3 ms·

I recently finished "The Cosmic Revolutionary's Handbook"[0], which can best be described as "Here's all the data, please, tell me how your crazy theory fits AL

by mabbo 6y ago

I recently finished "The Cosmic Revolutionary's Handbook"[0], which can best be described as "Here's all the data, please, tell me how your crazy theory fits ALL of it, because it's really hard to do that". I really illuminates the difficulty in building any theory that fits all the data, and the incredible beauty that our current best theory of the universe truly is. Gravitational wave events like these are just more data, making the challenge that much more difficult. And yet, the model still fits pretty well, barring a few pieces. But we don't throw the baby out with the bathwater, because it's still a really good model. And the best part about that is that we can use the model we have to make predictions, and most of the time they're pretty good. Predictions about things we can never see or experience or know. We can see into the past and future. [0]https://www.amazon.com/Cosmic-Revolutionarys-Handbook-Beat-Bang/dp/1108486703 https://www.amazon.com/Cosmic-Revolutionarys-Handbook-Beat-B...

pp19dd 6y ago
Thanks for the read recommendation, it's now my next book on the list.
dnautics 6y ago
I'm skeptical. These days it's increasingly possible to create models with enough free parameters and complicated math that you can sift through noise with the right filters and get what you were looking for. This is even more troublesome for unimodal phenomena. There has only been a singular multimodal observation that correlates with a GW observation; and that's troubling especially since the number of unimodal observations is increasing, and so the statistics of how often we would expect random signals to correlate (with the three prongs of the detector, and then with a multimodal observation) is getting murkier and harder for third parties to review. The truly disciplined experiment is to do a double-blinded study: over a timespan record events in secret. Then augment the GW data with 3x fictional events (say, fix the times and magnitudes but randomize azimuth/altitude). Challenge EM observers to correlate with these events, break the seal, and see if multimodal observations pass a simple chi-squared test.

by mabbo 6y ago

pp19dd 6y ago

Thanks for the read recommendation, it's now my next book on the list.

dnautics 6y ago

I'm skeptical. These days it's increasingly possible to create models with enough free parameters and complicated math that you can sift through noise with the right filters and get what you were looking for. This is even more troublesome for unimodal phenomena. There has only been a singular multimodal observation that correlates with a GW observation; and that's troubling especially since the number of unimodal observations is increasing, and so the statistics of how often we would expect random signals to correlate (with the three prongs of the detector, and then with a multimodal observation) is getting murkier and harder for third parties to review. The truly disciplined experiment is to do a double-blinded study: over a timespan record events in secret. Then augment the GW data with 3x fictional events (say, fix the times and magnitudes but randomize azimuth/altitude). Challenge EM observers to correlate with these events, break the seal, and see if multimodal observations pass a simple chi-squared test.