I knew it!
No all those fucktards who said I was wrong and flamed me can shut the hell up.
One more thing about the dice.
Moderator: Community Team
Forum rules
Please read the community guidelines before posting.
Please read the community guidelines before posting.
-
TuckerCase
- Posts: 72
- Joined: Fri Feb 03, 2006 8:55 pm
hi lack,
i'm statistics-trained and a programmer, and i'd like to know more about exactly how you use the dice file. in particular:
i) when there are less than 3 v 2 armies, which numbers do you take?
ii) how does the program calculate which line to take using the number from random.org?
iii) if the purpose of using a dice file is to ensure a better short-term spread so that players "feel" more randomness, wouldn't you rather want to reduce the size of the file rather than increase it?
I have no proof, but I think which we as computer game players might get too used to expecting a certain "reasonable" kind of spread in our random numbers because - i suspect - the pseudo-random algorithms we normally play with deliver "too-perfect" distributions.
i'm statistics-trained and a programmer, and i'd like to know more about exactly how you use the dice file. in particular:
i) when there are less than 3 v 2 armies, which numbers do you take?
ii) how does the program calculate which line to take using the number from random.org?
iii) if the purpose of using a dice file is to ensure a better short-term spread so that players "feel" more randomness, wouldn't you rather want to reduce the size of the file rather than increase it?
I have no proof, but I think which we as computer game players might get too used to expecting a certain "reasonable" kind of spread in our random numbers because - i suspect - the pseudo-random algorithms we normally play with deliver "too-perfect" distributions.
-
lackattack
- Posts: 6097
- Joined: Sun Jan 01, 2006 10:34 pm
- Location: Montreal, QC
Each line has 5 numbers, A1 A2 A3 D1 D2 so if it is 2v1 I discard A3 and D2.smloh wrote:when there are less than 3 v 2 armies, which numbers do you take?
It always takes the first line, and then deletes it.smloh wrote:how does the program calculate which line to take using the number from random.org?
I could ask random.org for a dice roll in realtime each time you click attack. But that would slow things down, and if random.org was down for maintenance Conquer Club would be screwed. So I built up a list of 500,000 dice rolls in advance and saved it as a file. I didn't doctor the file in any way, and I recycle it because I don't want to have to build new files all the time.smloh wrote:if the purpose of using a dice file is to ensure a better short-term spread so that players "feel" more randomness, wouldn't you rather want to reduce the size of the file rather than increase it?
And by the way, numbers from random.org are not pseudo-random. They are randomly derived from fluctuations in atmospheric pressure.
sounds theoretically fine. i don't think there's anything wrong with it. of course, it's always going to be possible to analyze one part of the file or other or even the whole thing and find that the percentages don't tally to the "ideal" expected distribution, but that is exactly what is to be expected from random distributions.
what i was saying is that perhaps people are too used to pseudo-random rather than the true random from random.org. from what i know, i think pseudo-random generally tends to generate a more "ideal" spread, so you might be less likely to end up in the low-probability outcomes. tallying up the pseudo-random numbers from excel tended to get me to a pretty representative normal / uniform distribution pretty quickly when i used to fiddle with them for quick analysis.
so, perhaps the system is too good! i wonder if people would complain less if you just gave them a regular computer-generated pseudo-random number each time. with multiple games going on, it's not like anybody can predict anything anyway.
the other way would be to ensure a "perfect spread" by using a dice file which contains all the 6x6x6x6x6 combinations, and randomly pick one each time. dunno if it would "feel" better on the scale of individual games and 5-10 dice rolls, but there's a chance it might!
what i was saying is that perhaps people are too used to pseudo-random rather than the true random from random.org. from what i know, i think pseudo-random generally tends to generate a more "ideal" spread, so you might be less likely to end up in the low-probability outcomes. tallying up the pseudo-random numbers from excel tended to get me to a pretty representative normal / uniform distribution pretty quickly when i used to fiddle with them for quick analysis.
so, perhaps the system is too good! i wonder if people would complain less if you just gave them a regular computer-generated pseudo-random number each time. with multiple games going on, it's not like anybody can predict anything anyway.
the other way would be to ensure a "perfect spread" by using a dice file which contains all the 6x6x6x6x6 combinations, and randomly pick one each time. dunno if it would "feel" better on the scale of individual games and 5-10 dice rolls, but there's a chance it might!
comming from a computer tech's angle.... I can say there is nothing TRULY random on computers. It's all programmed in, and you can only truly simulate randomness. One method of possibly simulating randomness a bit better is what I call the slot machine method. You basically cycle numbers like the wheels in a slot machine turning. When you click on roll it instantaneously stops the cycling. You'ld have to have each die rotating at a different speed to simulate this correctly. You'ld also need to start the dice at different numbers each time to get away from the feel of the roll that someone might be able to develope. I dont' think you should neccessarily show this graphically... just have it in the background.
The method of removing rolls from a list seems to be not wise... as you are decreasing the possibility of rolls happening again.... even with 500,000 rolls you are in essence basing it more on something that isn't as random as a physical roll of the die.
just some thoughts....
The method of removing rolls from a list seems to be not wise... as you are decreasing the possibility of rolls happening again.... even with 500,000 rolls you are in essence basing it more on something that isn't as random as a physical roll of the die.
just some thoughts....
-
supermarcol
- Posts: 153
- Joined: Mon Jan 30, 2006 7:19 pm
- Location: Montreal, Qc
I don't see anything wrong with picking a dice roll from a list. Appart from the fact that computer generated random numbers are not completely random (although random.org must be pretty close to that) its just like rolling the dice 500 000 times and writing it down, then each time its your turn you take one. You're not removing possibilities, its just that all the possibilities have been preestablished prior to your game. It doesnt make much of a difference if the roll is figured out 2 days ago or now as long as each one is random.
-SuperMarcol
-
fubarcivic
- Posts: 31
- Joined: Tue Jan 24, 2006 5:18 pm
I agree with fishfleas.
I have no beef with the "randomness" or how the dice are designed. I simply play with what I have. However, deleting the first line of 500,000 is eliminating the possibility that the same dice roll can happen again. For example, when we play risk in real life, or any dice related game, although it is extremely rare, it does happen that you will roll the exact same thing twice. However, by eliminating the line of text for that roll, you are taking the possibility of repeated rolls out. Randomness is described as:
Random: Having no specific pattern, purpose, or objective.
By creating a file with 500,000 pre-determined rolls is essentially creating a pre-determined pattern.
Now, having said that, I would much rather have the 500,000 pre-determined rolls than using random.org (no pun). As lackattack mentioned, if random.org goes down, so does conquerclub!
You may not agree with me, but that's my personal opinion. Other then that, great job with the site!!! Cheers all!!!
I have no beef with the "randomness" or how the dice are designed. I simply play with what I have. However, deleting the first line of 500,000 is eliminating the possibility that the same dice roll can happen again. For example, when we play risk in real life, or any dice related game, although it is extremely rare, it does happen that you will roll the exact same thing twice. However, by eliminating the line of text for that roll, you are taking the possibility of repeated rolls out. Randomness is described as:
Random: Having no specific pattern, purpose, or objective.
By creating a file with 500,000 pre-determined rolls is essentially creating a pre-determined pattern.
Now, having said that, I would much rather have the 500,000 pre-determined rolls than using random.org (no pun). As lackattack mentioned, if random.org goes down, so does conquerclub!
You may not agree with me, but that's my personal opinion. Other then that, great job with the site!!! Cheers all!!!
-
supermarcol
- Posts: 153
- Joined: Mon Jan 30, 2006 7:19 pm
- Location: Montreal, Qc
the second line could simply be the same as the first one. There are only 7776 possibilities of roll, and that's counting the rolls that are the same but in different positions. so its bound to happen in the 500 000 rolls file that you get the same roll twice, thrice or more in a row. Although they are predetermined, the predetermined file is random. I don't see a problem with that. I think the rolls are perfect, everyone gets some bad luck once in a while.
-SuperMarcol
-
fubarcivic
- Posts: 31
- Joined: Tue Jan 24, 2006 5:18 pm
Like I said, that was my personal opinion, but I don't mind the 500,000 pre-determined rolls. I would hate to be in the middle of a game, only to find out that it crashed due to something completely out of lackattack's hands (i.e. randmon.org crashing).
I just wish we had the ability to play more than 4 games at once, but that's what you get for being a cheapskate lol
I just wish we had the ability to play more than 4 games at once, but that's what you get for being a cheapskate lol
-
supermarcol
- Posts: 153
- Joined: Mon Jan 30, 2006 7:19 pm
- Location: Montreal, Qc
Hrm.. I read up on the randomness of random.org and I'm impressed.
Maybe for fun I'll analyze the #'s too, and post some fun statistics, being the comp sci/math person I am.
Like.. the frequency, entropy, and general bias towards attacker/defender given x vs y dice.
Pretty sure these #'s wont disappoint though.
Also, instead of throwing out dice... Why not just have them in a constant string of #s?
e.g. 1345543212356 and then throwing out x numbers when x dice are rolled.
Maybe for fun I'll analyze the #'s too, and post some fun statistics, being the comp sci/math person I am.
Like.. the frequency, entropy, and general bias towards attacker/defender given x vs y dice.
Pretty sure these #'s wont disappoint though.
Also, instead of throwing out dice... Why not just have them in a constant string of #s?
e.g. 1345543212356 and then throwing out x numbers when x dice are rolled.
Zip File
I have begun to review the dice file and will post my comments shortly.
In the meantime, my preliminary analysis indicates that there is a very big problem with using this file! I don't want to get into too much detail because contrary to what Lack said, I do NOT believe it is that hard to use this file to one's advantage. Especially during slow play times.
I have to agree with smloh. The right way to do this is to list all the possible outcomes and to have the program randomly select one of these outcomes. THAT is the way to ensure true randomness!
However, I would offer a slight modification to smloh's proposal of listing all 7776 combinations. There are not always 7776 combinations (i.e. some people only attack 2-1). Really, what we need are 4 separate files, one for each of the possible dice that could be rolled - that is to say, during a game there can either be 2, 3, 4, or 5 dice rolled. Each file would contain all the possible outcomes and random.org could randomly select one of these outcomes.
Alternatively if it made the programming easier, we could duplicate the "3 dice" and "4 dice" files and create 6 separate files, corresponding to each of the possible attacks (1-1, 1-2, 2-1, 2-2, 3-1, 3-2). That may eliminate the need for the computer to identify which outcomes belong to the attack and which belong to the defender. But I'll leave that up to you programmers to discuss.
I want to reiterate thought that this file should not be used!
P.S. As a little taste of the potential problems with this file, count how many times the combination 1;1;1;1;1 arises and how many times the combination 6;6;6;6;6 arises.
In the meantime, my preliminary analysis indicates that there is a very big problem with using this file! I don't want to get into too much detail because contrary to what Lack said, I do NOT believe it is that hard to use this file to one's advantage. Especially during slow play times.
I have to agree with smloh. The right way to do this is to list all the possible outcomes and to have the program randomly select one of these outcomes. THAT is the way to ensure true randomness!
However, I would offer a slight modification to smloh's proposal of listing all 7776 combinations. There are not always 7776 combinations (i.e. some people only attack 2-1). Really, what we need are 4 separate files, one for each of the possible dice that could be rolled - that is to say, during a game there can either be 2, 3, 4, or 5 dice rolled. Each file would contain all the possible outcomes and random.org could randomly select one of these outcomes.
Alternatively if it made the programming easier, we could duplicate the "3 dice" and "4 dice" files and create 6 separate files, corresponding to each of the possible attacks (1-1, 1-2, 2-1, 2-2, 3-1, 3-2). That may eliminate the need for the computer to identify which outcomes belong to the attack and which belong to the defender. But I'll leave that up to you programmers to discuss.
I want to reiterate thought that this file should not be used!
P.S. As a little taste of the potential problems with this file, count how many times the combination 1;1;1;1;1 arises and how many times the combination 6;6;6;6;6 arises.
-
supermarcol
- Posts: 153
- Joined: Mon Jan 30, 2006 7:19 pm
- Location: Montreal, Qc
I counted 54 1;1;1;1;1 and 78 6;6;6;6;6. Statistically there should be around 64 of each. Now since this is random, I don't expect it to be a perfect distribution. So for this list we get more 6;6;6;6;6 than 1;1;1;1;1 but it'll all even out in the end. Thats what statistics are, using gigantic numbers. 500000 seems very big sure, but there are 7776 possibilities. As for having user hack in that list, I don't know enough to know if its a threat. Maybe it is. But from a random and statistical point of view, I don't think the way the dice are determined is bad (hmmm maybe droping off numbers does make it wrong, that could be why I always lose to that last 1 , but as far as 3vs2 is concerned I think its good). You need to picture it in your head. Imagine 500 000 monkeys that throw numbers in a room, and then these are put in the order they come out on a big list. If you play with that list, the dice roll will be just like if you threw the dice right then (assuming perfect dice and the heat from repeated throws doesn't change anything 
). I don't think the dice should be changed, but seeing how many people are ready to go to extra length to cheat and better their scores, I wouldn't be surprised if someone tried to predict his roll with that list.
[EDIT]Just though of it, it would be very easy to figure out the next roll
assuming low playing hours.
, but as far as 3vs2 is concerned I think its good). You need to picture it in your head. Imagine 500 000 monkeys that throw numbers in a room, and then these are put in the order they come out on a big list. If you play with that list, the dice roll will be just like if you threw the dice right then (assuming perfect dice and the heat from repeated throws doesn't change anything ). I don't think the dice should be changed, but seeing how many people are ready to go to extra length to cheat and better their scores, I wouldn't be surprised if someone tried to predict his roll with that list.[EDIT]Just though of it, it would be very easy to figure out the next roll
assuming low playing hours.-SuperMarcol
Take it a step further
Supermarcol...Take it a step further.
1. The goal is not perfect distribution since you are correct that everything should equal out over time, but in this case it will not equal out over time since we are using a closed file. We are essentially generating randomness of an irregular file. To simplify, imagine I gave you a 7-sided die and told you the numbers one through six are definitely on there. You don't know what's on the 7th side, so you won't know what number is more likely to come up. The odds will be skewed, but will it be to your advantage or disadvantage?
2. There are not always 7776 combinations - only if you attack 3-2
3. The difference between 78 and 54 is HUGE (over 40%)
4. I'm glad you came to the realization about down playing times.
We need to have randomness from a closed set (i.e. so long as any combination of dice has an equal chance of coming up, we are all happy). The way to do that is seperate files.
1. The goal is not perfect distribution since you are correct that everything should equal out over time, but in this case it will not equal out over time since we are using a closed file. We are essentially generating randomness of an irregular file. To simplify, imagine I gave you a 7-sided die and told you the numbers one through six are definitely on there. You don't know what's on the 7th side, so you won't know what number is more likely to come up. The odds will be skewed, but will it be to your advantage or disadvantage?
2. There are not always 7776 combinations - only if you attack 3-2
3. The difference between 78 and 54 is HUGE (over 40%)
4. I'm glad you came to the realization about down playing times.
We need to have randomness from a closed set (i.e. so long as any combination of dice has an equal chance of coming up, we are all happy). The way to do that is seperate files.
-
supermarcol
- Posts: 153
- Joined: Mon Jan 30, 2006 7:19 pm
- Location: Montreal, Qc