The paper explaining that this Nature cover story, besides being a publicity stunt, was also completely wrong, has so far attracted very little media attention. The first thing I’ve seen came out today at New Scientist, a publication often accused of promoting hype, but in this case so far the only one reporting problems with the hyped result. The title of the article is Google’s quantum computer simulation of a wormhole may not have worked. It contains an explanation of the technical problems:
The first problem has to do with how the simulated wormhole reacted to the signals being sent through it….Yao and his colleagues found that for each individual test, the system continued to oscillate indefinitely, which doesn’t match the expected behaviour of a wormhole.
The second issue was related to the signals themselves. One of the signatures of a real wormhole – and therefore of a good holographic representation of a wormhole – is that the signal comes out looking the same as it went in. Yao and his team found that while this worked for some signals – those similar to the ones the researchers used to train a machine learning algorithm used to simplify the system – it didn’t work for others.
…it seems that for this particular quantum system, the size winding would disappear if the model was made larger or more detailed. Therefore, the perfect size winding observed by the original authors may just be a relic of the model’s small size and simplicity.
There is a response from Maria Spiropulu:
“The authors of the comment argue about the many-body properties of the individual decoupled quantum systems of our model,” she says. “We observed features of the coupled systems consistent with traversable wormhole teleportation.”
Remarkably, Lenny Susskind throws the authors of the stunt under the bus:
“What is not so clear is whether the experiment is any better than garden-variety quantum teleportation and does it really capture the features of macroscopic general relativity that the authors might like to claim… only in the most fuzzy of ways (at best),” he says.
I remember being at a student journal club discussing this work in which a curious computer science undergrad with little to no knowledge of wormholes etc. pointed out that the result would have to be reproduced with “untrained operators” to properly establish the claim. I’m surprised that nobody pointed out this basic and blatant flaw earlier.
I wouldn’t have thought it possible, but I agree with Susskind.
Can anyone explain Spiropulu’s response? The only way I can think of making sense of it is if she mixed up “decoupled” and “commuting.”
I’m not sure why it’s shocking to see Lenny express scepticism about a result that has met with almost universal scepticism from experts in the field who were not co-authors of the paper.
@E.S. I believe the authors argument is that even though the terms in each “boundary Hamiltonian” all commute, the coupling between the two sides doesn’t commute with them, and so the coupled system is interesting, or even “wormhole-like”, even if the individual decoupled systems are not. I find this argument uncompelling. An actual holographic simulation of a wormhole should absolutely have to thermalise without any coupling between the two sides.
This is off topic but Springer is expecting to release a Handbook of Quantum Gravity this year or next. The ncatlab has a page with published chapters on arXiv so far which may be of interest to readers:
Much more CDT and LQG than I would have guessed.
Susskind was in some sense part of the original publicity stunt, co-authoring the explanatory article that appeared in Nature with the technical article, see
“A holographic wormhole traversed in a quantum computer”
I’m assuming he recently read the Yao et al article and realized that the original article he had helped promote was likely technically wrong. Not shocking that he should now express skepticism, for shocking, see what Harvard is sending out, just got this yesterday in the newsletter they send out to their alumni
Last I heard from the editor at Quanta and the director of the IAS, they were standing behind this work as a huge breakthrough, since that was what they were hearing from experts in the field. Maybe things have changed, will be interesting to see if Quanta now takes seriously my suggestion that they should re-report this story.
Perhaps the interesting question is why the journalists at Quanta would take it upon themselves to decide that something *was* a huge breakthrough when such a large number of experts in the field held that it was not only not a breakthrough, but was nonsense. It’s a vision of science journalism that I hope they do not decide to adopt uniformly.
More from Lenny (“Lost in the landscape”):
The Quanta article (and associated video) was produced under embargo conditions, where the claims of a “breakthrough” were being kept secret and could only be discussed with a very limited range of people other than the authors. Even after the “breakthrough” was made public, the initial reaction from a lot of experts was to try and find a way to be supportive, since it was excellent PR for their field. It took a while for many experts to look into the actual result and realize there was nothing there. More recently, the article by Yao et al has made clear that what was there was not just vacuous but wrong.
I’m curious to see what Quanta does about this. Last time I was in contact with them, I was advising them they should re-report the story. Their response was that they were going to listen to what experts had to say (not me…).
It seems to be spreading. Not sure what to make of this exactly, but it sure seems like wormhole-style hype to me.
Yes, outrageous hype, courtesy of the University of Bristol press office. Anyone at Bristol willing to contact people there to tell them to not do this kind of thing?
Anyone who wants to skeptically study the claim from Bristol about “counterfactual teleportation” might want to start with this critique by Lev Vaidman of the original concept: