[home]

## Gambler's fallacy dice

The point is to simulate rolling two dice, but speed up convergence to the expected values. It keeps track of previous dice rolls, and weights the numbers to make them converge faster to the expected values. Useful for Settlers of Catan so no one complains about all those 3's that keep coming up more than the 8's.

### Python version

This is in progress. The current version is untested but now has some options.

Minimal version Press Enter to roll the dice.
### Mathematica version

You can see a simplified version on the Wolfram Demonstrations website: http://demonstrations.wolfram.com/RollingTwoDiceWithWeightedHistory/
The script is written in Mathematica, and requires the (free) Wolfram CDF player to use in-browser.
The CDF player unfortunately doesn't work in Linux, so you'll have to download the Mathematica nb or cdf file instead.
The version here is a simplified version of the original one (the original additionally keeps track of whose turn it is and displays the appropriate picture when resources are produced, but currently doesn't run properly in the CDF player). Here are the download links: original simple

TODO: these files do work, but they are in the process of being updated/improved.

- How weights are calculated can be adjusted by the RATE slider (higher number implies heavier weighting for numbers that haven't shown up for a while, so faster convergence but easier to predict).
- Randomness can be adjusted by the RANDOM slider -- this is the probability that any dice roll is actually just random wrt expected distribution, instead of incorporating the weighting and history. One of the checkboxes toggles the option of notifying you whether the roll was random or based on weighted history.
- The FIRST variable ensures that the first FIRST rolls will be random (no history).