apparently very difficult to reproduce.

Thank you sir. I shall suffer on lol!

[Imperius]

Cimarb, on 02 July 2014 - 10:37 AM, said:

That would be awesome if you could have it toggle DPI setting when you zoom in... Not sure if that would be as feasible as doing it manually when you need it, but...

Well to toggle between snipe and brawl I press up or down since I can pick the DPI from 50-8500 some crazy sensitive number ( most likely for 4K.) I have it at 100 then jumps to 1000. Got to get those snap shots in al-la call of duty... *rolles eyes* it does help in "brawling" situations to switch so easy. I was going to take off the right side wing, but it holds my pinky so well I don't get pinky drag, I thought I I would hate it, but I actually really love it

You probly could macro in it since it does have macro software. Don't know yet spent past two days installing, and tweaking Windows 8.1 for SSD performance and games.

Pros: Comfort, quality, Highest DPI I've seen on a mouse.
Con: PRICE, damn it's not cheap.

[IronChance]

Karl,

The new matchmaker launch has been smooth as silk and I've been enjoying it all night long. I'm not sure how well balanced the group queue is right now, but I'm really impressed that all the changes were put in and it works as well as it has been on first launch.

Nice job, man.

[Modo44]

Now that the matchmaker makes sense, there are two obvious improvements:

Track group and solo Elo separately. You keep talking about adding a set modifier, but that has obvious issues. You can not predict who sucks the same (or even more) when grouped up. Pushing group Elo manually would introduce new problems, specifically: unduly punishing bad players by asking them to carry unrealistically hard. Tracking group Elo would automatically adjust for every player's group play (in)ability.

Track chassis Elo separately. Seeding issues were mentioned for this. One solution would be using MWO-specific battle values (i.e. averaged Elo values for each chassis -- already tracked) to help with initial seeding and re-seeding of chassis Elo. Again, no manual adjustments, just working with heuristics based on Elo.