links 12/30/25: https://roamresearch.com/#/app/srcpublic/page/12-30-2025

• https://distill.pub/2018/differentiable-parameterizations/ CPPNs make beautiful image parametrizations 
• https://towardsdatascience.com/reveling-what-neural-networks-see-and-learn-pytorchrevelio-a218ef5fc61f/ layer activations of image classification neural networks 
• https://www.fda.gov/news-events/press-announcements/fda-explores-new-contracting-approach-advance-public-health-innovation not sure what to make of this
• https://reason.com/2025/12/29/the-trump-america-ai-act-is-every-bit-as-bad-as-you-would-expect-maybe-worse/ yep, it's very over-broad.

• https://www.asimov.press/p/clinic-loop detailed information from "failed" clinical trials can be very useful in iterating therapies. worked for CAR-T!