facemaker329
 member, 7026 posts
 Gaming for over 30
 years, and counting!
Tue 26 Jun 2018
at 21:22
Re: When you just can't roll well
The discussion of improbabilities reminds me of another story...

A friend/former roommate of mine basically kit-bashed his own game together, cherry-picking aspects of several other systems that he liked, along with a few elements of his own devising.  One of those was his version of Luck.  It was, basically, a die pool hybridized with an attribute.  You could spend a Luck point in various ways, or he could have you roll in your Luck, 1d20...if you rolled lower than your Luck score, Fortune smiled upon you in some way...

When he first started using the system, there was also a rule that if you rolled a 1 on a Luck check, you gained a Luck point, buildingon the notion that those who rely on Luck tend to become lucky.  And, for most people, that worked out great...every session or two, someone might get a Luck point...

Then, one night, because of that rule, my character ended up starting the evening with a Luck of 12...and ended the night with a Luck of 18.  If it had been a standard D20 game, it would have been calamitous.  Next game session, he announced that the Gain Luck on a 1 rule now only applied until your Luck reached 15.  To the best of my knowledge, nobody else has actually gained enough Luck in any of his games for that rule to have any use...
DarkLightHitomi
 member, 1378 posts
Wed 27 Jun 2018
at 04:25
Re: When you just can't roll well
engine:
DarkLightHitomi:
As for the site's rng, I don't know for sure what Jase uses, but if it is a common one, or worse, the default for java/c++/etc, then it is well known for having skewed results in the first handful of results.

I'd be interested in reading details about this, if someone has a source. Searching seems to just bring up questions about how to skew results deliberately for whatever reason.

If it's a well-known issue, I imagine there are well-known ways to address it, and I'd like to understand those.



Not sure where to find it now. It had come up in programming class ages ago, and the professor said that pretty much any pro will use a better rng than the standard library for anything important, such as games. Of course, cryptography requires an entirely more profound rng as well.
engine
 member, 624 posts
Wed 27 Jun 2018
at 05:25
Re: When you just can't roll well
DarkLightHitomi:
Not sure where to find it now. It had come up in programming class ages ago, and the professor said that pretty much any pro will use a better rng than the standard library for anything important, such as games. Of course, cryptography requires an entirely more profound rng as well.

Given that, then, there's no reason to think that this site doesn't use a better generator. That and the fact that those who run this site were no doubt aware that people would be likely to question them over their methods. Sounds like it would be easy enough for them to obtain as good a random number generator as anyone here might need.
bigbadron
 moderator, 15589 posts
 He's big, he's bad,
 but mostly he's Ron.
Wed 27 Jun 2018
at 06:49
Re: When you just can't roll well
Yes.  On a number of occasions Jase has shown that the dice roller is, statistically, equal to real dice (tested over sets of a quarter of a million rolls, or more).

And still people complain that it rolls low.  Except when they need low numbers, when it rolls consistently high.  :)
DarkLightHitomi
 member, 1380 posts
Wed 27 Jun 2018
at 08:24
Re: When you just can't roll well
It is a trait of the default rng for c++ (and I believe java) that the first few results favor low. The issue being only with small samples, seemingly having been missed when first implemented because the original programmers went for easy rather than good.

It is a problem that literally can't be discerned in large sample sizes. And if he simply tested it like we did in class, by seeding once and generating 250k numbers between 1 and 6, well that is most certainly not rigorous, nor sufficient, and known to specifically miss this issue.

I'll see if I can find it again in my book to post the reference.
engine
 member, 625 posts
Wed 27 Jun 2018
at 13:27
Re: When you just can't roll well
DarkLightHitomi:
I'll see if I can find it again in my book to post the reference.

Thanks, I'd really like to understand this. Not that I'm concerned about the rolls on this site, but I've heard this claim before.

Assuming it's true, though, and assuming the roller on this site suffered from the same defect, how would anyone be able to tell that a particular set of rolls were the result of it? Or is the idea that if there's reason to doubt the rolls, then a GM is justified in throwing out results that seem fishy to them?
bigbadron
 moderator, 15590 posts
 He's big, he's bad,
 but mostly he's Ron.
Wed 27 Jun 2018
at 14:11
Re: When you just can't roll well
DarkLightHitomi:
It is a trait of the default rng for c++ (and I believe java) that the first few results favor low.

Then it's probably just as well that neither of those is used here.  :)

Additionally, checking up on old threads it looks like I estimated low.  jase generally tests with sequences of rolls running in the tens of millions.  Additionally he has, in the past, made the testing code available to users, so that they can run their own test sequences, and analyse the results online.

So hundreds of millions of rolls by multiple users, with the result showing that the roller is as random as real dice rolled at a table.

Also, for all those people who used to complain about low rolls, he switched the results display over one time (without telling anybody), so that instead of displaying the roll it displayed (X-roll) where X was the die size+1.  For example,  a roll of 4 on a d20 was displayed as a "17", and a 17 showed as a "4".  If the roller was consistently rolling low, the displayed results should have been high...

People still whined that they kept getting low results.
engine
 member, 626 posts
Wed 27 Jun 2018
at 14:21
Re: When you just can't roll well
bigbadron:
Additionally, checking up on old threads it looks like I estimated low.  jase generally tests with sequences of rolls running in the tens of millions.  Additionally he has, in the past, made the testing code available to users, so that they can run their own test sequences, and analyse the results online.

Cool. I think, though I'm not sure, that this is simply a matter of faith for some. That's why I want to understand DarkLightHitomi's position and the reasons for it, and what it would mean if the roller weren't noticeably flawed.

bigbadron:
People still whined that they kept getting low results.

Ah, stuff like that makes me smile.
V_V
 member, 751 posts
 You can call me V, just V
 Life; a journey made once
Wed 27 Jun 2018
at 14:37
Re: When you just can't roll well
In reply to bigbadron (msg # 37):

I do have a genuine question, which jase may or may not want secret. When you say it's like "real dice", do you mean to say it eventually favors certain dice for certain character aliases? If so, that would be quite astute, meta and crazy, but sort of cool!


Spoiler for my pursuit of rigged dice: (Highlight or hover over the text to view)
I rolled a die using a remote controlled lego tower. IT rolled roughly twenty times a minute, and after many, many, many attempts worked for weeks without the die falling out of the rig, or the rig powering down or getting caught. I did this over the period of 11 months with different dice. Sometimes together, sometimes apart. It rolled roughly 2.5 million rolls, between down time, margin of slightly faster rolls, and multiple dice being logged. In the end, I rolled them all about 1,000 times, by hand, to see a trend. Then put a total of four dice in the rig to roll for 2.5 years. After which the rig finally broke down, and I never deemed to replace it. It was in my room running non-stop, because I liked the sound of the fan, for sleeping and home noise, but eventually forgot about it. *chuckles* Then one day it just stopped, and I thought; what just happened. Apparently dust had just acculumated. Which is when I took the dice out. 


Dice, especially copper and plastic (the most common) favor a side after long enough of rolling. The best example is the platonic solid cube and the octohedron. This is why casinos give away their dice after enough rolls.

I have a d6 IRL that rolls "6" 5/12 times.  So just over 1/3 of the time. I have a couple d20s that roll high (12+) numbers just under 70% of the time. The d20's in fact were not rigged, they just naturally got that way, or if they were, they were rigged before I got them.

So my question, if does the dice roller actually "learn" by the alias used?


Spoiler for if you think I'm a dirty cheater: (Highlight or hover over the text to view)
Anyway, just out of my defense, I never used the rigged dice for anything but meta. The god form D&D Olidammarra used it in a game. I had a box of "charity" dice that players could use to "reroll" a bad roll if they would either donate to the game table (food, drinks, mapping, etc..) or when I was running for the LGS to get them revenue and myself free cards (as payment from the store for raising interest). So I never used them to "cheat" anyone but myself, or the one instance of the god of luck.


Anyway, many people actually think dice rollers are "bad" because of the very fact RL dice show favor.

I actually rolled 10,000 (tiny I know) rolls myself on RPoL a couple times, and just counted the net sum (though I DID record each die, THAT got tedious) with mean. So I know that the dice roller is actually quite accurate to a truly random roll. I think it rolled like 123k for all rolls total sum. Which is slightly above the regression to the mean. Which is to be expected (as in not exactly mean) with only 10,000 rolls.

This message was lightly edited by the user at 14:39, Wed 27 June.

bigbadron
 moderator, 15591 posts
 He's big, he's bad,
 but mostly he's Ron.
Wed 27 Jun 2018
at 15:00
Re: When you just can't roll well
V_V:
When you say it's like "real dice", do you mean to say it eventually favors certain dice for certain character aliases?
Only if said character alias sacrifices sufficient quantities of snack foods and non-alcoholic beverages to the dice gods, who squat on their mountain of 25 sided d20s, the fabled Mount WTF, and laugh every time some mortal rolls the exact number that he needs least.
Genghis the Hutt
 member, 2545 posts
 Just an average guy :)
Wed 27 Jun 2018
at 16:13
Re: When you just can't roll well
I had a friend who bought a new set of dice for a little mini-con that we had.  They were big dice -- the d20 must have been at least 3" across.  Anyway, he kept rolling 1's all night, he just rolled poorly.  He was going to throw the dice away but I asked him to give them to me instead.  I took them home and rolled the d20 enough to determine where it was likely to roll -- 1's most often, but 17's some of the time.

Then when I was about to have a total party kill when I was running a table at a con, I'd break them out and talk about how I loved killing a party and how these were my kill dice.  Tended to be very exciting with either 1's or near crits.

Then afterward I'd tell the table this story. :)
GreyGriffin
 member, 201 posts
 Portal Expat
 Game System Polyglot
Thu 28 Jun 2018
at 18:21
Re: When you just can't roll well
I just wanted to pipe up here and note that streaks of luck have a profoundly greater impact on a PbP environment, where one roll is the culmination of a day's activity, and a "bad" streak of 2-10 rolls can last a week (or longer!), whereas that streak would be "burned out" by an hour's play at tabletop.  And especially, given the apparent quantity of high-rp, low-crunch games played, a bad roll with a good stat can go so far as to undermine a character concept.  (Did your smart character fail the one knowledge roll this week?  Heck, this month?)

No matter how perfect the dice, the extended, drawn-out psychological consequences of bad rolls, especially subsequent bad rolls, is especially stinging to us forum players.
praguepride
 member, 1302 posts
 "Hugs for the Hugs God!"
 - Warhammer Fluffy-K
Fri 29 Jun 2018
at 01:52
Re: When you just can't roll well
Genghis the Hutt:
I had a friend who bought a new set of dice for a little mini-con that we had.  They were big dice -- the d20 must have been at least 3" across.



This becomes a big issue in cheap "novelty" dice. I'm guessing it's poor quality manufacturing because there isn't a high demand for precision giant dice. My guess is these would fail the salt water test hard.

I read a reddit post about a guy who runs PFS and how his players would cheat by bringing in "loaded" dice. They would go to the big bins of dice and test them to find ones that hit 19s or 20s so he would have to have a glass of salt water and whenever someone rolled a crit on a die he would test it.

He had to test the dice repeatedly because they would try to do things like have a regular die, wait for it to get tested and then swap it out for an identical looking rigged die...

Cheaters gonna cheat I guess...


Found the thread:
https://www.reddit.com/r/Pathf..._full_comprehensive/
ChromaticNewt
 member, 10 posts
Fri 29 Jun 2018
at 02:02
Re: When you just can't roll well
GreyGriffin:
No matter how perfect the dice, the extended, drawn-out psychological consequences of bad rolls, especially subsequent bad rolls, is especially stinging to us forum players.


And I think this is the real issue for PbP. The dice roller is what it is and, whether we can comprehend big data or not, it is random.

So, how do we handle what seems like bad luck for an character, particularly in critical situations, when it just seems that they can't do anything right. Systems that have a single all or nothing pass score (rather than a sense of partial, total, or critical success, with a lower range for total fail) seem to be at the root of the problem.

Should we take die rolls away from players completely? Have the GMs roll, and fudge when necessary, and let the player know the result? Do players really want to be able to roll the dice themselves?
Hunter
 member, 1446 posts
 Captain Oblivious!
 Lurker
Fri 29 Jun 2018
at 02:38
Re: When you just can't roll well
ChromaticNewt:
And I think this is the real issue for PbP. The dice roller is what it is and, whether we can comprehend big data or not, it is random.


Technically, computers use an algorithm to generate random numbers.  As such, they're not really random.
praguepride
 member, 1305 posts
 "Hugs for the Hugs God!"
 - Warhammer Fluffy-K
Fri 29 Jun 2018
at 03:30
Re: When you just can't roll well
Hunter:
Technically, computers use an algorithm to generate random numbers.  As such, they're not really random.


If you want to be technical, nothing is random. A die roll hits a number because of a million physics variables acting upon it from air resistance to material density of die and surface and angle of the roll and the momentum on the die etc. etc.

And yet, for all intents and purposes, it's a random enough generation of numbers that over large sets of data all numbers of equal probability of being rolled.

Same thing with an RNG generator. Unless it's a BAD RNG generator (which Jase has published datasets showing his isn't) then it's effectively random enough to qualify as just as random as anything else we use to generate random numbers.
DarkLightHitomi
 member, 1384 posts
Fri 29 Jun 2018
at 04:27
Re: When you just can't roll well
Ha, without a Bigcrush score that says good, I don't consider a rng as anything more than adequate.

I don't know what he uses, but I'd want to see the BigCrush results before I call it "good."

That said, I don't really feel his rng needs to be better than what it is, though it'd be nice.

This message was last edited by the user at 04:28, Fri 29 June.

engine
 member, 627 posts
Fri 29 Jun 2018
at 05:00
Re: When you just can't roll well
ChromaticNewt:
So, how do we handle what seems like bad luck for an character, particularly in critical situations, when it just seems that they can't do anything right. Systems that have a single all or nothing pass score (rather than a sense of partial, total, or critical success, with a lower range for total fail) seem to be at the root of the problem.

In part, but other systems just sort of kick the can down the road. Characters have to be able to fail, and if they're able to then sometimes they will.

As I've probably said already, I think the root of the problem is that the usual ways to fail are uninteresting, at best. At worst, they make the character (and by extension the player) seem foolish. Death is a very common default way for characters to fail, and while a lot of people find ways to make the best of it, it has a lot of problems. What I've found is that if a group can find ways that the players don't mind failing, ways they'd still see as consequences that the characters would hate, but which the players would find interesting and engaging, then failure, and therefore the luck of the dice, matter a whole lot less.

ChromaticNewt:
Should we take die rolls away from players completely? Have the GMs roll, and fudge when necessary, and let the player know the result? Do players really want to be able to roll the dice themselves?

Ugh, no don't hide and fudge. No one is a perfect liar and when the players realize what the GM is doing they're likely to be displeased.

That said, I bet we don't need randomness in RPGs at all. And I'm not talking complete free-form either. It's possible to have interesting, strategic games without randomness - chess, for instance. Although I don't know of any such, I bet there's a way to have RPGs that operate that way: moves and countermoves, and no randomness at all.
GreyGriffin
 member, 202 posts
 Portal Expat
 Game System Polyglot
Fri 29 Jun 2018
at 05:14
Re: When you just can't roll well
engine:
What I've found is that if a group can find ways that the players don't mind failing, ways they'd still see as consequences that the characters would hate, but which the players would find interesting and engaging, then failure, and therefore the luck of the dice, matter a whole lot less.

Sounds like someone needs some Burning Wheel.
swordchucks
 member, 1495 posts
Fri 29 Jun 2018
at 13:20
Re: When you just can't roll well
engine:
As I've probably said already, I think the root of the problem is that the usual ways to fail are uninteresting, at best.

This is why at least some new games (I'm specifically thinking about Delta Green here, but I've seen it elsewhere) have a big warning at the front that you should only roll dice when failure can be interesting.  While you'll never get around a few necessary cases where the fail state is "it didn't work", I prefer it when the PCs don't have to roll in most of those situations and just succeed.

You do still get into situations where absolutely no one is rolling well and the game goes in a weird direction you don't really want to play.  Which is why I also prefer games that have a safety mechanic to shore up against a streak of bad luck.
NowhereMan
 member, 219 posts
Fri 29 Jun 2018
at 13:37
Re: When you just can't roll well
In reply to swordchucks (msg # 50):

Makes me think of the Gumshoe system, which makes finding "core clues" automatic. It's interpreting those clues that matters.
engine
 member, 628 posts
Fri 29 Jun 2018
at 14:04
Re: When you just can't roll well
GreyGriffin:
Sounds like someone needs some Burning Wheel.

I don't need it, I hope. I borrowed the books once and almost couldn't close them quickly enough, though I did like what I saw of the abstract money system. I didn't know it didn't use random numbers, but that's probably not enough to entice me.

I should also have remembered that there's always Amber Diceless Roleplaying, but that always sounded rather kooky to me.

swordchucks:
This is why at least some new games (I'm specifically thinking about Delta Green here, but I've seen it elsewhere) have a big warning at the front that you should only roll dice when failure can be interesting.

True. I'm surprised to hear that Delta Green says that, but good on them. Anyway, yeah, this is something try to apply to every game I play, and it has only improved my experiences.

swordchucks:
While you'll never get around a few necessary cases where the fail state is "it didn't work"

Easily gotten around: "It didn't work, but as a result of your failure, you learn/notice/encounter this interesting thing...." The thing not working isn't boring; the thing not working and nothing coming of that is boring.

swordchucks:
You do still get into situations where absolutely no one is rolling well and the game goes in a weird direction you don't really want to play.

How? Why is that direction a possible one if no one wants to play it?
Hunter
 member, 1448 posts
 Captain Oblivious!
 Lurker
Fri 29 Jun 2018
at 15:55
Re: When you just can't roll well
engine:
ChromaticNewt:
Should we take die rolls away from players completely? Have the GMs roll, and fudge when necessary, and let the player know the result? Do players really want to be able to roll the dice themselves?

Ugh, no don't hide and fudge. No one is a perfect liar and when the players realize what the GM is doing they're likely to be displeased.


I personally prefer to roll on the table where the players can see, I now only use a GM screen for reference and to hide my notes/upcoming material.    Most players seem to prefer it that way, it seems.
DarkLightHitomi
 member, 1385 posts
Fri 29 Jun 2018
at 20:09
Re: When you just can't roll well
For any roleplaying game, the GM's job is to make things interesting and to keep things mobile (to avoid roadblocks that prevent story moving forward, yet leaving it to the players to actually move things onward at their own pace).

That includes when the characters fail.

How many awesome movies are there where the main characters fail and fail and fail till they win? Lots.

For that matter, how often are protagonists common folks that are handling the problem because they are forced to or because no one else will?

To quote Red from Angry Birds "Now I see the fate of the world rests on idiots like me, and that's terrifying."

Far too many people are afraid to be something less than a hero, afraid to be the ordinary person in extraordinary circumstances. GMs handling failure in such a poor way is probably a major cause of that.

Number 1 piece of advice for DnD, "Always fail forward."
praguepride
 member, 1307 posts
 "Hugs for the Hugs God!"
 - Warhammer Fluffy-K
Fri 29 Jun 2018
at 23:08
Re: When you just can't roll well
This has gotten off track but I used to build convention LARPs for years and the rule we followed was rules of threes. If there was something vital to furthering the plot you needed to provide at least three ways of getting there because people miss things or fail to interpet all the time.

Same thing with secrets and information. If you want it to come in play you need three people to know it. Some people forget or hoarde information or dont understand what it means but chances are at least one player figures it out