Accidental Repost

mimzy630

I don't know how to delete repeated posts, so I'm doing this instead.
Here's the newest post:
https://forum.deadbydaylight.com/en/discussion/43474/shrine-of-secrets-and-statistics#latest

Cardgrey

@mimzy630 said:
Here's all my data for what I'm about to say: https://docs.google.com/document/d/1Bn3M1tOrXYAdgEaYUj1cp2R8RcgjsD0rI9TTaS94ViQ/edit?usp=sharing . Feel free to check it, criticize it, tell me I'm bad at math, tell me my sources are wrong, or call me an idiot. Even if I made some minor errors counting weeks and such, the general trends are way too pronounced to be ignored.

Hi, I'm Pharos#2088, AKA that guy who complains in #memes about BBQ not being on the Shrine, AKA mimzy630, AKA Aigis of Reason. I've done a pretty in-depth analysis on the Shrine of Secrets week by week, and whether it is "fair" (i.e., does every teachable perk have an equal chance of appearing each week.). Obviously, the answer I've come up with is no, no it isn't. I started by looking at BBQ & Chili, half because the week it was on the Shrine my laptop was broken, and half because I suspected its absence for so many weeks would be a statistical anomaly. Well, we're living in the worst timeline, because at the time of writing this (January 24th, 2019), the chance of BBQ not being on the Shrine since October 31st-November 7th, 2017 is 1.89%. This is an anomaly, to say the least. To say more, this is likely intentional rigging by either BHVR or pressure from Vertex to make more money from the license due to it being $4.99 base price rather than $6.99. Now, I don't know the real reason. In fact, 1.89% isn't that low, and it is possible. For several months now, I've been lurking the #memes channel in the Dead by Daylight Discord, ranting every Tuesday about the new Shrine. However, I hadn't formally typed up my data. Now I have, and now someone can really check for themselves if the Shrine is rigged.

Of course, BBQ being so popular, I thought maybe focusing on it gave me a bias. So, I next did a full analysis week-by-week to see if the chance of paid and chance of free perks on the Shrine was proportional to what was expected by statistics. Naturally, again, it wasn't. No big surprise there. However, the trend was so durable, and at times so large, that it really struck me: 115 weeks and 460 trials netted a -6.7% weighted average, or average fewer paid perks. Normally, this would be pretty minor and insignificant, but with the backing of 460 trials, this represents a trend. I didn't both calculating the standard deviation of significance (R) value because I honestly don't remember how, but any data with such strong negative correlation and over 400 trials at least deserves further review.

To close, I do like this game. I have about 358 hours and counting, but these findings are (pun intended) rather insidious. I'd rather not think that BHVR is actively lying about the Shrine of Secrets being "random" and "fair," but that seems to be the only option.

To any DBD player reading this, I hope you share it. To BHVR Staff, I ask that you do at least one of the following:

Make a statement that the Shrine of Secrets is rigged

Make the Shrine of Secrets truly random

Release the formula that determines the weekly Shrine of Secrets

Notice none of those say "Rig the Shrine of Secrets." If random chance means never seeing BBQ on the Shrine again and having to shell out that $4.99, so be it. I don't care. This game is cheap at $20 with constant sales and I don't mind supporting the devs. The only thing I mind supporting is lying. These are clearly not outliers. Time and number of trials are on my side to say that and make the data significant. This is a trend, and I want to see it gone or recognized.

TL;DR: It’s probably rigged. The chance of getting BBQ, and more important, the overall chance of getting paid perk is far lower than statistics would suggest.

• Pharos#2088

Hell ya it’s rigged... I needs me some shards...
Shards don’t grow on 🌲 lol

Lagoni

@mimzy630 said:
TL;DR: It’s probably rigged. The chance of getting BBQ, and more important, the overall chance of getting paid perk is far lower than statistics would suggest.

Cumulative binomial probability.

mimzy630

@Lagoni said:

@mimzy630 said:
TL;DR: It’s probably rigged. The chance of getting BBQ, and more important, the overall chance of getting paid perk is far lower than statistics would suggest.

Cumulative binomial probability.

Did I mess up horribly? I'm not discounting that, but still, it shouldn't change the overall trend.
Please, I encourage you to tell me how misguided I am.

Milord

mimzy630 said:

I'd rather not think that BHVR is actively lying about the Shrine of Secrets being "random" and "fair," but that seems to be the only option.

To any DBD player reading this, I hope you share it. To BHVR Staff, I ask that you do at least one of the following:

Make a statement that the Shrine of Secrets is rigged

Make the Shrine of Secrets truly random

Release the formula that determines the weekly Shrine of Secrets

I don't want to defend them, but at the same time, it's a misconception that "random" and "fair" have to go hand in hand. Afaik, they never stated anywhere that the shrine is meant to be fair; they just state that it's random. This means that perks don't have to have equal chance to appear, as long as they don't hand-pick the perks, it's still random.

Now, would it be good if all perks have equal chance to appear? Yes, ofc. But they don't need to admit a lie that they never make in the first place. If people get the wrong idea due to bhvr's lack of explanation, then it's the people's fault from starting with a wrong set of assumption.

AlwaysInAGoodShape

@mimzy630

I next did a full analysis week-by-week to see if the chance of paid and chance of free perks on the Shrine was proportional to what was expected by statistics.

It isn't. By the results you can quite clearly see it's RNG. Just look at all the newer killers and to the degree that they are new, their perks have appeared less than others.

---

Make the Shrine of Secrets truly random

Be careful with what you wish for, because this exactly is the problem. You don't want pure randomness. You want a rotation of perks mixed in with some minor randomness.
Similar to games like Heartstone, for every card pack you open there's a chance of getting a legendary. The longer you haven't gotten a legendary, the higher the chance is that you get one. In about every so 50 card packs you open you get a guaranteed legendary.
It's exactly this type of pseudo-randomness that you want.

If you want pure randomness than there's a possibility of a perk never appearing (Surveillance), which from a point of randomness isn't a problem, but from a point of your ability to get it, it is. Randomness is just a concept. If it doesn't appear then you didn't really have a "chance" to get it. You just didn't have a chance. There is only 1 universe and you only experience 1 timeline.
If pure randomness produced let say the same 8 perks for the full lifespan of DBD (highly improbably), then this would be absolutely a shame, rendering the Shrine useless.

You should ask for Pseudo-Randomness instead.

thesuicidefox

Shrine needs to be reworked anyway. Ever since they changed shards it's honestly not worth it, especially if you have more than 50% of the perks unlocked already.

AntiJelly

I don't even think it's rigged, they just made the algorithm pure RNG. There is a chance we'll never have some perks on the shrine. It needs to be pseudo random, like @AlwaysInAGoodShape said.

mimzy630

@AlwaysInAGoodShape said:
@mimzy630

I next did a full analysis week-by-week to see if the chance of paid and chance of free perks on the Shrine was proportional to what was expected by statistics.

It isn't. By the results you can quite clearly see it's RNG. Just look at all the newer killers and to the degree that they are new, their perks have appeared less than others.

---

Make the Shrine of Secrets truly random

Be careful with what you wish for, because this exactly is the problem. You don't want pure randomness. You want a rotation of perks mixed in with some minor randomness.
Similar to games like Heartstone, for every card pack you open there's a chance of getting a legendary. The longer you haven't gotten a legendary, the higher the chance is that you get one. In about every so 50 card packs you open you get a guaranteed legendary.
It's exactly this type of pseudo-randomness that you want.

If you want pure randomness than there's a possibility of a perk never appearing (Surveillance), which from a point of randomness isn't a problem, but from a point of your ability to get it, it is. Randomness is just a concept. If it doesn't appear then you didn't really have a "chance" to get it. You just didn't have a chance. There is only 1 universe and you only experience 1 timeline.
If pure randomness produced let say the same 8 perks for the full lifespan of DBD (highly improbably), then this would be absolutely a shame, rendering the Shrine useless.

You should ask for Pseudo-Randomness instead.

That would be nice, but I'm alright with the Shrine being fully random, so long as it truly is. That would be consistent with what they are saying, which is adequate to me.

AlwaysInAGoodShape

@mimzy630

Then why did you make a post complaining about the shrine? It is already completely random. lol
That is the whole issue.

mimzy630

@AlwaysInAGoodShape said:
@mimzy630

Then why did you make a post complaining about the shrine? It is already completely random. lol
That is the whole issue.

Err, well, the data suggests that it is not completely random, which would at least be fair. Instead, it seems to be neither fair in perk occurrence nor in randomness, with the latter being BHVR's promise.

mimzy630

@AntiJelly said:
I don't even think it's rigged, they just made the algorithm pure RNG. There is a chance we'll never have some perks on the shrine. It needs to be pseudo random, like @AlwaysInAGoodShape said.

If it is random, then what a timeline we occupy. Very anomalyous indeed.

AlwaysInAGoodShape

mimzy630

The data shows exactly that it is truly random.

Depending on when they were released, their perks on average show up less as the pool simultaneously increases.

The most recent killers have their perks on the shrine the least. The oldest killers have their perks on the shrine the most. Everything indicates randomness.

And to be purely theoretical: pure randomness can produce any possible results, meaning that you cannot say that "the data suggests that it is not completely random".

Hoodied

@mimzy630 said:

@AlwaysInAGoodShape said:
@mimzy630

Then why did you make a post complaining about the shrine? It is already completely random. lol
That is the whole issue.

Err, well, the data suggests that it is not completely random, which would at least be fair. Instead, it seems to be neither fair in perk occurrence nor in randomness, with the latter being BHVR's promise.

If its random you cannot tell because the results will also be randomized you could get self care 20 times and rancor twice, agitation 5 times and empathy 6, you will never know

AlwaysInAGoodShape

@mimzy630 said:

@AntiJelly said:
I don't even think it's rigged, they just made the algorithm pure RNG. There is a chance we'll never have some perks on the shrine. It needs to be pseudo random, like @AlwaysInAGoodShape said.

If it is random, then what a timeline we occupy. Very anomalyous indeed.

Nothing strange about it. The more recent the characters, the less often their perks have appeared.
Some perks see overappearance, some see under appearance and there's no deducable pattern.

Thats what you asked for in your OP. Sadly the Devs listened to you and not to me (:

mimzy630

@AlwaysInAGoodShape said:
mimzy630

The data shows exactly that it is truly random.

Depending on when they were released, their perks on average show up less as the pool simultaneously increases.

The most recent killers have their perks on the shrine the least. The oldest killers have their perks on the shrine the most. Everything indicates randomness.

And to be purely theoretical: pure randomness can produce any possible results, meaning that you cannot say that "the data suggests that it is not completely random".

I think you misunderstand. By looking at perk trends over time, over 460 perks, there has been a significant underrepresentation of paid perks. This is, of course, within the realm of probability, but a -6.7% correlation after 460 trials usually tends to denote a trend, not randomness. If the Shrine was random, then it should near 0%, especially with so many trials. It is was between +@/-3%, it would probably be insignificant. It's nearly double that.

Forgive the estimate, I'm not at my computer and can't do standard deviation yet.

AlwaysInAGoodShape

@mimzy630

that is not how randomness works.
Pure randomness could even mean that the shrine would only show 8 perks and all the others would never even appear on the shrine.
Only with pseudo randomness could you expect to see a 0%. that's the whole point of randomness.