Wu Kong misses a lot more since a patch couple of weeks ago

I too see a huge problem with WuKong, even though I love him. When fighting normal enemies, I NEVER miss, which is normal. But with the debuff, I’ve seen 10+ misses IN A ROW multiple times in a battle. The math doesn’t add up. No, these aren’t bosses and no, there is no other debuff or special involved. Something isn’t right.

1 Like

It’s still a problem, after the fix. All someone has to do is watch a normal level fight, not a Titan or boss or a raid, and it STILL happens.

This was happening even before the latest patch. I believe 2 patches back.
I just chucked it up to they’re making it much harder for people using wu.

Post vid or detailed math, otherwise you’re just telling a story. (And it seems to be a story that has been pretty thoroughly debunked, above.)

Is it an issue with Wu Kong specifically or host the luck of random generation? It does seem misses increased, but it also seems the occurrence of special ability increase from leveling up a hero has decreased also.

SG patched it so that all misses were displayed; the bug (or feature) was that some of the miss spam was suppressed before.

I too have had my WTF WU! moments where I’ll completely whiff on 8+ tiles on titans, but this isn’t a common occurrence.

That could be the new hot acronym for 2018.
“FUWU!”

3 Likes

I have had an increase in the occurrences of a three hit totally missing, which is the luck on the random generation perhaps. Wish the special ability increases where as prevalent as the Wu Kong misses. :slight_smile:

Yeah but the likelihood of missing all three tiles on a single 3-match is like 10% which is pretty frequent in terms of even a single titan fight.

Sure, I’d prefer a flat doubling of damage from Wu rather than the RNG fest that is progression titans or high-level titans in general, but sloppy math wise that’s basically what it works out to over the longer run… just sometimes everything is going right on a board and Wu F’s it up (though my reaction seeing 1/10 strong color tiles hit when everything is setup perfectly and wasting many seconds in some combination of shock and disgust doesn’t help).

FUWU! is apropos for that situation, but usually I get a good score commensurate with the good board.

1 Like

I posted this in a different thread on the same topic over a month ago. Wu comes with the strange phenomenon of psyching people out so they don’t use him. Hey, I’m fine with that. The more players that quit using Wu, the better for me as I take advantage of his awesome ability (also making it less likely they nerf him if people complain about him).

8 Likes

I see. I was found that unfair to use Wu Kong in raid battle. The enemy us ability of Wu Kong show increase attack power for 2-3 time every attack but when my heroes attack to use ability of Wu that miss attack for 60-70%.
I would like to tell that is fair to battle, is’t it?

No, 3 misses at 32% chance each is .32^3 (.32 x .32 x .32) That should only happen 3.2768% of the time i.e. about 1 in 30 chance so SHOULD be very rare event indeed. Something obviously changed, or the RNG is really crap and each roll of the RNG is NOT independent. That is the most likely situation. The RNG is seeded upon some time clock, and the same seed is VERY likely to happen. So when the first tile misses, then it is very likely for all tiles to miss (since the RNG is seeded to the same value for all tiles).

If so, then this IS a bug, and long standing. It just has shown up recently when they ‘fixed’ the missing ‘miss’ messages.

NOTE, does the triple (or quad) miss happen more often when the shots are horizontal, vs vertical? If so, that may play into the theory of the RNG reseeding itself. If the RNG does that, then the developers should please go back to college level algorithms 101 (i.e. very basic). The RNG used should be a good one (with independent results for each pick, AND there should be no reseeding of the RNG, just take the next value returned.

But seeing triple misses (if wu really is 32% miss) should be rare. quad misses should happen just over 1% of the time, and and 8 gem miss which was posted above, should happen only 1 out of 9094 times (so should be a true unicorn!!)

I was a senior systems engineer for 25 years, so do know w.t.f I am talking about when it comes to properly and poorly designed software. The RNG from what I have seen (only able to look at end results), does appear to be much less than pseudo ‘random’. If SG is going to advertise that all game play is solely governed by RNG, (which in a way really sucks since you may be given good ascension materials, BUT not what you need for a LONG time), then they should at least get the RNG right. It does not have to be a crypto strong RNG, but it at least should be properly designed and implemented to function as properly as possible!

NOTE, I do not think the misses are ‘just’ wu kong. I have seen blinding by Gan ju (25%), Bane (35%) and Justice (35%) also have WAY too many full whiff shots.

For bane/justice (35%), a triple miss should happen 4.29% of the time. For Gan Ju, triple miss should happen only 1.56% of the time. From observations, I see Gan Ju triple misses about 1 out 10 times, which is WAY too high. I will start to focus in just on Gan Ju (I see him a lot in raiding). I will leave him somewhat alone, so let him target my heros, and make some observations over time. Unfortunately I do not yet have Wu Kong (he is on my ‘really’ wanted list to help for titans).

Now, in hind sight, from this message, if blinding worked like this: ALL tiles (of each color), are either hits or misses, and all results from a special (if it is a multi enemy) are either hits or misses), AND the chance for all hit or all miss HONORS the percentage chance, then that would be 100% acceptable blinding. Yep, it sucks when you fire something off and whiff, nothing. But in the same line of thought (for wu), 2 out of 3 times (over the LONG haul), you should get 100% impact WITH wu kongs attack bonus. Possibly it is just trying to intermix each gem with a percentage chance of success (and then F’ing up the usage of the RNG), is the base problem here.

5 Likes

100% agree. A game like this with so many RNG elements should have proper RNG algorithms. Things like streaky Wu misses are annoying. But what really concerns me: if you’re a spender and pay with real money for hero summons, the results also depend on this crappy RNG algorithms. Once I got 4 Carvers out of one 10 pull… (and it wasn’t even a green elemental pull)

2 Likes

If this is a problem with RNG dependencies on VERY close in timing requests, then it could also be possible to impact the 10x pulls (or things like multi win with loot tickets). In the same line of thought, you are REALLY lucky, then you might get 3 Domintia on a single 10x pull, if the timing hit just right.

Just because (IF this WAG is in fact true), multiple fast RNG requests are not always independent, does not mean that OVERALL (on the long haul), that things would not even them selves out. But I do stand by my claim. If the game advertises that all elements in the game are based on the RNG, then they should work to get it right, and get the chances to be independent one pull of the RNG to the next…

2 Likes

Btw my Carvers didn’t came 4 in a row, though. Maybe it was really a coincidence. But nevertheless it’s hard to keep the faith in the RNG algorithms when you always see this streaky behavior…

As for what Zarsten listed. I will do some stats here. I will totally remove 4* and 5* from the mix (I assume he was using the normal epic summon, since he listed it was not green only).

So we are only focused on 3*, and will simply assume that 4* and 5* do not exist. We will also ‘assume’ that each 3* has an even chance of a pull. So there are currently 24 3* (I am also ignoring the new 3* sand special characters). So each chance for a carver is 1/24.


WARNING!!! Skip reading if you don’t care about the math.

So each pull should be 1/24 chance to getting a carver. The ‘raw’ chance for 4 carvers (in a row), would be (1/24) ^ 4, or .0003014% chance. However, that is 4 in a row. Since we have 10 pulls, that 4 could be #1, #2, #3, #4 pull, or #1, #2, #3, #5, etc. So we need to find out how many groupings of 4 events are in 10 pulls. Fortunately, there is a very easy formula for computing this. It is called combinatorial function. The formula for computing how many different 4 events there on on 10 chances, the formula is: C(n,r) where n==10 and r==4 which is 10! / (4! x (10-4)!) which is 3628800 / (24 x 720) which is 3628800 / 17280 which is 210. So there are 210 different ways to pick 4 ‘special’ events out of 10 total events.

Ok, so the real percentage is 210 x .0003014% or .0633% which is about 1 in 1580 tries.

End of math.


SO this happening is certainly rare, but NOT out of the ordinary. If there were 20 reports of this happening over a month, but there were 25000 10x pulls that month, then all would be just fine (a little heavy, but withing statistical normal, due to a very small sample size). But in the same situation, over that same 25000 10x pulls, if there were 65 times where a 4x same 3* hero pull happened, then the results are absolutely dubious, and the RNG implementation is almost 100% the culprit.

NOTE, this also makes assumption that all 24 3* have identical percentage chance of being pulled. IF THIS IS NOT the case, then the math shown above does not work, and would require much more complex math. In that case, pulling 4 bane one would have to use the percentage chance of pulling a bane (which we as end users would NOT know), vs the 1/24 chance if all chances are equal. I do know that percentage chance of a 4* is much less than percentage change of a 3*, and 5* are even less. But I do not know if all 3* have same chance, (or all 4*, or all 5*). there may different percentages built into the chances, and the percentage may vary between players. I always seem to get a lot of about 5 or 6 heros, and I have heard others complain of always seeing a lot of the same heros (but often much different than the ones I see). Could it be that there is a random seed attached to each of our accounts, used to seed this thing, and that is why we often get ‘similar’ pulls ???

It would be nice to know, but I certainly am not going to hold my breath about hearing anything from the SG development crew. I will just keep watching release notes, and hopfully if they do fine things to ‘fix’, there will be a little nugget of information, such as ‘improved overall randomness within the RNG’ or something like that.

But I am pretty much done beating this dead horse. From the outside it does appear that there may be issues in the RNG system (and which have become much more apparent with the ‘fix’ to the ‘miss’ messages). It may only be on very close in timing events, but it may also be larger and more systemic. But now we will just see if SG does anything to improve this, or not.

I do want to run some counts of gem attacks when my heros are hit with Gan Ju. If I put together enough information, it will become apparent if ‘overall’ he blinds at 25% chance. It will also be apparent if there are large occurrences of multiple back to back to back events which fall outside of statistical normal. But other than reporting any hard statistical observations, I will just do the waiting game on SG.

2 Likes

Wu has already been tested and he’s fine. It’s psychological. This kind of post pops up with every update and every update Wu hasn’t changed.

1 Like

And people still don’t understand probability, and interpret it as a guarantee.

If an event has a probability of 1, aka 100% that still isn’t guarantee that the event will happen. If the probability is 0, that still isn’t guarantee that the event won’t happen. And same for every other number in between.

Lottery has insanely low probability of winning, and yet, every now and then, someone gets the jackpot.

If y’all made a thread “I got 4 duplicates in one 10-pull, did you?” I wonder how many folks would show up and me too the thread? :thinking:

Might be a good way to start collecting data…

2 Likes

Those ‘tests’ were that somebody counted the hits and misses and the result was in the expected range. Yes. Nobody denies that. But still the streaky behavior of quite long series of hits and series of misses is NOT fine. Please read the first post of GloriousCodger again.
I’m software developer, too. I have experiences with random ray generatiin for raytracing. I know what I’m talking about. When you look at Wu’s (and other) streaky behavior, it’s obvious that the used RNG algorithms have insufficient quality.

3 Likes