I have a new twitter - @MaxDiffusionRL
AND a new bluesky - stochastickea.bsky.socia
Some claude logs for exploration
https://claude.ai/share/7173910e-c0a5-4419-ba46-3bd7d4d1c4b5
https://claude.ai/share/e007b850-94fe-4444-9a4f-74a4933763a6
https://claude.ai/share/8fb60d5b-9c03-4e07-89e3-770b06ee7fbf
https://claude.ai/share/7d99d8e8-e7c4-464d-9770-26c0e91278c5
https://claude.ai/share/605befc6-bad8-4630-93cc-dcdf604cb36d (AND MANY OTHER THEMES OF PREVENTING HUMAN X-RISK)
https://claude.ai/share/3794ccb8-bb04-4dda-a05d-4d6bdb04f218
I have a high level of betweenness-centrality/breadth. I sometimes generate useful amounts of entropy (way easier when I’m broadly/sparsely distributed). On occasional bursts, I can trigger exponential growth trajectories in multiple people (eg someone once told someone else “go meet me and something interesting will happen”). I can occasionally be highly entertaining. I like open-ended pre-disciplinary areas of study.
Once upon a time, I was the Quora celebrity that all the Quora employees knew. Adam d’Angelo said “I’m important for Quora”. AI causes global weirdening - maybe some interface with AI may make Quora rescueable again. I still ask the best questions on Quora! https://www.quora.com/profile/Alex-K-Chen
One of my main interests is preventing brainwave drowsiness/resetting brainwaves to “focus” mode when I can (sometimes a special kind of confidence is enough). [though the easiest solution is still ending your long break from adderall]
Retaining plasticity for multiple “cycles” is what I live for! (you may screw up one cycle and may need to wait 4 years for the next “cycle”).
I lean into the “strength of WEAK TIES” - it is a uniquely serendipity-maximizing (if illegible!) strategy. I aim into the feeling of not being boring, even if my strategies sometimes seem to produce some bursts of interestingness punctuated by long periods that seem stagnant.
While straightforwardedness is impossible, I am at least guaranteed to be non-boring in some way.
I am interested in neuromodulation (including way less limited TMS/tFUS), out-of-distribution generalization, polychromic objectives**,** alternative computing architectures, and making fat tailed distributions more Cauchy. I like highly expressive geometric language. Maybe Ilya is right!!