Why not use AI to solve this problem?

[BradKML]

There are tools like AlphaEvolve designed to find better algorithms to solve a problem, and in the case of Axelrod, the job is to consistently beat the tournaments in many forms (different turn sizes, multiple noise levels). https://github.com/codelion/openevolve

Noted this a whole ago, but the stars are aligning with new research strategies #1434 #1468
Also, it is a good idea to add extra constraints like code complexity ("cognitive complexity") #740

Also getting LLMs to:

• add in new strategies in the codebase Desired New strategies #379
• Population change in environment Advanced Moran Processes and Population Dynamics #535
• All the refactoring for NumPy and Python 3.13 (too many issues to count)

Yes there are tools like Claude Flow or BMAD for accelerated but a lot are product-centric not issue-centric, but at the same time SWE-Bench has a good list of SOTA tools like OpenHands that can do similar things https://www.swebench.com