For general readers:
A week or so ago, I wrote about my skepticism concerning the claim of a “detection” of the photon ring that’s widely expected to lie hidden within the image of a black hole. A nice article in Science News appeared today outlining the current controversy, with some quotes from scientists with differing views (including a dissenter from the EHT itself). It’s a good read: to the point, well-written, and with high standards.
Background: The image in Figure 1, created by the Event Horizon Telescope (EHT) in 2019, shows the “accretion disk” of material circling and eventually falling into the giant black hole at the center of the galaxy M87. But the image is too blurry to easily reveal the photon ring, whose appearance in a perfect telescope might look like that in Figure 2, which shows a simulation of the region around a black hole.
Unlike the accretion disk, the photon ring would be an effect purely of curved space-time around the black hole, which is expected to lens and focus light from the accretion disk into a narrow circlet. As such, the ring would be a smoking-gun signature of Einstein’s gravity at work. That’s why any claim of detection, using fancy image processing or any other method, merits both attention and critique.
3 Responses
Also a smoking gun, not being direct evidence, merits both attention and critique.
We computer types have a long and well-established track record of subconsciously imprinting our hopes and expectations into software that we claim is interpretation indifferent. The devil is in the details, but healthy caution is extremely advisable for any type of AI-synthesized “final” result.
I see that claim often, but I don’t see a long and well-established track record of anything like it.
Yes, ascertainment bias has been identified by the Cochrane institute but “We have not identified studies reporting the empirical impact of ascertainment bias.” And ML can be tricky but its “expectations” aren’t human.
I would expect that claims of verified bias are heavily biased. 😎