change the dice generator.

[Deckeon]

You complain about repeated dice outcomes but you do not give data. If I understand correctly, in "Space Empires" you roll 10sided dice. There is a 1/100 chance of a 3-fold repetition (or a 1/1000 chance of a specific number repeating 3-fold, e.g. 2-2-2). Since this is a rare event you need quite some games (how often are these dice rolled per game?) to analyze.

Please note that rare events cluster if they are random, i.e., several of them may happen in a short time span and then they will not happen for a long time. If a 1/100 event happens exactly once in a 100 time span it is *not* random!

It CERTAINLY happens more than 1 in a 1000. Fair enough, I will try to attempt to get data. Space empires is difficult to get it you must trust me on the data, there is a lack of trust on this topic. Edit: no I can get the data (manually from game logs and will track doubles too as I think that's the actual problem)

If the topic comes up over and over and over and even in Catan showing bias to certain sets of numbers in individual games, over and over, you guys might want to think about be a little more open minded to the possibility that even a very small chance of (artificial/nonrandom) number repetition can be game breaking. We have lost players in space empires on BGA over this. Rage quits and at least as many forum wars over the topic as Catan.

[Jontia]

Good luck; Here's a great write up of the sort of method you'll need to test BGA's catan dice and some tools to help you.

Dice Testing

As covered on the page, you'll need around 300 rolls for a minimum viable test of a d6 I belive. At around 70 rolls a game rough estimate for a base game 4 player catan, you'll need your next four complete games to cover two set assuming you intend to test the red and yellow die individually?

[Bruce Idaho]

I agree. The numbers 4, 5, 9 and 10 always seem to come up more often than 6 & 8 do.

[Jellby]

I agree. The numbers 4, 5, 9 and 10 always seem to come up more often than 6 & 8 do.

Mi guess: always, except most of the time.

Unless you mean *the sum* of 4, 5, 9, 10 comes more often than the sum of 6, 8. Then that's expected (but should not happen *always*).

Another possibility is that key word is *seem*. That's subjective and then we have to take your word for it. If that's what it seems to you, there's nothing to discuss. But it's probably not what others see.

[Suoivax]

I agree. The numbers 4, 5, 9 and 10 always seem to come up more often than 6 & 8 do.

go grab two dice and roll them about 20 times and write down each result.

[BarnardsStar]

I’ve conducted a practical experiment on this subject… initial facts: the site serves about a quarter of a million games per day. Based on number of games over the last 20 minutes and current hourly game rate, Catan accounts for about 3% of all BGA games, so thus about 7500 games per day or 52,500 games per week. The average number of rolls each player does in a game of Catan, according to site statistics, is about 20.

I quickly whipped up some Python code to simulate a 3-player game’s rolls—60 rolls. (3p seems very common; I’m happy to run these numbers for 2 or 4 players.) I then look at that game to see what the most common dice roll was, giving tie-breakers to numbers farther from 7 (since those would stand out in people’s minds more), and furthermore, just in case, a secondary tie-breaker to the lower number (so if 5 and 9 were tied for most rolls, I returned 5). Thus, a single most-common roll for each game.

I then did this 525,000 times (to smooth out the numbers) and divided the results by 10 (to get occurrences per week).

Code: Select all

 2: 1.4,
 3: 130.1,
 4: 1205.8,
 5: 4614.9,
 6: 11092.8,
 7: 19674.3,
 8: 10124.9,
 9: 4358.6,
 10: 1171.1,
 11: 124.5,
 12: 1.6

So yes, you should expect around 3 games per week, somewhere on BGA, where either 2 or 12 is the most commonly rolled number, and 250 games per week where it’s 3 or 11. This is the nature of randomness.

When you can play 40 games of Catan per day (and I don’t know about you but IRL I don’t think I ever had a year in which I played it 40 times), you will see weird stuff happen. It’s pretty normal.

[Jontia]

When you can play 40 games of Catan per day (and I don’t know about you but IRL I don’t think I ever had a year in which I played it 40 times), you will see weird stuff happen. It’s pretty normal.

Played at lunch at work everyday for a couple of years. We saw some pretty weird stuff, that's what random means. You can only draw conclusions from a big set of data. "Look at my last game" is totally meaningless.

But honestly, this is a 12 page thread. And this is a minimum of the 2nd time pages or scripts have shown up to demonstrate how often odd things happen. I don't think you'll get any converts. But it was worth a try.