I came across http://game2.com/eote/ quite a while ago. About a month ago I discovered it has an error and the yellow dice aren't quite right. I tried sending an email to the listed address, but have yet to receive a reply. Is anyone here responsible for it or knows who is? It looks like it might use the same code for people who play on Google Hangouts, so I wonder if the yellow dice are messed up on there too.
Broken Die Roller
what is wrong with the dice? really hard to tell since they treat each symbol separately
w what its worth, I say get the actual app. its a great program and worth the 5 bucks. all the dice you'll need for just about any ham, from this to rimperial assault to dnd
He (like me) may not have an Android or iOS device available. That said, there are free iOS emulators for Windows that should run the app. I'm considering giving it a try, though I am also considering just buying it for my brother's iPad and borrowing it for game nights. 
Edit: Late night decisions = bad. Got an iPad emulator running (iPadian). But there is no way to get the app from iTunes into said emulator. Tried to find it on their own custom app store, but it is "under construction."
Going to just buy it for my brother's iPad and call it a $10 app now. Oh well. We'll use it for X-Wing, Armada, and EotE so I guess it's worth it.
I greatly enjoy the app. Cheaper then a set of dice, and I could even use "regular" dice on it. simple interface and nice "bling" factor
Really helped me during a game session where there was a multiple ship race and i could preset and save the dice pools.
Do you mean the probability generator? I'm not a statistician, but it does look a little wonky. On the d12s, according to the probability generator: (rounding to the nearest tenth)
Challenge die has an 8.4% chance of rolling a Despair
Proficiency die has an 8.3% chance of rolling a Triumph
Force die has an 8.2% chance of rolling double dark side pips
Other than that, I can't see what's wrong with the Proficiency dice.
I use Orokos.com to roll dice. It's a nice little program, but maybe more for Play by Post than for Hangouts.
Do you mean the probability generator? I'm not a statistician, but it does look a little wonky. On the d12s, according to the probability generator: (rounding to the nearest tenth)
Challenge die has an 8.4% chance of rolling a Despair
Proficiency die has an 8.3% chance of rolling a Triumph
Force die has an 8.2% chance of rolling double dark side pips
Other than that, I can't see what's wrong with the Proficiency dice.
I use Orokos.com to roll dice. It's a nice little program, but maybe more for Play by Post than for Hangouts.
IIRC, I think the odds are calculated by "rolling" a certain number of times and getting the odds from that. Depending on the sampling size this could be a bit off from the Expected value.
I too noticed an error, but with the red dice probabilities (it should have a lower chance of failure than it rolls). I contacted the site curator via email and he replied that he agreed and would correct it. No luck yet, though.
Edited by MarinerThere is another hangout dice for eote that went a bit unnoticed. It works very well and has some nice features. The other was giving us some sync problems to my group:
I too noticed an error, but with the red dice probabilities (it should have a lower chance of failure than it rolls). I contacted the site curator via email and he replied that he agreed and would correct it. No luck yet, though.
I tried it out several times, and got some pretty widely varying results — much more than would be expected, especially since I was supposedly running it in “Monte Carlo” simulation mode.
I don’t think it’s totally useless, but I would definitely be careful about making any hard-and-fast decisions based on output from this tool, if you were to “roll” it just once. Reload that same roll and run it again a few times, and you should get a sense for how much variation you can expect, and then you can use that to inform whatever decision you might be trying to make.
OTOH, I would like to point out that Litheon’s script is pretty authoritative — it doesn’t simulate dice rolls or anything, it actually works out the real probabilities and puts everything together mathematically.
So, to the precision indicated on screen, you should get the exact expected values. No need to run that multiple times, because the math doesn’t change between rolls, and it doesn’t use a random number generator.
But it is a CLI program, without a nice GUI or WebUI front-end.
Edited by miishelleHe (like me) may not have an Android or iOS device available. That said, there are free iOS emulators for Windows that should run the app. I'm considering giving it a try, though I am also considering just buying it for my brother's iPad and borrowing it for game nights.
Edit: Late night decisions = bad. Got an iPad emulator running (iPadian). But there is no way to get the app from iTunes into said emulator. Tried to find it on their own custom app store, but it is "under construction."
Going to just buy it for my brother's iPad and call it a $10 app now. Oh well. We'll use it for X-Wing, Armada, and EotE so I guess it's worth it.
The problem is that in the roller the red die and the yellow die have the same (opposite) distribution, but on the physical dice the yellow die has one more success.
The creator just got back to me and he has fixed the yellow die. He said he's going to check the code on the Hangouts tool too, to make sure it's correct.
I too noticed an error, but with the red dice probabilities (it should have a lower chance of failure than it rolls). I contacted the site curator via email and he replied that he agreed and would correct it. No luck yet, though.
I tried it out several times, and got some pretty widely varying results — much more than would be expected, especially since I was supposedly running it in “Monte Carlo” simulation mode.
I don’t think it’s totally useless, but I would definitely be careful about making any hard-and-fast decisions based on output from this tool, if you were to “roll” it just once. Reload that same roll and run it again a few times, and you should get a sense for how much variation you can expect, and then you can use that to inform whatever decision you might be trying to make.
Potentially the results could be found without a Monte Carlo method. For example, 2 Ability or Difficulty dice each have 64 possibilities. Same with 1 Ability + 1 Difficulty. The main problem comes from the fact that symbols can cancel each other out, so reducing the 64 down to the true results is a bit harder. I've worked on this issue with my Dice Roller as a basis and found that I needed to extend the tables that held the results for various dice to speed up the sum up and reduction of some of the possible combinations. So, instead of just having a table for each single die then working through the combinatorics of each die with each other die in the stats (3G2Y1B4P or 2G3P2S) instead have tables prebuilt for things like 1G1P and 1Y1R, etc... as the total combinations are actually less than the number of sides per die multiplied together. 1G1P should result in 64 combinations, but when the sets are reduced (cancel successes with failures, advantages with threat), the results will be less than that with some results just being duplicate and so needing to count them more than once instead.
Monte Carlo really depends heavily on the sampling size in each run to try and obtain the regression to the mean as the sampling of samples resulting in an approaching the true stats type response.
Potentially the results could be found without a Monte Carlo method.
Correct, and that’s what Litheon’s program does.
Well, at least his probability generator — he’s also got a random dice roller, but we’re not talking about that one. 
For example, 2 Ability or Difficulty dice each have 64 possibilities. Same with 1 Ability + 1 Difficulty. The main problem comes from the fact that symbols can cancel each other out, so reducing the 64 down to the true results is a bit harder. I've worked on this issue with my Dice Roller as a basis and found that I needed to extend the tables that held the results for various dice to speed up the sum up and reduction of some of the possible combinations. So, instead of just having a table for each single die then working through the combinatorics of each die with each other die in the stats (3G2Y1B4P or 2G3P2S) instead have tables prebuilt for things like 1G1P and 1Y1R, etc... as the total combinations are actually less than the number of sides per die multiplied together. 1G1P should result in 64 combinations, but when the sets are reduced (cancel successes with failures, advantages with threat), the results will be less than that with some results just being duplicate and so needing to count them more than once instead.
I don’t remember the details of what Litheon did in his program, but I know that he was able to find a number of optimizations that he was able to apply to the process.
You might want to take a look at his code and see if there is anything you can “borrow” that would help improve your program, or if there are any additional optimizations you have discovered that he might want to consider.
Monte Carlo really depends heavily on the sampling size in each run to try and obtain the regression to the mean as the sampling of samples resulting in an approaching the true stats type response.
Right, and I think that’s what the program the OP mentioned was trying to do, but seems to have failed at. At this stage, I trust Litheon’s program more than anything else I’ve seen, if only because I’ve seen his code and I actually understand at least some of it.
However, I’d love to take a look at the code for your program, too — if you’re willing.
I've been working on my own roller, off and on today, as an excuse to learn Python. To help my GM (and others), I plan to eventually publish some tables that illustrate what combinations of dice are likely to have at least one success. My intent is that this ought to help sizing encounters.
The game2 roller does not appear to have the failure probability fix to the red die that I noticed.
I've been working on my own roller, off and on today, as an excuse to learn Python. To help my GM (and others), I plan to eventually publish some tables that illustrate what combinations of dice are likely to have at least one success. My intent is that this ought to help sizing encounters.
The game2 roller does not appear to have the failure probability fix to the red die that I noticed.
Heh, I wrote mine in perl using Wx back in the Beta for EotE
It is definitely a good app to practice creating. These dice make for some more interesting issues since it isn't like the normal numeric dice where adding and subtracting is all that is needed, heheh.
The game2 roller does not appear to have the failure probability fix to the red die that I noticed.
What do you think is wrong with it? It seems correct to me.
Heh, I wrote mine in perl using Wx back in the Beta for EotEI've been working on my own roller, off and on today, as an excuse to learn Python.
And Litheon’s roller is written in Ruby. So, we have all three languages represented, which I think is very cool!
If we can all confirm that the respective algorithms used are correct, then we have triple-layer protection against a language-level bug that might potentially cause one program to go astray.
Awesome!
The game2 roller does not appear to have the failure probability fix to the red die that I noticed.
What do you think is wrong with it? It seems correct to me.
From the dice conversion table in the core rule books, a yellow proficiency die has 12 sides, 6 which gives one success (including the triumph) and 2 sides that give two successes. The red challenge die has 5 sides which result in one failure and 2 sides that result in two failures. If you Monte Carlo 1 yellow die roll 100,000 times, it shouldn't have the same success probability distribution as the red die has in failures.
There is a similar difference between the Ability and Difficulty dice, but the game2 roller appears to simulate that correctly.
The game2 roller does not appear to have the failure probability fix to the red die that I noticed.
What do you think is wrong with it? It seems correct to me.
From the dice conversion table in the core rule books, a yellow proficiency die has 12 sides, 6 which gives one success (including the triumph) and 2 sides that give two successes. The red challenge die has 5 sides which result in one failure and 2 sides that result in two failures. If you Monte Carlo 1 yellow die roll 100,000 times, it shouldn't have the same success probability distribution as the red die has in failures.
There is a similar difference between the Ability and Difficulty dice, but the game2 roller appears to simulate that correctly.
Like I said before, the yellow die and the red die were the same in the roller. This was incorrect. The programmer fixed it by adding a success to the yellow die. The red die is fine. If it's still wrong for you, you may need to clear your browser history or something.
Like I said before, the yellow die and the red die were the same in the roller. This was incorrect. The programmer fixed it by adding a success to the yellow die. The red die is fine. If it's still wrong for you, you may need to clear your browser history or something.
Voila! Browser history cleared, it works now! Thanks
Edited by MarinerI got another email from the programmer. Apparently the Hangout version had the same flaw, but is now fixed. I wonder how many online groups will suddenly find their rolls more effective.
I used this format before switching to the FFG dice roller. I didn't notice any glaring faults, and still occasionally use it when making some example rolls