I thought I'd share this with you all. It is one of the things you miss out on when playing table-top. Unless, that is, you count what passes for a dice rolling substitute in ToEE: http://archive.gamespy.com/comics/dorktower/archive.asp?nextform=viewcomic&id=1147 I think the RNG for this game has all of the "Lucky Dice".
That was funny, isn't that the same strip and character that throws the guys PDA against the wall, when the RNG doesn't go his way? You posted that one elsewhere IIrc. The RNG being such a touchy subject with some people, I will use general discussion to bring this up. ToEE started out with a different RNG then Troika switched it over to this in Patch 2 Beta 5 If anyone knows of a game that has a really good RNG, and we can figure out which RNG it is using. Plugging it into ToEE is not beyond the realm of possiblity. Yes it would take a .dll hack and quite a bit of testing. But I'm just saying... is it worth the effort?
Perhaps some back channel help could be enlisted by those in the know...*cough - Agetian*...so this wouldn't be such a challenge. I for one would rest much better knowing that my friends here wouldn't be getting their dungeon parties licked to death all the time by the giant frogs!
Hey Spike, do u remember posting a link to a comic with a cute little sorceress and her demon familiar (also a cute little thang)? I check out these comix when u post them then forget where they r and months later think "I wonder what's accumulated now?"
The link is broken. I was checking that site a couple of times per week (along with my other comics... ) on the off chance that the guy was going to do some work. Well, he's gotten part 5 done, but it needed to cleaned up & re-inked & whatever else you do with comics before posting them on the internet. Last update was June, then the link went goofy. So, if I find it again, I'll post it here. Shouldn't be hard to find {Pawn, by Frederick K.T. Andersen}; I just need to look. BRB... http://en.wikipedia.org/wiki/Webcomic http://www.pawn.se/index.html The pawn link takes you to Surftown, a pay domain. It may still be be in there, but there are too many free ones for me to do that. Try Hellbound, it's quite funny: http://www.hellboundcomic.com/Comic/Hellcomic.php
In the changelogs for patch 2 there is also a reference to altering the RNG to fix a problem of it never actually rolling maximum numbers (beta 6). The Mersenne twister actually has its own file which asks for feedback etc - despite being 4 years old it might be worth a try. The contact info is in the .txt file mt19937ar in the main Atari folder but it's out of date. Here's the current homepage of one of the co-inventors: http://www.math.sci.hiroshima-u.ac.jp/~m-mat/eindex.html No idea if it's any use, but there is a 2005 version written for 64 bit machines available for download from there as well as an updated version from March 2004 of the one used by ToEE which was done because it was realised that the output often gave results too close to each other...
Hmm, interesting reading. I wonder if it was actually activated? The patch notes say that they can "plug in" different randomizers, so it might be as simple as someone leaving off a flag to activate the MT . . . with the range that this thing claims to have, it shouldn't have clumping problems.
I personally do think that there is a problem with the output. But then, I'm equally aware that this just could be a 'run' of results which are part and parcel of any random event. The vexed issue of when is an odd result an odd result
If there was one best RNG, every game would use it. There is no such thing as a single best methodology to make a perfect or near perfect RNG. There is no such thing as true random numbers, especially when you add die rolls, for example 2d6 or 3d4. In real life, statistics show there are series, sequences and groupings that tend to occur. These go outside of true randomness. The idea of a true RNG is like the idea of cold fusion. IMO - Give people a good RNG and they will find an excuse to complain about it. [EDIT - humor] There is a reason rolling dice in games is called DIE ROLLS.
Well, that's why all the ones on computers are called pseudo-random number generators. ;-) You'll note from the MT website that they also mention that there's no set way to test for the "best" randomizer, but they do have some criteria for what constitutes unnatural clumping or repetition. I've seen just a little too much oddity from the ToEE roller to fully trust it . . . anyone remember the bit with Pool of Radiance 2, where I guess someone claimed that the roller was deliberately tweaked to throw you numbers that would fail in order to make the game more "challenging"?
C'mon folks, I can't be the only guy here who knows that so-called RNG's follow an algorithmic pattern, and that they do 'cause they are created by algorithms. It is a formula, and is therefore goverened by the rules that define that particular fomula ( of which there are many, but few in 'public use' that seem truly random to the casual observer). Mathematics is a stern science which does not really allow for the possibility of 'true' randomness as rolling dice in RL does. Thr fact that any of you notice the lack of true randomness is a tribute to you, and your ability to see beyond the immediate present.
What people probably do not know [rather realize] are some things like a 2d6 die roll provides 36 possible die roll combinations to roll the 2 dice and get a result of 2 to 12 from 36 possible combinations. die rolls for 2d6 -> combos ****************************** Roll 2d6 = 2 -> 1 time / 36 combos = 2.77 % Roll 2d6 = 3 -> 2 times / 36 combos = 5.55 % Roll 2d6 = 4 -> 3 times / 36 combos = 8.33 % Roll 2d6 = 5 -> 4 times / 36 combos = 11.11 % Roll 2d6 = 6 -> 5 times / 36 combos = 13.88 % Roll 2d6 = 7 -> 6 times / 36 combos = 16.66 % Roll 2d6 = 8 -> 5 times / 36 combos = 13.88 % Roll 2d6 = 9 -> 4 times / 36 combos = 11.11 % Roll 2d6 = 10 -> 3 times / 36 combos = 8.33 % Roll 2d6 = 11 -> 2 times / 36 combos = 5.55 % Roll 2d6 = 12 -> 1 time / 36 combos = 2.77 % ****************************** Roll 2d6 = 2-12 -> 36 combinations What this means is that when 2d6 is rolled, there are 6 combinations to roll a 7 and there is just 1 combination to roll a 2 or a 12. This results in the bell curve that was referred to before, and means a 7 can occur 1:6 [16.67 % chance] compared to a 2 can occur 1:36 [2.77 % chance] or compared to a 12 can occur 1:36 [2.77 % chance]. * Randomness should normally behave within certain boundaries [ideal textbook], but the real world [chaotically] would have it go outside of the textbook boundaries [have runs of the same number] to be or have true randomness [the unexpected]. My point is that trying to make a better RNG is probably hard or difficult and would probably be far fetched. If a RNG is broken, it absolutely should be fixed - I agree. Otherwise, it's not even near perfect, but it does work.
Yeah, that phenomenon provides the famous bell curve that shows up in the beginning of the DMG for the ol' 1st Edition of AD&D, when you roll 3d6. The problems is that the Temple RNG tends to throw suspicious clumping numbers on flat rolls like d20. (Clumping on the RNG will tend to make the bell curve on multiple dice rolls more extreme, too.)
We know they switched to the Mersene Twister from the release notes in realease 2, because the original RNG had clumping problems. What we don't know is which version of the Twister used in this game. Looking at the twisters revision notes, it has been updated since 2003 when the game was released, The older versions of the Twister apparently had some clumping problems too, that since have been improved on. Supposedly anyway. I have been trying to find out which version is actually used in ToEE and hopefully will have an answer soon. After that it would just be hacking in the plug-in for the newer version. If that is the direction we decide to go.
