Scaling up Connectomics
Posted: Fri Oct 06, 2023 2:16 pm
https://wellcome.org/reports/scaling-connectomics
https://cms.wellcome.org/sites/default/ ... tomics.pdf
Scanning the connectome would be a critical step in revival. Good progress is being made, and I just realized that there was a major flaw in my predictions. I was assuming that the electron microscope that would eventually be capable of scanning a human brain would be scanning large diameter slices. I now realize that instead of a million slices, we might instead be talking about (tens/hundreds of?) thousands of tiny cubes, each of which then get sliced thin for scanning using conventional technology. This means that the massive expensive electron microscopes that I had been assuming were necessary just aren't. All we're really waiting for is for the AI analysis to improve. This is a very important realization because software can improve very quickly. Since it now seems to just be a software problem, I'm moving up my predicted date a few decades. So I now think we're looking at more like 90 years to revival instead of 110 years. I'm wrong either way, but it's fun to try to increase prediction accuracy.
https://cms.wellcome.org/sites/default/ ... tomics.pdf
Scanning the connectome would be a critical step in revival. Good progress is being made, and I just realized that there was a major flaw in my predictions. I was assuming that the electron microscope that would eventually be capable of scanning a human brain would be scanning large diameter slices. I now realize that instead of a million slices, we might instead be talking about (tens/hundreds of?) thousands of tiny cubes, each of which then get sliced thin for scanning using conventional technology. This means that the massive expensive electron microscopes that I had been assuming were necessary just aren't. All we're really waiting for is for the AI analysis to improve. This is a very important realization because software can improve very quickly. Since it now seems to just be a software problem, I'm moving up my predicted date a few decades. So I now think we're looking at more like 90 years to revival instead of 110 years. I'm wrong either way, but it's fun to try to increase prediction accuracy.