Occultist Zero Heal Frequency, bugged ?

[macnjus]

Latley, maybe over the last 4 days or so, I have noticed that my level 1 and 2 occultists with upgraded skills and gear have been having an awful lot of heals for zero, this being by far the most common number I have been seeing for heals.

One of mine now has a range of 0-14 for heals, I have assumed that the distribution is supposed to be uniform within that range, is that a bad assumption ? For example, with zero being by far the most frequent number, I would say that 70% of the time the heals are between 0 and 4.

I don't recall seeing such low heal numbers until recently, has something changed ? Anybody else notice this ? Is the distribution not supposed to be uniform ?

[wckowalski]

Not sure, but I don't think all skills that have a range of a number as an effect have an equal chance of every number in that range.

[Lu]

Excuse my skepticism, but how large was the sample you took before coming up with the number 70%?

When people come to the forums to complain about something RNG based like this (low heals, enemy crit rate, dodging, etc). it's usually after one or two particularly unfortunate dungeon runs where everything seemed to go wrong. There's already a natural tendency to notice and remember when an Occultist rolls a 0 far more than the countless occasions where they roll a number between 5 and 15 because it's so outrageous and frustrating when it happens. When it happens multiple times throughout several runs it can seem like the game is bugged or otherwise unfairly skewed towards giving the player 0 heals - a suspicion that grows even further when other people come along and complain about something similar happening to them.

Unless you've actually tested this over hundreds of dungeon runs and have statistics that clearly show 0 zero heals being the most common, I would simply chalk it up to bad luck.

[wckowalski]

Also bear in mind there are enough people playing that there are bound to be a few that get phenomenal strings of some sort of bad luck (especially given the number of different factors in the game that could have such a bad streak), and when that happens they are likely to come here and complain about it, so posts like this are bound to happen.

[macnjus]

Yes, I well understand that RNG is RNG. I will start tabulating some data, but my anecdotal is based on the past 30 or so hours of play and it was quite clear to me that zero was by far the most common number over that period.

For example. I had a fight recently that started off with 2 zeroes back to back, which if the distribution is uniform from 0-14, should occur only 1 time in 225. I have observed many such double zeroes over the past several days, and I probably have observed well under 225 heals.

But, as I said, will start data collection on Occultist heals.

[Tom Cruise]

I've always thought it was a weighted scale towards the lower numbers, 0-3's are far more frequent than 6+ in my experience. In terms of data collection, I wouldnt really deem crit heals as high heals as the crit roll is most likely separate from the damage/heal roll. In light of that i'd definitley say the mechanic is weighted to giving you bad heals more foten than good heals.

[wckowalski]

I think it's just you. Remember there are thousands of people playing this, so there are bound to be a few people coming here and complaining about really bad luck on their occultist heals. And yes, sometimes you can get crit heals that heal for less than the maximum non-crit heal.

Also, I'm not entirely sure that all numbers are supposed to have an equal chance, maybe 0 is twice as likely as other numbers or something, but not as bad as you are describing in my experience.

[aardvarkpepper]

Originally posted by macnjus:

But, as I said, will start data collection on Occultist heals.

Yeah taking systematic observations &c is the right way to go about it. Maybe there's an unknown interaction between DD's code and processors or engines or something that makes supposedly random number generation actually less random.

I commend your initiative.

I know that means you'll sleep better at night.

But yeah seriously good thinking, keep it up.

[Answulf]

Confirmation Bias

"In psychology and cognitive science, confirmation bias (or confirmatory bias) is a tendency to search for or interpret information in a way that confirms one's preconceptions, leading to statistical errors."

[wckowalski]

I'm pretty sure you can tell what the spread is if you talk to someone who's worked on modding this, and the info is clear in the game files.

[aardvarkpepper]

@Answulf: You've seen the papers on college students and even professors mapping data on to statistical curves? Turns out a load of studies were flawed because they used computers to map data on and show correlations that weren't there, especially with limited data. 'course I mentioned it was an unsound practice myself decades ago, and I got the same dismissive answers that I'm seeing on these forums today. But bad science is bad science.

Sure confirmation bias is a problem. But there is also possible confirmation bias by people saying there is no problem and that it's all in the OP's head. They're saying that there is no problem, but on what basis do make that assumption (and it is an assumption)? It's sheer ego is what it is. People don't want to deal with something, they say it's all in your head.

Bridge swaying a lot in the wind? All in your head. Oh maybe you see it but no it's got to be safe, it's just got to be. But that's how you get things like the Tacoma Narrows Bridge collapse. All the science up to that point said that it would be fine, then after observation and circumstance proved that not to be the case, the books were rewritten.

Science is not the monolithic infallible entity some apparently think it is. Real world applications show science to be wrong again and again and again. Science once accepted that the sun goes around the earth. It took centuries for that to change.

Well don't think I'm a Luddite, certainly there are benefits to science. But blindly believing that there is some absolute truth is the antithesis of science.

Suppose the RNG (random number generator) code in Darkest Dungeon is written in such a way that it interacts with operating systems or random number generation seeds in such a way that supposedly RNG numbers are actually not RNG but repeat? Maybe it hasn't been seen before. Maybe it only happens on particular systems. But it is not unthinkable that it happens. Even when individual components act in a certain way, combinations of elements can act differently; that's the process of emergence.

Life itself is supposedly a combination of chemical processes. It is pretty ridiculous to shave off a beard hair (or whatever), and look at that discarded hair and think that some combination of processes of those cells could be walking around and eating cheesy poofs (or whatever).

Now is a computer that complex of a machine? How about operating systems? Is it *just possible* that combinations of machines and operating systems and software may combine to create unpredicted behavior?

Well it is, you know.

Even if it's "in the game files" who's to say how those game files are properly processed by the computer &c? Yes in all probability even if the original poster did properly observe and note down statistical deviation, perhaps that's only an outlier result. Why not. But it is not impossible that it is *not* an outlier result, and sneering at the OP is simply inappropriate, though I could think of additional pejoratives, I think it better to leave them unsaid.

now let's hug

[Answulf]

Originally posted by aardvarkpepper:

Sure confirmation bias is a problem.

Glad we agree!

[aardvarkpepper]

Originally posted by Answulf:

Originally posted by aardvarkpepper:

Sure confirmation bias is a problem.

Glad we agree!

I've been outmaneuvered

wp Answulf wp