49 replies to this topic

[Nightbird]

Testing done on testing grounds... if anyone wants to do the test in live, please go ahead.

With the 8 C-UAC/2 build, each with a 17% chance of Jamming on a double tap, we expect to see on a double tap with all 8 guns:

0 Jams 22.5% of the time
1 Jams 36.9% of the time
2 Jams 26.5% of the time
3 Jams 10.8% of the time
4 Jams 2.8% of the time
5 Jams 0.5% of the time
6 Jams <0.01% of the time
7 Jams <0.01% of the time
8 Jams <0.01% of the time

In this experiment, I double tapped, record the number of Jams, waited for all weapons to unjam, and repeated.

I saw the following after 68 double taps:
0 Jams 42 times (61.7%)
1 Jams 7 times (10.3%)
2 Jams 1 times (1.5%)
3 Jams 7 times (10.3%)
4 Jams 11 times (16.2%)
Rest 0 times

Even if the sample size is small, the number of 0 Jams is far too high, as is the number of 4 Jams. The occurrence of 1, 2 Jams are also too low. The probability of encountering this extreme a finding when the Jam chance code is performing correctly is <0.0001, therefore the chance of a bug present is >99.99%

Please feel free to repeat this experiment to confirm the percent of Jams occurring is off.

Edited by Nightbird, 27 September 2018 - 02:16 PM.

[LordNothing]

pgi doesnt fix bugs, they tweak numbers and hope nobody noticed that they didnt fix the bug.

short of a complete uac mechanics rewrite i think were stuck with all over the place performance. like i had one game where my uac10 adder jammed 10 times in a row during a long brawl against a guy with questionable aim. i still got him down to cherry ct with my backup lasers and had i been able to use my uac10 that would have been a decisive victory rather than a shameful defeat. if it had jammed half as much (thats still 5 jams in a row) id have won that easy.

i think uacs should put out at most 1.5x damage but use more sane jam mechanics that doesnt jam cold or as often or for as long. the offitial word is that uacs are for non-sustainable surge damage but the mechanics dont allow for that in practice. it should only really jam if you are spamming it.

Edited by LordNothing, 27 September 2018 - 03:07 PM.

[YueFei]

Nice work investigating this. Would not surprise me if it's true.

Even Blizzard at launch of WoW messed up on Warrior's miss chance. Took awhile for them to admit the mistake, and it was only after players sent proof after parsing their combat logs to show that their actual miss chance was more like ~20%, not the 5% it should have been.

[Nightbird]

LordNothing, on 27 September 2018 - 02:57 PM, said:

pgi doesnt fix bugs, they tweak numbers and hope nobody noticed that they didnt fix the bug.

short of a complete uac mechanics rewrite i think were stuck with all over the place performance. like i had one game where my uac10 adder jammed 10 times in a row during a long brawl against a guy with questionable aim. i still got him down to cherry ct with my backup lasers and had i been able to use my uac10 that would have been a decisive victory rather than a shameful defeat. if it had jammed half as much (thats still 5 jams in a row) id have won that easy.

Let's keep it scientific please, anecdotal evidence doesn't help anyone

[LordNothing]

Nightbird, on 27 September 2018 - 03:03 PM, said:

Let's keep it scientific please, anecdotal evidence doesn't help anyone

of course but long strings of jams like that should never happen. they are frustrating and too random for what is supposed to be a game of skill.i dont want jams to go away they should just happen for a reason other than rngesus says so. you also see long streams of no jams which are equally as frustrating to the enemy and that shouldnt happen either. both are present in your data just not to the extent of things ive seen (and i admit those are outliers).

Edited by LordNothing, 27 September 2018 - 03:15 PM.

[VonBruinwald]

LordNothing, on 27 September 2018 - 03:09 PM, said:

of course but long strings of jams like that should never happen.

But can, statistically.

[Nightbird]

LordNothing, on 27 September 2018 - 03:09 PM, said:

of course but long strings of jams like that should never happen.

The chances of flipping a coin 10 times and getting heads 10 times is extremely low, but the chances of getting 10 heads in a row when flipping a coin 1 million times is extremely high. Without a mathematical context, anecdotes are useless.

VonBruinwald, on 27 September 2018 - 03:11 PM, said:

But can, statistically.

Yep.

[LordNothing]

again thats exactly whats wrong with the uac mechanics. there are no heuristics to determine if the weapon is performing above or below its expected operating performance. when the rng says jam what needs to happen is the game needs to look at the previous jam history (% of shots that have jammed) and if you are performing below the expected performance of the gun then you veto the jam and let the gun fire. likewise if the jam history is much lower then you might want to step up the jam chance to bring a much needed hault to an endless barrage. behavior remains random but now its kept in check.

Edited by LordNothing, 27 September 2018 - 03:25 PM.

[Nightbird]

LordNothing, on 27 September 2018 - 03:24 PM, said:

again thats exactly whats wrong with the uac mechanics. there are no heuristics to determine if the weapon is performing above or below its expected operating performance. when the rng says jam what needs to happen is the game needs to look at the previous jam history (% of shots that have jammed) and if you are performing below the expected performance of the gun then you veto the jam and let the gun fire. likewise if the jam history is much lower then you might want to step up the jam chance to bring a much needed hault to an endless barrage. behavior remains random but now its kept in check.

That wouldn't help if the 8 C-UACs either all jammed (17%) or all not jammed. The data would say it's jamming correctly.

[LordNothing]

if each of the guns has its own performance monitoring it would work. if you monitor them all as a single unit then you can get all kinds of weird results. guns will pingpong between being hot (lucky) and dead (jam happy) independently of eachother. so as you add guns you more or less get the same number working around the same time. adding guns will actually normalize performance a little bit over single guns but it would still more or less have the same damage output per gun.

[Nightbird]

LordNothing, on 27 September 2018 - 05:44 PM, said:

if each of the guns has its own performance monitoring it would work. if you monitor them all as a single unit then you can get all kinds of weird results. guns will pingpong between being hot (lucky) and dead (jam happy) independently of eachother. so as you add guns you more or less get the same number working around the same time. adding guns will actually normalize performance a little bit over single guns but it would still more or less have the same damage output per gun.

So many things wrong with this logic

[Y E O N N E]

Did each round of firing stop as soon as any number guns jammed?

[Nightbird]

Yeonne Greene, on 27 September 2018 - 05:51 PM, said:

Did each round of firing stop as soon as any number guns jammed?

Yes, I double tap, record result, double tap, record result... etc. Gives all guns time between shots to cooldown and unjam.

[Aidan Crenshaw]

I don't think statistics work on such a low sample-size. Let's talk again when there are about 10k samples.

[Nightbird]

Aidan Crenshaw, on 27 September 2018 - 10:06 PM, said:

I don't think statistics work on such a low sample-size. Let's talk again when there are about 10k samples.

It works just fine. If a coin is supposed to land heads 22.5% of the time, and after 68 tosses 42 (or 62%) land as heads, there's a < 0.00000001% chance you have a coin that lands heads 22.5% of the time. Betting against that is pretty much suicide.

I work in medicine R&D, where we test new treatments on people to determine if it works or not, and people die when it lands as tails. We don't have the luxury of 10k body bags before you're willing to call the treatment a dud.

Edited by Nightbird, 27 September 2018 - 10:28 PM.

[ShiverMeRivets]

Nightbird, on 27 September 2018 - 02:09 PM, said:

Testing done on testing grounds... if anyone wants to do the test in live, please go ahead.

With the 8 C-UAC/2 build, each with a 17% chance of Jamming on a double tap, we expect to see on a double tap with all 8 guns:

0 Jams 22.5% of the time
1 Jams 36.9% of the time
2 Jams 26.5% of the time
3 Jams 10.8% of the time
4 Jams 2.8% of the time
5 Jams 0.5% of the time
6 Jams &lt;0.01% of the time
7 Jams &lt;0.01% of the time
8 Jams &lt;0.01% of the time

In this experiment, I double tapped, record the number of Jams, waited for all weapons to unjam, and repeated.

I saw the following after 68 double taps:
0 Jams 42 times (61.7%)
1 Jams 7 times (10.3%)
2 Jams 1 times (1.5%)
3 Jams 7 times (10.3%)
4 Jams 11 times (16.2%)
Rest 0 times

Even if the sample size is small, the number of 0 Jams is far too high, as is the number of 4 Jams. The occurrence of 1, 2 Jams are also too low. The probability of encountering this extreme a finding when the Jam chance code is performing correctly is &lt;0.0001, therefore the chance of a bug present is &gt;99.99%

Please feel free to repeat this experiment to confirm the percent of Jams occurring is off.

Interesting. Thanks for doing this.

I find the trend very puzzling. In the data, a chance for a jam is generally TOO LOW by a significant amount - in theory, you should double tap with no jam at all only 22% (between 1:4 and 1:5), but the data shows you did not jam at all more than half of the time (61%). But when we break down the types of jams there is a huge excess of quadruple jams relative to the other possibilities.

It seems that guns somehow affect each other. The way to test this is to do 68 experments with a single gun and compare the jam chances to the theoretical values.

[Shifty McSwift]

Never been a big fan of double or nothing mechanics like that, random chance stuff is fine in the smaller dosages, as it is far more manageable to balance but as others mentioned that dynamic has ended up quite messy in implementation, I would have thought UACs as stream fire was a good idea for scaling out the difference in potential, and then you could have less jams/more stream etc etc, play with those differentials to suit, but we ended up with half stream acs half stream uacs and its just messy and weird.

[joergen]

How did you calculate the grouped Jam Chance? When each uac has their own 17% calculation the number of grouped jams should be very random from test to test.

Singel Jam Chance on each uac was 13.6% on your test. Did you get through all ammo?
(68 x 8= 544 double taps, 74 jams total)

[El Bandito]

61.7% chance of no jam? I wish!

[Nightbird]

joergen, on 28 September 2018 - 01:20 AM, said:

How did you calculate the grouped Jam Chance? When each uac has their own 17% calculation the number of grouped jams should be very random from test to test.

Singel Jam Chance on each uac was 13.6% on your test. Did you get through all ammo?
(68 x 8= 544 double taps, 74 jams total)

That 13.6% isn't far enough from 17% for me to call it out as a problem. I expect there to be a bug with the randomization code when a group of guns is fired, maybe time is used as the randomization seed and not enough time is passing between the calculation of each individual gun, making groups of them share the same result.

El Bandito, on 28 September 2018 - 03:12 AM, said:

61.7% chance of no jam? I wish!

Do the experiment then, you have an UV