One morning I woke up to see the front page of every newspaper across the world covered with pictures of what looked like a big orange doughnut. What could this represent? I learned that the doughnut picture was rendered by combining images taken from telescopes all around the world.
Petabytes of data had been mailed to one location for analysis and a few groups of PhD students wrote some code to combine and overlay the images. The student responsible for combining all of the work into one image got something that looked like a ‘black hole’ and newspapers and magazines from around the world put it on the front page, even though each individual image in the data set showed no donut, just blur.
I thought: PhD student work is great, but it often contains mistakes, so I sure hope her advisor went through her code very carefully. If you overlay petabytes of images with individually tuned contrast ratios and construct your algorithm so that the overlay is centered around a region of interest, it is certainly possible to create a black hole in the middle of your region of interest. You’d think that sort of scrutiny is routinely applied to new results, but in the darkened conference rooms in which such data is presented, skepticism is expressed in hushed and veiled terms. Sometimes, nothing stands in the way of an idea marching ahead.
The first article I read had a bit of detail on how the algorithm was developed:
Bouman adopted a clever algebraic solution to this problem: If the measurements from three telescopes are multiplied, the extra delays caused by atmospheric noise cancel each other out. This does mean that each new measurement requires data from three telescopes, not just two, but the increase in precision makes up for the loss of information.
I wanted to see her presentation of her data, but what I found was a TED talk. The thing that concerned me the most from her TED talk was when she said at 6:40 (I paraphrased):
Some images are less likely than others and it is my job to design an algorithm that gives more weight to the images which are more likely.
as in, she told her algorithm to look for pictures that look like black holes and, lo and behold, her algorithm found black holes by ignoring the data that didn’t look like black holes.
LIGO did something similar in their algorithms, so if they got away with it, why can’t she?
Finally, I found a real, academic presentation from shortly after the media blitz and front page news stories. It is an hour long and the technical stuff starts about ten minutes in. At 14:40 she talks about the ‘challenge’ of dealing with data that had an ‘error’ of 100%. At 16:00 she talks about how the ‘CLEAN’ algorithm is guided a lot by the user – as in, the user makes the image look how they think it should look. At 19:30, she said, “Most people use this method to do calibration before imaging, but we set it up so that we could do calibration during imaging.” Gaaaah! At 31:40, she shows four images that look the same in the amplitude domain – showing the extent to which this measurement relies on information in the phase domain. An image with a hole or without a hole looks the same in the amplitude domain. At 39:30, she says that the phase data is unusable and the amplitude data is noisy. To me, this sounds like she just contradicted herself.
A clever commenter who read this material when I first posted it on quora.com wrote:
Take a look at this picture of my cat.
You don’t see a cat? You just need to apply the right cat-shaped filters.
When I see a picture of a black hole, I see that our academic system has succumbed to the overwhelming noise of our dark age.
I should mention pre-existing biases; my default setting is skeptical. I don’t really believe that black holes exist because I think that theorists got drunk on general relativity and invented them. Astronomers got drunk on interpreting the meanings of tiny dots of light and convinced themselves that they had seen these invisible, theoretical beasts of the night sky.
Seiously, any time you use a singularity to describe a physical phenomenon, it usually turns out to be wrong. Just think about the equations governing water swirling around in a basin. If you use one type of equation, you get a singularity in your basin and if you use another type of equation, you don’t.
That the whole universe is full of singularities floating around in space requires a suspension of disbelief with which I am not really comfortable. I am not alone in this, apparently. Believe it or not, there are still scientists out there who do not believe that black holes exist.
Some black hole researchers will display a picture of a star with a dark spot in it and then follow that up with a simulation showing the same picture. There is something important to know about simulations. If you have a picture of something, it is easy to make a simulation that copies the picture. What is hard is to make a simulation of something you have never seen before, predict how and where you will see it, and then record an observation of it. That is really the only way to do science. Any other route can lead to self-trickery.
Here is a precursor to the black hole image that made Katie Bouman’s PhD work famous.
Saturation effects must be rather difficult to deal with in such images, but, as we saw recently, that hasn’t stopped them from ending up on the front page of newspapers across the world.
Here is a study which claims to support the existence of black holes, but really just tracked some stars near the center of our galaxy. They write
“The stars in the innermost region are in random orbits, like a swarm of bees,” says Gillessen. “However, further out, six of the 28 stars orbit the black hole in a disc. In this respect the new study has also confirmed explicitly earlier work in which the disc had been found, but only in a statistical sense. Ordered motion outside the central light-month, randomly oriented orbits inside – that’s how the dynamics of the young stars in the Galactic Centre are best described.” Unprecedented 16-Year Long Study Tracks Stars Orbiting Milky Way Black Hole
I believe their measurements, but I don’t always believe the interpretation which scientists give to their results.
A lot of money goes into making these sorts of simulations and studying black holes, so one should expect resistance to any change in belief.
Although the author of this video does not come out and question the existence of black holes, I like how he describes the way in which astronomers pick unlikely scenarios out of thin air and use them to explain the qualities of blurry blobs of light. I find it amazing how ‘artist’s renditions’ and just-so stories pass as science in this day and age.
Science should produce progress, but when it swirls around in an eddy of self-citation, you end up with a black hole – in a figurative, not literal sense.
(I first posted this material on quora.com and the photo in the header is from Viccissitudes by Jason deCaires Taylor. underwatersculptures.com)