Decades of neuro research wasted because "we wanted to be REAL SCIENTISTS." @BrentWRoberts at 15:00 on @fourbeerspod
NIMH funded only candidate-gene and fMRI studies, because SCIENCE, and both of those turned out to be science-free mirages.https://fourbeers.fireside.fm/22
-
Show this thread
-
@BrentWRoberts dutifully pursued the 5-HTTLPR mutation, which was supposed to cause depression. By 2010 he realized it doesn't and abandoned the project. Yet, after seemingly conclusive evidence, funding inertia means we're still wasting time & $$ on it.https://slatestarcodex.com/2019/05/07/5-httlpr-a-pointed-review/ …5 replies 3 retweets 24 likesShow this thread -
Replying to @Meaningness @BrentWRoberts
I don't want to get too Pollyanna-ish and claim the system self-corrected gracefully, but it does self-correct. Eventually self-referential piles of bs come tumbling down which is better than most human systems of codifying information.
1 reply 0 retweets 0 likes -
Similarly, there was obviously more to be wrung out of backprop, but it had to spend 15 years in the wilderness for hw to becomes so overpoweringly good that examples would overcome the stigma from the 2nd wave neural net prophets of the late-80's.
1 reply 0 retweets 0 likes -
We had a 16K CM-200 back in the day and, for our use case (ocean and atmospheric modeling), it was a pretty nifty machine, but that was mostly due to the Weitek FPUs. The base CM-2 was more of an expensive curiosity.
1 reply 0 retweets 0 likes
Yup. The CM-200 was a GPU before GPUs were invented. Or, less charitably, a Cray with an alien parasite attached. Or, the spiritual successor to the PDP-12… a floorwax AND a dessert topping. I left long before its design, and can’t take any of the credit/blame!
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.