Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Stargrasper]

While I'm sure this is just statistically really bad luck, I keep getting possessed dwarves.  Back in 40d, about three-quarters of all moods were possessions.  Now in v31.0x, ALL of my moods thus far have been possessions.  Okay, clearly possessions must be weighted more heavily than any other type of mood, but this is just absurd.

[Fourdots]

It's just bad luck - the RNG has decided that you have sinned against it, for whatever reason.

Although, come to think of it, possessions do seem to be fairly common in my forts ... though not to the degree you're seen.

[0x517A5D]

Observational selection fallacy, maybe?

I've been meaning to dig into .31's mood code and figure out what's changed, but so far when I have the urge to, I don't have the time to, and vice versa.

So I just attach a debugger, change a certain memory address from 2 to 0, and continue playing.

[Stargrasper]

Observational selection fallacy, maybe?

I've been meaning to dig into .31's mood code and figure out what's changed, but so far when I have the urge to, I don't have the time to, and vice versa.

So I just attach a debugger, change a certain memory address from 2 to 0, and continue playing.

I don't personally see the fallacy.  Considering that 100% of moods I've seen in v31.0x have been possessions, I don't see that there's anything I could be ignoring in the first place.  Of course, maybe it is if you can think of something I'm missing.

I'd love to see what's under the hood in the mood code.  Be sure to let us know whenever you happen to get around to it.

[smigenboger]

Make them miserable, and you may get something else XD

[o_O[WTFace]]

Well, I'm pretty sure that in 40d it was 1/3 chance for a possession and 1/3 chance for each of the other 'standard' moods.  It could have changed in 31.X but I haven't had enough moods to get a decent sample size. 

My guess is, you are that person waaaayyy at the edge of the bell curve who just happened to get the bad end of every random mood roll.  Maybe a kitten soap pyramid would please the RNG god?

[Greep]

It's indirect if I'm right.  Dwarves now work till their thirsty and hungry, so their mood is generally lower.  Possession seems afflicts dwarves that have a moderately lower mood.

[hjd_uk]

Dicegod.
My current fort sees only Strange-Moods, not seen a possesion yet.
Are you on Terrifying terrain? What is the general state of the Dwarves' happiness?

[Karnewarrior]

Yeah, if you flip a coin 100 times, with 1000 people, someone is going to get all-heads or all-tails. Looks like your that person.

[Stargrasper]

Yeah, I agree it's almost surely luck.  Although if you REALLY want to use statistics, I think a hypothesis test would probably demonstrate my sample is beyond chance at any common level of significance.  Sadly, my sample size is too small and I don't know about the normality of the parent population...and my sample would certainly fail an Anderson-Darling test...so I really can't run a binomial probability hypothesis test.  And I don't remember off-hand how to run non-binomial probability tests.  So I can only conjecture that my results are beyond chance.

For those of you that don't understand the above (probably most of you), it's probably beyond random chance that anyone would get my results.  It's statistically possible, but REALLY unlikely.  Still, I don't understand why I would be getting these results.  At least with a fell mood, I lose a dwarf but get a legendary out of the deal...

[msilenus]

Yeah, if you flip a coin 100 times, with 1000 people, someone is going to get all-heads or all-tails. Looks like your that person.

I think this becomes true about half the time if you have 633825300114114700748351602688 people instead of 1000, or else 10 coin flips instead of 100.  Otherwise, it's usually untrue.

Edit: Sorry --9 flips instead of 100.

[Stargrasper]

Yeah, if you flip a coin 100 times, with 1000 people, someone is going to get all-heads or all-tails. Looks like your that person.

I think this becomes true about half the time if you have 633825300114114700748351602688 people instead of 1000, or else 10 coin flips instead of 100.  Otherwise, it's usually untrue.

Edit: Sorry --9 flips instead of 100.

Statistically, the probability of getting all tails in 100 coin flips is 1/1267650600228229401496703205376, or 7.88860905221012E-031%.  Note, that it's times ten raised to the negative thirty-first... approximatively 0%.  That statistical probability of that happening at least once within 1000 samples is still darn near zero.

[msilenus]

True.  Note the original claim was about getting all tails or all heads,which is why my expected number of people is about half your probability.  Though I think your number makes more sense for what we're talking about re: posession runs.

If the OP has seen a significant number of moods, and hasn't seen a single non-possession, that's pretty interesting.  What "significant" means is a little fuzzy, and would vary from person to person.  I'd probably say that if he saw something like a ten moods without seeing a single non-posession, then something is definitely* up in his fort that's at least skewing the normal probability to something other than 33%.  (If there were non-posessions mixed in there, then it might be observational bias, but the claim was that there weren't.)

* Within .001%.

[Stargrasper]

True.  Note the original claim was about getting all tails or all heads,which is why my expected number of people is about half your probability.  Though I think your number makes more sense for what we're talking about re: posession runs.

If the OP has seen a significant number of moods, and hasn't seen a single non-possession, that's pretty interesting.  What "significant" means is a little fuzzy, and would vary from person to person.  I'd probably say that if he saw something like a ten moods without seeing a single non-posession, then something is definitely* up in his fort that's at least skewing the normal probability to something other than 33%.  (If there were non-posessions mixed in there, then it might be observational bias, but the claim was that there weren't.)

* Within .001%.

I call "significant" to be greater than or equal to thirty observations.  This is because in statistics, it's commonly said that if sample size is at least thirty, you can assume normality of the distribution of the data.  Normality makes testing data easier.

And to be specific, I believe it was 10 moods since switching to v31.0x, I've seen nothing but possessions across various forts in all four v31 builds.  If you're wondering, the reason I've seen so few moods overall is because I've been very busy with real life.  This is a small sample, but it still seems beyond chance.

[Retro]

I've found that my forts vary between either about 75% possessions or  like 10% possessions. Some forts practically every mood is one, some I never get them. I think it might not be entirely random, either. The RNG used to always pick the same words over and over in 40d for fortress and group names, while in 0.31 I'm finding it's a lot more random (for example, I see 'abbey,' 'barricades,' and 'mansion' a whole lot less). Maybe the things Toady was having the RNG do were drawing the same number before and he fixed it? Dunno.