Within the next few days, I expect a huge amount of attempted replication experiments (of various quality) for and variations upon the Weight-Space Programming paper. This will likely be similar to what happened with the large number of neuron steering websites that popped up soon after the "neural networks are already clean" essay dropped. This is likely due advanced AI lowering the barrier of entry to such citizen science style mech interp discoveries, and this trend is likely to continue. It reminds me of the constant (and likely almost all agentic driven) improvements done by members of the community to the time complexity of various mathematical problems after the OpenAI drop (see "https://bigospeedrun.com"). All this generally seems to be a good thing, although the massive influx in both signal and noise may make it increasingly hard for researchers to sort through it all. I expect this trend to accelerate rapidly in the near future.