![]() |
| ||
|
Author | Message | Page: 2 1 [all][bottom] |
Egleris member, 187 posts Fri 3 Jul 2020 at 08:34 |
One of the things about true randomness though is that concentration of numbers (both on the high and low end) is a thing that can happen, isn't it? So a generator that can give you a string of low (or a string of high) results is actually one that is working much better than one who'd only give you equally distributed number - the latter, in fact, is one that isn't working properly at all. Personally, in eleven years on the site I haven't really found any prevalent patterns in the rpol roller that are worth speaking of. I've had times where the same number was rolled multiple times in a small sequences, others that were all high, or all low, or all over the place, and I don't think I've ever seen it gamed yet. So... I think it's working fine, for whatever value anecdotal evidence of one person might have. | |||||||
Imladir member, 35 posts Fri 3 Jul 2020 at 08:56 |
In the end (i.e. with a large enough sample) though, you have to get uniformly distributed numbers. It's just that in most cases, we tend to focus on samples that just aren't big enough. | |||||||
Bruiser419 member, 17 posts Fri 3 Jul 2020 at 15:57 |
I'll admit, I did try the site with a different browser, and rolled similar if not worse. It is what it is I guess, even if it's super frustrating. | |||||||
bigbadron moderator, 15905 posts He's big, he's bad, but mostly he's Ron. Fri 3 Jul 2020 at 16:06 |
You will find that if you change your target numbers so that you need to roll low instead of high (eg: roll d20 with a target of 4- instead of 17+)... the roller will somehow know this and will always roll high so it can continue to frustrate you. | |||||||
Zag24 supporter, 640 posts Fri 3 Jul 2020 at 21:26 |
Untrue in practice, as long as the computer has constant human interaction and all you're looking for are dice-sized random numbers. If you take the milliseconds digits at the time when a human has done something, these are perfectly random. | |||||||
facemaker329 member, 7226 posts Gaming for over 30 years, and counting! Fri 3 Jul 2020 at 21:27 |
| |||||||
Imladir member, 36 posts Fri 3 Jul 2020 at 21:36 |
But with the same seed, you will always have the same sequence (which is quite useful when you want to debug a random process). No matter when it's asked, by whom or anything. Hence the pseudo. But we can still consider it random, because it will stick to the desired distribution pretty closely (more or less depending on the algorithm), so it can be considered, for us, random...but once again, that's because we don't know the seed, that's all. | |||||||
Zag24 supporter, 641 posts Fri 3 Jul 2020 at 21:49 |
Absolutely true
Absolutely untrue. You're assuming that the numbers are generated by an algorithm. That's one way, and I agree those are not truly random. But you can also acquire (note that I don't say generate) random numbers from the randomness that humans generate as they interact with the computer. For instance, it took me about a minute to write this note. For the sake of argument, let's say that it is between 40,000 and 120,000 milliseconds. If the computer were timing it, rounded off to the nearest millisecond, and then took that number modulo 1000, it would have a truly random number between 1 and 1000. It isn't the computer generating the random number, it is my interaction with the computer. This message was last edited by the user at 21:52, Fri 03 July 2020. | |||||||
Imladir member, 37 posts Fri 3 Jul 2020 at 22:00 |
In any case, no one generates random numbers like that. You could, theoretically, re-seed a random algorithm before you roll the dice. But it has a cost (in time), so it's not done either, especially for something that's (mostly) good enough as it is. | |||||||
Zag24 supporter, 642 posts Fri 3 Jul 2020 at 22:31 |
This message was last edited by the user at 22:35, Fri 03 July 2020. | |||||||
Imladir member, 38 posts Fri 3 Jul 2020 at 22:37 |
| |||||||
Kagekiri member, 195 posts Sat 4 Jul 2020 at 01:55 |
| |||||||
facemaker329 member, 7228 posts Gaming for over 30 years, and counting! Sun 5 Jul 2020 at 07:21 |
| |||||||
Zag24 supporter, 644 posts Sun 5 Jul 2020 at 16:32 |
| |||||||
facemaker329 member, 7229 posts Gaming for over 30 years, and counting! Sun 5 Jul 2020 at 17:54 |
| |||||||
jase admin, 3764 posts Cogito, ergo procuro. Carpe stultus! Wed 15 Jul 2020 at 14:19 |
Instead I'll point you to what you've all actually rolled -- all of the die rolls analysed into a table for you at https://rpol.net/info.cgi?action=analysedice. We've been recording them for some time now so you can have fun checking the bias of 6.6 million rolls. Correction... the lack of bias... The page talks about colouring but you'll see none as everything is so close to the expected. Enjoy! (c; | |||||||
bigbadron moderator, 15908 posts He's big, he's bad, but mostly he's Ron. Wed 15 Jul 2020 at 15:53 |
| |||||||
slogworx member, 13 posts https://slogworx.com Thu 17 Sep 2020 at 22:40 |
Big fan of the site, and I appreciate that you guys track those die rolls. Lots of fun numbers to play with! We need a basis for comparison, though, which until recently has been difficult to find. My hypothesis is that random number generation used for gaming online typically utilizes cryptography/computer science libraries, and they are TOO RANDOM--moreso than a standard hand-rolled die. This is what skews the rolls from what we expect in a table top game. For funsies I compared 700 rpol.net die rolls to 700 hand die rolls from the cast of Critical Role during campaign 2 (stats available via https://critrolestats.com). The numbers revealed that an rpol.net d20 roll is 10% more likely to roll a 2-10 over an 11-19 compared to a human rolled d20. As your site stats may confirm, the likelihood of rolling any given number is "a fair chance" given total rolls towards infinity. But compounded over time in a limited campaign, the distribution of rolls compared to hand rolled dice means we're 10% more likely to fail at shit. Personally, I find that to be better opportunity for storytelling, as some of our most memorable gaming experiences are when we fail. But it's food for thought regarding random number generators for RPG gaming. If anyone wants to review my data, it's here: https://slog.link/ER This message was last edited by the user at 23:21, Thu 17 Sept 2020. | |||||||
donsr member, 2053 posts Thu 17 Sep 2020 at 23:13 |
It the end it doesn't matter, everyone has the same dice roller.. you can always wait until have your made rolls to do PC house cleaning. For My games i change the outcomes needed.. sometimes you need to roll high..somethings low... the mods i use in my homebrew system withh either add or subtract accordingly. let's see the dice roller beat that! | |||||||
MalaeDezeld member, 125 posts Thu 17 Sep 2020 at 23:34 |
I would explain it the other way: human hand rolling dice isn't random enough. They skews the results toward what they want. Or at least those that aren't hated by their dice ;) Human not being good random generator is the reason dice tower exist. | |||||||
Skald moderator, 911 posts Whatever it is, I'm against it Fri 18 Sep 2020 at 04:39 |
| |||||||
jase admin, 3779 posts Cogito, ergo procuro. Carpe stultus! Fri 18 Sep 2020 at 06:51 |
Coincidentally (and I'm sure you'll be absolutely shocked to hear it) the human rolls favoured high results. 60% of the rolls made with real dice were 11 or over -- so I'd really hope that our stats would be 10% lower. Which (by another great coincidence) puts our 11-20 rolls at 50%. Exactly as it should be! | |||||||
slogworx member, 15 posts https://slogworx.com https://slog.link Fri 18 Sep 2020 at 13:22 |
What I found most interesting is that the frequency of 1s and 20s are pretty much the same. | |||||||
Gaffer member, 1651 posts Ocoee FL 45 yrs of RPGs Fri 18 Sep 2020 at 13:30 |
Well... yes. | |||||||
slogworx member, 19 posts https://slogworx.com https://slog.link Thu 22 Oct 2020 at 23:18 |
I really enjoyed this conversation and the points made. Humans absolutely tend to favor dice that roll well for them, and perhaps that is less fair for determining the story trajectory than a random number generator. It's also kind of satisfying to be able to come up with a theory for why rolling on the site is a different experience from the table. I am happy to have contributed to the subject, and have since embraced my horrible rpol.net rolls knowing the statistical details. May you all fail 10% more than usual, and tell better stories for it! :) |
[top] |