Jump to content


Match Maker, Match Maker, Make Me A Match

MM Match Maker Devs Patent

  • Please log in to reply
128 replies to this topic

SenatorTH #61 Posted 15 December 2018 - 01:27 AM

    Senior Master Sergeant

  • Member
  • 3240 battles
  • 118
  • Member since:
    07-04-2018

I'll just copy my post from another thread.

--------------------------------------------

Everything was more or less fine with MM before 2.0.7 iirc, battles contained same tier mostly, waiting time was the same as now or even shorter. Suddenly from nowhere that new concept of MM appeared with mixed tiers, and since then all the complains started - sadly no one is listening.

 

Let's take standard situation with 4 humans each side.

 

WAS: I am t4, have 3 teammates t4, we fight against other 4 in t4 planes, all ground is t4. Very honest situation, isn't it?

 

IS: two possibilities.

1) I am t4, one more t4, two teammates t3. Other team contains 2x t4, 2x t3, ground and bots are t4 level. I don't care much about t3, as I fight vs t4. Situation for me is about the same as was earlier - fighting against equal opponents.

2) everything the same, but I am one of those t3's. Situation sucks completely - I fight vs higher tier, bots are higher tier, ground targets are higher tier, even the altitude for high-AAs is different. In fact it looks like predetermined conditions for loss. Remember, we're not considering any skills, just pure matchmaking.

Now if we add to this that +2 tier allowance, the situation with matchmaking comes anecdotic at all, like one t3 vs flight of two t4.

 

Conclusions you can make by yourself.

---------------------------------------------------

Zoney45 wrote a brilliant phrase: "when the devs put up a special bulletin trying to smooth down the players, then you know it's an issue."

Yeah, when everyone and their dog says it is wrong, they just issue an article explaining why it is good, LOL.

I entered this game with high enthusiasm, and was playing quite a lot. Until my patience lasted. Now I've started to skip days and even weeks - and feel much better - no frustration, no anger. Hell, this game instead of bringing the joy just makes you suffer. Probably the day of my wowp deletion is coming, as I do not believe anything will change to better anymore. It will change for sure, but for better? Nah...



Zoney45 #62 Posted 17 December 2018 - 09:04 AM

    Master Sergeant

  • Member
  • 952 battles
  • 43
  • Member since:
    11-10-2018

Nah, no balance issues here -- everything makes perfect sense. See how it all adds up? FUN!  :D

 

(As an aside, this was a map with three objectives in a straight line -- two military bases on the ends and a garrison in the center. I had pinged the two outer bases at the outset, but one of my "teammates" kept hammering on pinging the garrison throughout the match. I hit "negative" but he really really had a jones for that garrison. Maybe he mistook it for an airfield.)

 


Edited by Zoney45, 17 December 2018 - 09:33 AM.


Zoney45 #63 Posted 18 December 2018 - 05:06 AM

    Master Sergeant

  • Member
  • 952 battles
  • 43
  • Member since:
    11-10-2018

Here's another fun situation. This mission lasted 5 minutes. Look at the comparisons in wingman aircraft. On my team two ordinary tier-IV (fighter + multirole). The other team has wingmen with two tier-V specialist fighters.

 

The matchmaking apparently "thinks" it's being equitable by making sure each team has a flight-up pair, and never you mind the laughable disparity in aircraft. Please don't do me any favors. I'd rather just have raw horsepower than this worthless "finesse." I'd have traded both those tier-IV flights for another lone specialist tier-V fighter with a high-damage player. The player in the XP-36F was making a go of it but it was hard to get off the ground with the Dynamic Duo in their upper-tier specialists.

 

I sent one of 'em packing in a face-off but not before he and his buddy first took me out unawares.

 

Add to it it's a small-area, high-density battle with a short resource count (three closely spaced objectives including two airfields) so the dominant match-up just strafed back and forth a few times. (I give 'em credit for having a fairly disciplined flight pattern but with few on the other team to challenge and disrupt your formation it's not microsurgery.)


Having experienced it, I'm convinced frustration with matchmaking lopsidedness is a leading cause of player departure. I'm still digging the game but I haven't really had time to develop a full-blown permanent- rage-quit mentality.

 

It might suggest something that I'm here typing these short essays rather than flying.

 

See the happy results.

 


Edited by Zoney45, 18 December 2018 - 05:47 AM.


CorvusCorvax #64 Posted 18 December 2018 - 06:47 AM

    Captain

  • Member
  • 4012 battles
  • 4,033
  • [OWSS] OWSS
  • Member since:
    01-26-2012
Yeah, sometimes you just can't buy a good match.  Last weekend, I am flying MRFs, and the other team seems to be getting human bombers.  Okay, switch to HF.  Now flighted humans up tier from me in LFs.  Okay, dang it, I'm going to fly my Ta-152.  Finally! But the matchmaking was not working out.  A FW-190D does not match up at all with an RB-17.  There is just no part of that match up that comes out well for the MRF pilot.  Unless the human in the RB-17 is a complete potato.

Zoney45 #65 Posted 19 December 2018 - 12:41 AM

    Master Sergeant

  • Member
  • 952 battles
  • 43
  • Member since:
    11-10-2018

View PostCorvusCorvax, on 18 December 2018 - 06:47 AM, said:

Yeah, sometimes you just can't buy a good match.

 

Precisely. 

 

Referring back to my previous post, the matchmaking is "shrewd" enough to try and even out paired flights, but has no recognition for the vast difference between two paired tier-IV fighters that might be just beginning to grind for upgrades, versus paired fully upgraded specialist fighters of the next-highest tier that probably have several advanced-to-ultimate pieces of service equipment installed.

 

I just sat down and had four quick flights. Two of them I barely did anything (I mean I was a brick out there) and our team won, the other two I topped off the list and we lost. It's perplexing when stuff like that happens in waves or patterns.

 

I've seen other players on these forums complaining about "rigged matches" -- I'm not going that far, I don't think they're "rigged"; but there's definitely an over-large element of chance and luck for what is supposed to be a skill- and reflex-based tactical game.

 

Matchmaking idiosyncrasies aside, I have to temper my wailing and moaning by reiterating that the developers captured some excellent dynamics when they pieced this game together.

 

But back to the wailing-wall, here was another LOL result for posterity -- putting aside the W/L stats which I didn't bother digging into, notice the two specialist aircraft for two human players on the winning team vs none for the losing side, in addition to the 2:1 ratio of upper-tier fighters for human players on the winning team.

 


Edited by Zoney45, 19 December 2018 - 12:58 AM.


Perrigrino #66 Posted 29 December 2018 - 12:44 PM

    Senior Master Sergeant

  • Member
  • 5738 battles
  • 319
  • [OWSS] OWSS
  • Member since:
    10-02-2013

As for RNG, and how it affects matches, I will start adding some links to enhance the discussion, since it was one of the questions to the Devs : 

 

https://www.random.org/randomness/

 

Excerpt:

"

Pseudo-Random Number Generators (PRNGs)

As the word ‘pseudo’ suggests, pseudo-random numbers are not random in the way you might expect, at least not if you're used to dice rolls or lottery tickets. Essentially, PRNGs are algorithms that use mathematical formulae or simply precalculated tables to produce sequences of numbers that appear random. A good example of a PRNG is the linear congruential method. A good deal of research has gone into pseudo-random number theory, and modern algorithms for generating pseudo-random numbers are so good that the numbers look exactly like they were really random.

The basic difference between PRNGs and TRNGs is easy to understand if you compare computer-generated random numbers to rolls of a die. Because PRNGs generate random numbers by using mathematical formulae or precalculated lists, using one corresponds to someone rolling a die many times and writing down the results. Whenever you ask for a die roll, you get the next on the list. Effectively, the numbers appear random, but they are really predetermined. TRNGs work by getting a computer to actually roll the die — or, more commonly, use some other physical phenomenon that is easier to connect to a computer than a die is.

 

PRNGs are efficient, meaning they can produce many numbers in a short time, and deterministic, meaning that a given sequence of numbers can be reproduced at a later date if the starting point in the sequence is known ...."


Edited by Perrigrino, 03 January 2019 - 03:22 AM.


Perrigrino #67 Posted 29 December 2018 - 01:54 PM

    Senior Master Sergeant

  • Member
  • 5738 battles
  • 319
  • [OWSS] OWSS
  • Member since:
    10-02-2013

here is an excellent explanation and demonstration of just how the wrong "seed" can affect matches, typical where we see in wowp the same instance of being down-tiered, or up-tiered, but there is obviously more to it and would be nice for wowp devs to jump in here and explain what they use. Btw, it is not that complicated, or secret, in terms of HOW it is used. And just to be clear, I was not asking for specifics at the time:

 

 https://www.youtube....h?v=JGwSEbnJGR0

 


Edited by Perrigrino, 03 January 2019 - 03:22 AM.


Perrigrino #68 Posted 30 December 2018 - 04:54 AM

    Senior Master Sergeant

  • Member
  • 5738 battles
  • 319
  • [OWSS] OWSS
  • Member since:
    10-02-2013

and then there's this, which I include, not to be subversive, but to highlight the variations on the definitions of "balance" and "fairness" (read the comments below the main article): It is there for anyone who cares to look, so in the public domain, and provides for some interesting perspectives on Match Making, and shows how the debate continues- over 5 years and still a thorny issue, but never the less- enlightening:

 

https://forum.rockpa...-is-rigged/6718

 

Excerpts:

"I’m not sure I understand the problem. Players who lose a lot get given easier games, players who win a lot get given more challenging games. Seems to make sense.Are the microtransactions essential?

The problem with the system is that the players who play well, get given impossible odds. The game is unbalanced by its nature, so the issue becomes exaggerated too. The top tier tanks in any match make a huge difference, and when your team has complete idiots there, it becomes virtually impossible to do anything. Essentially, it doesn’t give them more challenging games, it forces them to lose.

 

"I don’t see a problem here. MM is supposed to create fair matches and this sounds like it does just that by weighing players according to their skill (to an extent). MM was broken when the game was released but it reasonably fine now, it could of course be better but for every game that you get screwed by it you get a chance to screw other people so it balanced out in the end and at least there aren’t matches with 4 tier differences anymore.

 

"What’s the point to play a game if your win or loose result does not link with your ability to play ? It’s not like we’re painting there, we play in a challenge that was supposed to smash testicles against, to see who’s are steel and who’s not, not to simply see beautiful colors on the screen.

 

"I’m more shocked that wargaming was allowed to file a patent which essentially describes every ELO-system out there. But back on topic, of course they have to make the game enjoyable for everyone and isn’t what they’re doing also based on any competitive situation in the real world? A sportsteam wins matches and climbs the ladder, their opponents get tougher and vice versa. People have been complaining about the matchmaker from the very beginning, even in closed beta people were [edited]and whining. But hey, that’s what happens if an automated system pits 30 strangers against each other. Oh and not wanting to brag or anything but after 10k games my w/l ratio has always been climbing slowly but steadily and currently sits and 55.5%. What made me quit the game was their move to make ‘premium-ammo’ available for everyone to buy with regular credits, but that’s in no way related to the matchmaker which in my opinion at least is working as intended.

 

"Any sports league will have a distribution where the mean of win/loss is about 50%. If a team plays well(so, it wins more than average), it will be placed in a higher league next season. Since WoT matchmaking doesn’t really work with seasons, they just measure over a period of time.

 

"The entire point of MM is that people have a challenging, but rewarding, experience. In other words better players get harder matches, this is a good thing.

 

"Well, heights are different, because one person being tall doesn’t intrinsically imply another person being short. When you’re talking about games (at least zero-sum games), one person winning obviously implies another person losing, so you would be more likely to find a normal distribution about the middle there - if everyone plays everyone else. The point of matchmaking is to maintain the kind of curve you’d get if everyone plays everyone else, even if everyone is not playing everyone else. If you had bad matchmaking that always pit the strongest available players against the weakest available players, you would see something more like a bimodal distribution."

 

"The problem is that you’re trying to use your w/l/d ratio as your WoT e-peen measurement:

it pulls up the losers […] it pushes down the good players...

 

"Teams should be balanced, we totally agree on that.

But they should be balance for the best to win, for fair play, not to unleash a fail fest.

Despite the teams being balanced your fighting chances are being nullified if bad players in the top vehicles decide the outcome.
You become spectator to who has the most imbeciles at the top.

Balanced teams with a random outcome beats the purpose of team balance in the first place.

 

"I’m really not sure what you’re trying to say here. Balance, to me, implies that good players play against good players, and not so good against not so good. So it may very well happen that two players/teams have the same win/loss ratio even though their capabilities may significantly differ because they’re not playing against the same quality of opponents. So what? It is more important to have everyone playing fair, challenging matches than for better players to have 6 out of 10 matches where they just roll over the opponents.

 

see link for more.

 

 


Edited by Perrigrino, 04 January 2019 - 08:40 PM.


Chuck_norris10 #69 Posted 30 December 2018 - 04:59 AM

    Major

  • Closed Beta Tester
  • 1914 battles
  • 6,374
  • [300SP] 300SP
  • Member since:
    03-02-2013

RNG's are what decides if you win or lose on a slot machine.

I dont think you can beat them but it has been tried.:(

 


 

 


 

Perrigrino #70 Posted 30 December 2018 - 05:06 AM

    Senior Master Sergeant

  • Member
  • 5738 battles
  • 319
  • [OWSS] OWSS
  • Member since:
    10-02-2013

your absolutely right, glad you brought it up:

 

https://www.wired.co...casinos-no-fix/

 

and should point out there are crooks on every continent, maddof, enron, etc..


Edited by Perrigrino, 03 January 2019 - 03:25 AM.


Captain_Underpants53 #71 Posted 30 December 2018 - 07:00 AM

    Captain

  • Member
  • 21604 battles
  • 3,078
  • [A-S-S] A-S-S
  • Member since:
    04-17-2017

View PostPerrigrino, on 29 December 2018 - 11:54 PM, said:

and then there's this, which I include, not to be subversive, but to highlight the variations on the definitions of "balance" and "fairness" (read the comments below the main article): It is there for anyone who cares to look, so in the public domain, and provides for some interesting perspectives on Match Making, and shows how the debate continues- over 5 years and still a thorny issue, but never the less- enlightening:

 

https://forum.rockpa...-is-rigged/6718

 

 

 

Thanks for that.  I was going to start a new thread about why I am getting into so many more matches where I am lower tier to the other players.  Matches where I am higher tier are much more scarce.  I have put this question directly to the devs on more than one occasion.  They ignore it and it lays there stinking like a hot mess.

 

The only one who has even obliquely, but not directly, said anything about the MM was blindfold.  Who assured us all that karma, or win%, had nothing to do with  the MM.  Well, what is the tier your plane assigned to if not MM?

 

So again, I ask the developers to answer the question.  What determines your plane tier in any battles?  Thanks in advance for ignoring the question.


MSgt, USAF, (ret)

Perrigrino #72 Posted 30 December 2018 - 07:55 AM

    Senior Master Sergeant

  • Member
  • 5738 battles
  • 319
  • [OWSS] OWSS
  • Member since:
    10-02-2013

View PostCaptain_Underpants53, on 30 December 2018 - 07:00 AM, said:

 

Thanks for that.  I was going to start a new thread about why I am getting into so many more matches where I am lower tier to the other players.  Matches where I am higher tier are much more scarce.  I have put this question directly to the devs on more than one occasion.  They ignore it and it lays there stinking like a hot mess.

 

The only one who has even obliquely, but not directly, said anything about the MM was blindfold.  Who assured us all that karma, or win%, had nothing to do with  the MM.  Well, what is the tier your plane assigned to if not MM?

 

So again, I ask the developers to answer the question.  What determines your plane tier in any battles?  Thanks in advance for ignoring the question.

 

No problem cpt.

 

From the game Tutorial 8 video, I surmised that the PRNG algorithm in WoWP MM uses a similar  "SEED"  (also info from Random.org), which may account for why we are seeing this "pattern", therefore not random, but there is randomness- to what degree?. 

 

What would be a better solution, I think, or at least one possible of many, is to reconfigure the algorithm so that players ALTERNATE between +1 up and -1 down, with each consecutive battle. That may at least alleviate some of the frustration of being downtiered, numerous times consecutively but removes some of the competitiveness, challenge and personal growth which seems to be an important element in advancement. Another option is to do away with TIERS completely. Another is to include TYPE of vehicle.

 

There are options....


Edited by Perrigrino, 30 December 2018 - 08:05 AM.


Captain_Underpants53 #73 Posted 30 December 2018 - 08:30 AM

    Captain

  • Member
  • 21604 battles
  • 3,078
  • [A-S-S] A-S-S
  • Member since:
    04-17-2017

A few hours ago I started to keep a written tally of how many times I am lower tier VS higher tier.  So far I am low man on the totem pole 10 times VS big man on campus 4 times.  There must be a better way for the MM to work.

 

I am going to keep this tally going.  For my own edification if nothing else.  Again, developers, what determines your tier in a battle?   blindfold, care to respond?


MSgt, USAF, (ret)

Perrigrino #74 Posted 31 December 2018 - 01:16 AM

    Senior Master Sergeant

  • Member
  • 5738 battles
  • 319
  • [OWSS] OWSS
  • Member since:
    10-02-2013

that's an excellent idea, keeping your own log of how often you were up-tiered / down-tiered and battle results. Would be great if more people did it. One could "almost" design an experiment around this. Problem is the sample group. Regardless, we know that the developers are gathering data. Is this an example of the kind of data that is gathered by them, analyzed by them? Where (which tier) players actually end up based on the assumptions built into the algorithm, and the resulting experience- win/loss/close call/slaughter.

 

It  is important for Wargaming that players spend time playing the game and thereby invest into it.The experiences of the players, as acknowledged in the patent's stated goals are very important, in terms of values such as fairness, competitiveness, balance, and fun. 

 

So Blindfold, can you let us know if the data gathered are used to analyze the end result of MM balance- Tier? battle results? Because so far the experience of the resent changes to MM in terms of distribution amongst the TIERS appears to be causing some serious problems. Whether that is true or not may be in the DATA?  

 

 


Edited by Perrigrino, 31 December 2018 - 01:36 AM.


Perrigrino #75 Posted 31 December 2018 - 06:02 AM

    Senior Master Sergeant

  • Member
  • 5738 battles
  • 319
  • [OWSS] OWSS
  • Member since:
    10-02-2013

fun food for thought:

 



Captain_Underpants53 #76 Posted 31 December 2018 - 07:21 AM

    Captain

  • Member
  • 21604 battles
  • 3,078
  • [A-S-S] A-S-S
  • Member since:
    04-17-2017

View PostPerrigrino, on 31 December 2018 - 01:02 AM, said:

fun food for thought:

 

 

Interesting!   :medal:
MSgt, USAF, (ret)

Captain_Underpants53 #77 Posted 31 December 2018 - 07:43 AM

    Captain

  • Member
  • 21604 battles
  • 3,078
  • [A-S-S] A-S-S
  • Member since:
    04-17-2017

View PostPerrigrino, on 30 December 2018 - 08:16 PM, said:

that's an excellent idea, keeping your own log of how often you were up-tiered / down-tiered and battle results. Would be great if more people did it. One could "almost" design an experiment around this. Problem is the sample group. Regardless, we know that the developers are gathering data. Is this an example of the kind of data that is gathered by them, analyzed by them? Where (which tier) players actually end up based on the assumptions built into the algorithm, and the resulting experience- win/loss/close call/slaughter.

 

It  is important for Wargaming that players spend time playing the game and thereby invest into it.The experiences of the players, as acknowledged in the patent's stated goals are very important, in terms of values such as fairness, competitiveness, balance, and fun. 

 

So Blindfold, can you let us know if the data gathered are used to analyze the end result of MM balance- Tier? battle results? Because so far the experience of the resent changes to MM in terms of distribution amongst the TIERS appears to be causing some serious problems. Whether that is true or not may be in the DATA?  

 

 

 

Well, my results from my first day of logging tiers.  I flew all fifteen of my bombers until they all had a victory.  So I wound up with 15 victories and 7 defeats,  Not very good but, whatever.

 

Of course, this is Tier III through Tier VIII.  I was higher tier 10 times for 7 victories and 3 defeats.   I was lower tier 12 times for 8 victories and 4 defeats.

 

The surprising thing, to me, was I wasn't lower tier nearly as often as in the last few days.  But maybe my perception and recollection are askew.  After all, I started this project simply because of how much more often I was lower tier.  I will continue in hopes of figuring it out.

 

Another thing I noticed though was when I was lower tier I usually didn't have all that much impact when we won.  However, when we lost, I frequently came in first in defeat.  Go figure.

 

I need more data to break it down to how many times on each tier I was the lower tier.  Not enough examples to show a trend about which tiers are the most likely to get you down tiered.  Must go fly, 'need input!'

 

 


MSgt, USAF, (ret)

Perrigrino #78 Posted 31 December 2018 - 09:34 AM

    Senior Master Sergeant

  • Member
  • 5738 battles
  • 319
  • [OWSS] OWSS
  • Member since:
    10-02-2013

thank you sir,

as my father use to say, "don't make kakaloosions" (conclusions) just yet- (and please don't take that personally, no slight intended, I was just reminiscing out loud). Wait til you get more battles in.

I too had one of those battles in bomber, where just me and a new guy in T3 bombers, who hit nothing, I bombed the crap outta everything, and his bots won the match, even though he only scored 1K to my 9K. The RNG works in setting up matches but also in damage. If his bots got an edge on mine, by taking them out, then I could see how that would remove my bots from the active population early on till they respawn, but the size of the discrepancy is hard to ignore and right now is all just speculation.

 

Here is another tidbit about RNG from- PC Gaming Discussion,

entitled, People are probability blind - and what games do to manage expectations 

which includes "other" developers talking about various games and probability and PRNG. Very interesting juxtaposed to this discussion

 

:https://forum.rockpa...ations/13608/55

 

Excerpts:

"Yup, which is (again) teaching people the wrong lesson about probability. They’re stacking the deck so they won’t be accused of stacking the deck - it would be funny if it weren’t so sad.

Firaxis have been doing this for years. Sid Meier once gave a speech about the winner paradox - how most people only complain when they lose, and when they do lose they feel cheated, even if it’s actually all fair and above board. I think it was Civ Revolutions in 2008 when they began adjusting the math to lessen the likelihood of repeated failure for the player.

 

"This strikes me as a system that’s only going to cause the bulk of players, who already seem to have a hard time grasping probability, to have even less realistic expectations the next time they encounter a game with true percentages.I mean, seriously, every luck-based game (Blood Bowl, XCOM etc) seems to have a seething forum full of people that think the AI must be cheating because 90% should mean a guaranteed hit. Speaking as an educator it’s legitimately horrifying.

When it comes to games is that players already know games already “cheat”. Games raise health and damage to create tougher enemies, give the AI side bonus or even infinite resources, read position and/or state data directly to determine AI reactions, etc… With that knowledge, it isn’t unbelievable to suspect that the game is also fixing random rolls. Funnily enough, games that actually rig their random rolls are most likely attempting to favor the player.

There also still seems to be a general distrust of pseudo-random number generation. People are more willing to trust physical dice rolls than they are a game’s RNG (and even then people will question whether the dice are fair.) Play a board game where one person makes every roll again and again against overwhelming odds, and people chalk it up to luck (particularly if they keep swapping dice around). Play a computer game where the same thing happens, and people believe the game’s RNG is busted or the game is cheating. (Differences are that you can’t actually see what the computer is doing the way that you see a physical die being rolled, and that you can’t “swap dice” if someone does question fairness.)

And, to be fair, the RNG in a game at its best isn’t going to cover the full range of possibilities. In reality, something that has a 90% chance of working might work 1000 times in a row, but it could potentially be impossible to have such a run. It doesn’t help that some games have worse designed RNGs than others, and some have particularly poorly implemented usage of an already iffy RNG design.

Two other big psychological issues are that success/failure is often binary and games end up in situations where the percentage chance doesn’t necessarily match what the player would believe to be realistic. These two can feed off of each other, such as when your trained soldier fails to even scratch the unarmored foe that he is standing right next to. In a real-time game, that kind of shot wouldn’t be guaranteed, because the target might through luck or skill dodge the shot. In a turn-based game, the player seems the foe just “standing” there are point-blank range, so you already have the player having a different perception of reality than what the game has. And then the player misses, and … nothing… The player gains nothing, because either he would hit or miss, with no middle ground. He won’t inconvenience the opponent (who realistically would have been spending time moving around while under attack), and he can’t scratch the opponent unless he hits (where a hit in turn may trigger a second check of the RNG for how much damage the hit delivers. So even a hit could itself turn into a scratch.)

And then you top all this off with a system that puts a lot of importance on a limited number of random checks, so things don’t get to even out over time. A single check breaking the wrong way, much less a full turn’s worth of “bad luck”, can wreck a scenario.

 

 

 


Edited by Perrigrino, 03 January 2019 - 03:27 AM.


Perrigrino #79 Posted 03 January 2019 - 01:26 AM

    Senior Master Sergeant

  • Member
  • 5738 battles
  • 319
  • [OWSS] OWSS
  • Member since:
    10-02-2013

Finally, after a couple of weeks of searching through the mighty internet and I found this gem. Wish someone had mentioned it earlier, since it is in the neighbourhood of WG, although it does not specifically address the issue of MM, it is about PRNG within the context of WG title:

 

"The accuracy of your shot is determined by a Random Number Generator. The formula is generated by a hamster on a wheel in each WG server. " -      :teethhappy:               http://wiki.wargamin...attle_Mechanics

 

Wait, there's more..

 

Accuracy and DispersionEdit

{name}
Standard Deviation Diagram

Every shot you take is dispersed randomly around the center of your aiming reticle, i.e. you will not necessarily hit exactly where you aimed. The actual dispersion amount is based on a Gaussian (normal) distribution curve and depends on your gun and the turret it is mounted to.

 

Gun AccuracyEdit

The accuracy value for a gun is given in meters at a range of 100m. The lower the value the more accurate your gun is. The value describes 2 standard deviations σ from the center of your aim. In other words, for a gun with 0.32m effective accuracy at 100m, 95.45% of all shots will land within 0.32m of the center of your aim at that distance. Dispersion amount increases linearly with distance, i.e. 0.32m effective accuracy at 100m translates to 0.64m at 200m and 1.28m at 400m

 

Follow the links above to review the entire article.

 


Edited by Perrigrino, 03 January 2019 - 07:50 AM.


Perrigrino #80 Posted 03 January 2019 - 01:47 AM

    Senior Master Sergeant

  • Member
  • 5738 battles
  • 319
  • [OWSS] OWSS
  • Member since:
    10-02-2013

wait there's more, mentioned by many here in 4ums, again this is more related to in game PRNG vs MM PRNG, but interesting none the less (from link above):

 

 

Client vs ServerEdit

Shots that fall outside of your aim or go into a completely different direction are usually caused by network or server lag. This can happen because every movement of your aim on your client has to be transmitted to the server first and executed there as well. So regardless of your latency the server aim will always lag slightly behind your aiming on the client. If you press CapsLock + 0 during a battle, your client will show an additional blue reticle which shows the last information the client has received about where the server is currently placing your aim, as well as the size of the aiming circle on the server. Since this is subject to latency as well, it is not entirely accurate either, but if you are having latency related problems it can be a good idea to use the server reticle and wait until server and client reticles match before you fire a shot. The server reticle is not displayed while Automatic Aiming is engaged.

Client vs ServerEdit

When you fire a shot, your shell trajectory is first calculated based on the data the client has. It is later updated with the data confirmed by the server. If the difference was big, then you may see shell tracers leave your barrel at an angle or change flight path mid-trajectory.

It is unclear whether only the aiming point or also the dispersion of the trajectory around this centre point is corrected based on server data. At least in v0.7.1, if you watch a replay, you can notice that each time you play it, your trajectory will be slightly different. This indicates that either dispersion is not synchronized between client and server, or that it is during a battle but that data is missing in the replay files. In any case, currently replays cannot be used to find out where your shot went exactly.

Ghost ShellsEdit

Players also like to claim ghost shells, i.e. shells that either disappear straight out of the barrel or pass straight through a target as if it wasn't there.

The first type is typically due to the player missing the visual tracer effect - watching the battle's replay from a different angle will reveal it to you.

The second type can in rare cases be caused by heavy lag in the client/server connection that causes a synchronization loss - what you are seeing on the client is not actually what is happening on the server, and what's happening on the server is what counts. It may also be caused by a mistake in the collision model of the opponent's vehicle. The collision model is a simplified version of the visual model of the target vehicle, but separated into various hitboxes. If two hitboxes are not perfectly aligned, they may create a void between them that a shell can pass through if it flies through at just the right angle. Finally, for some vehicles there are parts of the visual vehicle model that are intentionally not reflected in the collision model and a shell can pass straight through them. This is often the case for elements fastened to a tank's external hull, e.g. boxes or fuel tanks. In many cases these are just decoration.


Edited by Perrigrino, 03 January 2019 - 03:30 AM.






Also tagged with MM, Match Maker, Devs, Patent

1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users