Strategy weights

[EFHIII]

(Recommended: complete #9 before starting)

A way to have weight associated with certain strategies e.g. exampleStrats.random 20 for twenty Randoms

Possible implementation:

• Config file that gets analyzed by a separate script that looks at cache.json

Nice to haves:

• config file is as minimal & human-readable as possible and easy to edit by hand
• config file allows commenting out lines
• A reworking of results.html to work with weighted strategies
• A way to edit the weights inside of results.html
• A way to specify a weight range (as apposed to a specific value)
  • A way to visualize said weight range

[nekiwo]

Yes, I think this would be really cool as well. I hope someone implements it, but if I have time at the end of the day I might do it myself

[EFHIII]

A not about weight ranges: Each range adds a dimension to the results space; if you consider 0 ranges 0D, than 2 ranges is 2D

So one possible way to visualize 1 one weight range is with an image that assigns a color to each strategy, and going left to right, displays the colors top to bottom in an image. So if one strategy is 1st when there's 0% random but last when there's 99.9% random, you'd see on the left of the image, red is at the top, and going right, red moves towards the bottom of the image.

For 3 ranges, you could have a 2D matrix of these images, which would let you see everything, but it'd be a bit cluttered.

Alternatively, you could only display the color for whoever gets 1st which would make seeing 1 more range dimension easier at the cost of losing the position dimension.

[tesslovescode]

Working on an independent tool to do this

[tesslovescode]

Just so that we're not blocked on #9

[tesslovescode]

Submitted a pull request. Also, the independent tool can be found here: https://github.com/tesslovescode/PrisonersDilemmaTournament/blob/main/simulate_strategy_distribution.py

[tesslovescode]

(Recommended: complete #9 before starting)

A way to have weight associated with certain strategies e.g. exampleStrats.random 20 for twenty Randoms

Possible implementation:

* Config file that gets analyzed by a separate script that looks at `cache.json`

Nice to haves:

* config file is as minimal & human-readable as possible and easy to edit by hand

* config file allows commenting out lines

* A reworking of results.html to work with weighted strategies

* A way to edit the weights inside of results.html

* A way to specify a weight range (as apposed to a specific value)

  * A way to visualize said weight range

The current implementation reads results.txt It's configurable through some constants (minimal, editable) Can edit the weights within said constants No visualization yet, just ranks the strategies for each meta

[tesslovescode]

Here are the test files. This tool works for both the results.txt generated from this repo and the results.txt generated from the Cary's repo test_original_repo.txt test_discord_repo.txt