41 replies to this topic

[PhoenixFire55]

Skanderborg, on 26 June 2017 - 07:14 AM, said:

I like to elevate my self importance by being good at video games. Oh wait , no one cares.

Your level of understanding of what he said is about as good as your level of play ...

[Wyald Katt]

Yeonne Greene, on 26 June 2017 - 05:26 AM, said:

So the premise of the OP is pretty weak given that he has no idea what, exactly, PGI is seeing.

I want some of what they're having so I can go see things too.

I miss you, my sweet small wubsfu!

[VanillaG]

The stats page is worthless from a data mining perspective because it contains data from several years and multiple changes to the game. For the personal stats page to be even somewhat useful it needs to be broken down into into smaller segments so you can control for the various changes to the game. Ideally you could see the stats broken down on a monthly basis but then you run into a potential storage issue with having numerous smaller roll ups per person. I would be happy with only keeping the last 6 months of stats on the website with the ability to download the stats if I wanted to keep history.

[Y E O N N E]

Paigan, on 26 June 2017 - 05:32 AM, said:

That might be true, but you are making a similar assumption you insinuate the OP does.

The quering logic used to populate the UI MIGHT be buggy and thus falsify correct data.
It also MIGHT be correct and the error might reside in the collected/persisted data itself.

You do have a point, but not a terribly pertinent one. Nobody on this forum who isn't dev staff, myself included, is in a position to make any supposition on how flawed PGI's use of telemetry may be from this particular angle. Ergo, still weak.

Quote

Either way, the OP still has a point in that the PGI stats business ist least somewhat fishy.
Their data structure might be overly complicated and prone to making erroneous queries.
The balancing decisions might be based on buggy queries (maybe even the same that is used for the UI)
etc.

You can make a stronger case that their methodology is flawed, given that Chris has explained it several times, once even in writing on the MRBC website. The short version is that they tune stuff based on popularity measured against some nebulous concept of how popular they think it should be.

[Bud Crue]

Yeonne Greene, on 26 June 2017 - 09:20 AM, said:

The short version is that they tune stuff based on popularity measured against some nebulous concept of how popular they think it should be.

To quote from the last town hall:

"We have some internal values". That's the "short version".

That is literally all we have been told about their actual use of data to establish "balance". They have some internal values. Yup. The same "internal values" that they cited when nerfing the UACs on some seriously underplayed mechs, are the same values that demanded...DEMANDED...a .1 change in the duration of the ISLL. The same internal values that said the Victors -all of them- were just fine for three years. The same internal values that told them that all the Kodiaks -all of them at one time or another- needed nerffing. The same values that said the Arrow's machine gun quirks were OP. Etc. Etc. Etc.

They may have a rhyme and a reason for how they "balance" and how they "tune stuff" but all I know is that from my player/customer point of view their "internal values" are totally arbitrary and broken.

"We have some Internal values".
I swear that is going to be my new euphemism for when something makes no f***ing sense.

Edited by Bud Crue, 26 June 2017 - 09:35 AM.

[Y E O N N E]

Bud Crue, on 26 June 2017 - 09:35 AM, said:

That is literally all we have been told about their actual use of data to establish "balance".

It isn't, but you would have to go visit the MRBC website to find that out.

[Bud Crue]

Yeonne Greene, on 26 June 2017 - 09:45 AM, said:

It isn't, but you would have to go visit the MRBC website to find that out.

Can you give us a link, or is it readily apparent? I'm genuinely curious.

Edit: I'm looking around the forum posts on the MRBC website, and not finding anything. But if he has given an explanation that is actually useful for understanding some of the seeming nonsensical changes it seems a shame that they haven't posted it here. In any case I would like to read it if there is any insight to be gained.

Edited by Bud Crue, 26 June 2017 - 09:54 AM.

[Y E O N N E]

Bud Crue, on 26 June 2017 - 09:48 AM, said:

Can you give us a link, or is it readily apparent. I'm genuinely curious.

On my phone, don't know the BlackBerry code for a clean link, so here is the URL:

https://mrbcleague.c...php?storyid=240

[Bud Crue]

Yeonne Greene, on 26 June 2017 - 09:54 AM, said:

On my phone, don't know the BlackBerry code for a clean link, so here is the URL:

https://mrbcleague.c...php?storyid=240

Got it. Yes I have seen that. It doesn't really explain how their "internal values" are applied; how they actually use in game data to shif things...other than the meta...and even that is a pretty amorphous explanation that he gives. I want to know things like why crap mechs seemingly and randomly get occasional additional nerfs and why sometimes good mechs (sometimes meta mechs) get buffs? That's the kind of stuff I wish they would share with us. I mean if they have data and internal values that explain why a .1 duration changes is called for in a weapon, and yet are incapable for months on end of differentiating the comparative poularity and intrinsic value of a Kodiak 5 relative to a Kodiak 3, then I fear that no matter what Chris is using in his balance efforts is still akin to guessing. I'd like to be proven wrong on that.

[Deathlike]

Bud Crue, on 26 June 2017 - 09:35 AM, said:

To quote from the last town hall:

"We have some internal values". That's the "short version".

That is literally all we have been told about their actual use of data to establish "balance". They have some internal values. Yup. The same "internal values" that they cited when nerfing the UACs on some seriously underplayed mechs, are the same values that demanded...DEMANDED...a .1 change in the duration of the ISLL. The same internal values that said the Victors -all of them- were just fine for three years. The same internal values that told them that all the Kodiaks -all of them at one time or another- needed nerffing. The same values that said the Arrow's machine gun quirks were OP. Etc. Etc. Etc.

They may have a rhyme and a reason for how they "balance" and how they "tune stuff" but all I know is that from my player/customer point of view their "internal values" are totally arbitrary and broken.

"We have some Internal values".
I swear that is going to be my new euphemism for when something makes no f***ing sense.

I would love to have "internal discussions" with PGI, except reading all this stupid is giving me "internal bleeding" of the mind. At this point, we should just call this "telemetry" that has been used before.

Bud Crue, on 26 June 2017 - 10:04 AM, said:

Got it. Yes I have seen that. It doesn't really explain how their "internal values" are applied; how they actually use in game data to shif things...other than the meta...and even that is a pretty amorphous explanation that he gives. I want to know things like why crap mechs seemingly and randomly get occasional additional nerfs and why sometimes good mechs (sometimes meta mechs) get buffs? That's the kind of stuff I wish they would share with us. I mean if they have data and internal values that explain why a .1 duration changes is called for in a weapon, and yet are incapable for months on end of differentiating the comparative poularity and intrinsic value of a Kodiak 5 relative to a Kodiak 3, then I fear that no matter what Chris is using in his balance efforts is still akin to guessing. I'd like to be proven wrong on that.

Honestly, it probably has less to do with any sort of high level understanding of the mech relative to the game, but more to do with it being complained at on the lower levels, instead of the higher levels.

The Kodiak-3 being nerfed is warranted, but nerfing the other Kodiak variants is not. How that was interpreted was the Kodiak's as a whole getting nerfed... so obviously some sort of transitional/relational property is being applied in a blanket form (one is OP, so the others are technically OP), but not understanding the nuances of the different variants (particularly battlemechs as they obviously have different build characteristics than omnimechs).

At this point, subscribing to the Dartboard of Balance is better for the soul and mind. Then you can imagine what is going on...

[Bud Crue]

Deathlike, on 26 June 2017 - 10:26 AM, said:

At this point, subscribing to the Dartboard of Balance is better for the soul and mind. Then you can imagine what is going on...

Yup. And that dartboard brings us back home to "we have some internal values". Sigh.

[Wyald Katt]

Deathlike, on 26 June 2017 - 10:26 AM, said:

At this point, subscribing to the Dartboard of Balance is better for the soul and mind. Then you can imagine what is going on...

I left a most amusing ring of dart-sized holes on a wall of my childhood home's garage, when we finally moved out.

Thank goodness I wasn't using LRMs at the time, right?

Edited by Wyald Katt, 26 June 2017 - 10:39 AM.

[Y E O N N E]

Bud Crue, on 26 June 2017 - 10:04 AM, said:

Got it. Yes I have seen that. It doesn't really explain how their "internal values" are applied; how they actually use in game data to shif things...other than the meta...and even that is a pretty amorphous explanation that he gives. I want to know things like why crap mechs seemingly and randomly get occasional additional nerfs and why sometimes good mechs (sometimes meta mechs) get buffs? That's the kind of stuff I wish they would share with us. I mean if they have data and internal values that explain why a .1 duration changes is called for in a weapon, and yet are incapable for months on end of differentiating the comparative poularity and intrinsic value of a Kodiak 5 relative to a Kodiak 3, then I fear that no matter what Chris is using in his balance efforts is still akin to guessing. I'd like to be proven wrong on that.

Hence why I said "nebulous" earlier. We know the higher level methodology for determining whether or not something is OP or UP, but they haven't shared how they determine the specific fix for either case.

[Deathlike]

Yeonne Greene, on 26 June 2017 - 10:48 AM, said:

Hence why I said "nebulous" earlier. We know the higher level methodology for determining whether or not something is OP or UP, but they haven't shared how they determine the specific fix for either case.

To date, it's who yells the most and the loudest, particularly those that are not really good at the game.

Frankly, I almost think they want comp players to yell the most and the loudest (while not being obnoxious, which is kinda impossible given the stigma we get fairly or unfairly), but most of the time if you debate people that have an opinion that makes no sense... it ends up being "because reasons" instead of "this counters this more effectively in X situation and Y builds like it". When you leave it to people that don't truly maximize their mech (particularly in making the same mistakes in builds and/or usage), then it's easier to see the holes in their understanding.

At this point, it's easier to take the long approach and remind PGI what they did previously and then some to put into context how poorly one decision was made, so that everyone can remember what PGI did.

Edited by Deathlike, 26 June 2017 - 11:11 AM.

[Y E O N N E]

Deathlike, on 26 June 2017 - 11:11 AM, said:

To date, it's who yells the most and the loudest, particularly those that are not really good at the game.

Frankly, I almost think they want comp players to yell the most and the loudest (while not being obnoxious, which is kinda impossible given the stigma we get fairly or unfairly), but most of the time if you debate people that have an opinion that makes no sense... it ends up being "because reasons" instead of "this counters this more effectively in X situation and Y builds like it". When you leave it to people that don't truly maximize their mech (particularly in making the same mistakes in builds and/or usage), then it's easier to see the holes in their understanding.

At this point, it's easier to take the long approach and remind PGI what they did previously and then some to put into context how poorly one decision was made, so that everyone can remember what PGI did.

...so I should make my laser post and elucidate on why each value will work, then?

[DANKnuggz]

Karl Streiger, on 25 June 2017 - 10:06 PM, said:

No not this again.
Open you eyes. Open you mind.
You infidels are fools that fear what they can't unterstand.
See shots?
See hits?
See damage?

Consider that PGI can compare stats from patch to patch - this is everything you need.
Same for Mechs.
Absolute correct to use data instead of player XP.
Because otherwise you start to balance skill that would be stupid

Uhhh you do realize the OP was stating that the data she was showing was FALSE....
so you're saying they should balance based on FALSE data????

[Deathlike]

Yeonne Greene, on 26 June 2017 - 12:23 PM, said:

...so I should make my laser post and elucidate on why each value will work, then?

That feeds into the TL;DR mantra. I've written threads like that in the PTS area. That's what it came across to me as a consequence.

Edited by Deathlike, 26 June 2017 - 12:46 PM.

[Aim64C]

DANKnuggz, on 26 June 2017 - 12:31 PM, said:

Uhhh you do realize the OP was stating that the data she was showing was FALSE....
so you're saying they should balance based on FALSE data????

To be fair, I suspect the website is not pulling data from the servers correctly. The 'real database' should be much more thorough and you can construct each set of queries to break down the data however you like. In theory. If it's done right.

Granted, the website is our only window into what is/isn't on the server... so... to us, it doesn't look good. It -could- just be that the program setting up the queries for the website is bugged... could be that all of the data in the servers is par for the course and junk.

[Nightbird]

I've been trying hard to get marauders and kodiaks nerfed by using them continuously outside of events, and racking up dozens of kills and thousands of damage per death. Have I been successful?

[VanillaG]

DANKnuggz, on 26 June 2017 - 12:31 PM, said:

Uhhh you do realize the OP was stating that the data she was showing was FALSE....
so you're saying they should balance based on FALSE data????

You need to realize that the stats page is just a bunch of buckets that are updated after a match and has been running for a couple of years. It does not take into account any changes to the telemetry system nor does it correct any mistakes made collecting telemetry that have since been fixed. The fact that it contains such old data means it is worthless from data mining perspective.

My suspicion is the the stats logging system logs into multiple systems. One system is the one that we can see that are just rollup buckets that are incremented and a separate system that is actually used for analysis. The granularity of the analytic data would mean that it is impractical to have the same level of granularity for each player.