I had thought that the wormhole story had reached peak absurdity back in December, but last night some commenters pointed to a new development: the technical calculation used in the publicity stunt was nonsense, not giving what was claimed. The paper explaining this is Comment on “Traversable wormhole dynamics on a quantum processor”, from a group led by Norman Yao. Yao is a leading expert on this kind of thing, recently hired by Harvard as a full professor. There’s no mention in the paper about any conversations he might have had with the main theorist responsible for the publicity stunt, his Harvard colleague Daniel Jafferis.
Tonight Fermilab is still planning a big public event to promote the wormhole, no news yet on whether it’s going to get cancelled. Also, no news from Quanta magazine, which up until now has shown no sign of understanding the extent they were taken in by this. Finally, no news from Nature about whether the paper will be retracted, and whether the retraction will be a cover story with a cool computer graphic of a non-wormhole.
Update: Dan Garisto goes through the Jafferis et al. paper, noting “Turns out it looked good only because they used an average (a fact not specified in the article).” and ending with
The unreported averages for the thermalization and teleportation signal make a stronger case for misconduct on the part of the authors.
I don’t understand why Fermilab was planning a public lecture promoting this, and with what has now come out, it should clearly be cancelled.
Update: I like the suggestion from Andreas Karch
Quanta magazine could make a video where the wormhole authors share in vivid detail the excitement they felt when they realized that their paper isn’t just overhyped but actually wrong.
Update: Garisto has a correction, explaining that the averaging is not the problem with Jafferis et al., rather that the teleportation signal is only there for the pair of operators involved in the machine language training, not there for other pairs of operators that should demonstrate the effect. In any case, best to consult the paper itself. If Jafferis et al. disagree with its conclusions, surely we’ll see an explanation from them soon.
Update: The Harvard Gazette promotes the wormhole publicity stunt, with “Daniel Jafferis’ team has for the first time conducted an experiment based in current quantum computing to understand wormhole dynamics.” As far as I can tell, that’s utter nonsense, with the result of the quantum computer calculation adding zero to our understanding of “wormhole dynamics”.
Update: Video of the Lykken talk now available, advertised by FNAL as Wormholes in the Laboratory.
Just attended Lykken’s webinar. He did not address this refutation at all or any other voices of reason. I attempted to ask a question about this paper but it was not selected (the talk was in Zoom webinar format).
Andrew,
I also watched a bit of the webinar. It was completely misleading propaganda, no mention of anything skeptical or critical, he was selling this as a huge advance and success. Probably you weren’t the only one with serious questions at the end, but they just chose the “when do we get to travel through a wormhole” ones. It’s outrageous that Fermilab is doing this kind of thing.
Hard to know what it will take to stop more of it. Lykken will pursue the standard successful string theory tactic of just ignoring serious criticism, pretending it doesn’t exist.
I see that the grant funding this stuff:
https://pamspublic.science.energy.gov/WebPAMSExternal/Interface/Common/ViewPublicAbstract.aspx?rv=b31f0e55-de01-4635-91d1-2bf112daea20&rtc=24&PRoleId=10
is up end of August. Will anyone with some sense stop this from getting renewed or funded elsewhere?
“their paper isn’t just overhyped but actually wrong”
Phew! I was worried for a moment, there, that the paper was, not even wrong.
I suspect that unless there is public backlash at a level on par with the initial announcement, Jafferis et al. will subscribe to the policy of “we will not validate the criticism with a response”. They’re clearly playing the political PR game. No need to engage if the whole thing never picks up traction in the public arena.
Jim Eadon,
This is starting to become a serious problem with the title of this blog. If things keep going this way and I keep feeling I have to write about them, I may need to change title to something like “Utter Bullshit” to accurately reflect the nature of what I’m writing about.
suomynona,
So far the authors of the stunt are following the string theory playbook for dealing with criticism: pretend it doesn’t exist and keep playing the hype video.
The problem with calling this wormhole bullshit “bullshit” is that, in the unlikely event that mainstream media wants to report it, they can’t. Because the FCC, y’know? So it’s kinda self-defeating.
How’s that appointment of a medieval historian to head the Institute For Advanced Studies working out, in your opinion? The rub here is that there are structural reasons why these positions are occupied by the kinds of people occupying them; a scientist would have, if not knowingly lied about the wormhole, then played along, while, charitably, the medievalist simply did not know better and was duped. So which of the two given choices (the hypester-hoaxster-fraudster or the innocent) is, in your opinion, uh, preferable as director of IAS?
Lastly: at what point do we start to describe these people as what they are: swindlers, hoaxers, charlatans, liars, or, if you want to, simply “dishonest, money-seeking, self-serving careerists”. While some of these labels might be actionable, proper terminology is also important!
More_Nonymous,
It’s an interesting question about alternative universes to speculate how things would have been different if Dijkgraaf had still been IAS director. I doubt he would have have brought history into it, wouldn’t have made public comparisons to the experimental vindication of GR in 1919. I also doubt he would have said anything skeptical, preferring to join his colleagues in trying to figure out how to best exploit the hype to help with funding of their field.
The paper written by Norman Yao et al sounds compelling.
I am waiting for Jafferis et al.’s response to this paper.
I wrote two comments on the wormhole experiment. But I uploaded these comments in December and January. Anyway, I put them here:
https://arxiv.org/abs/2212.03568
https://arxiv.org/abs/2301.03522
The second comment discusses philosophical problems. I thought back then that the great achievement was to generate the sparse Hamiltonian, i.e., the greatest achievement was done by conventional computers rather than by the Sycamore. But if as Yao et al show that this sparse Hamiltonian is problematic, then I may have to update my second comment. I will wait and see…
Yao et al. claim that the systems considered in the Nature paper are unable to simultaneously satisfy scrambling + thermalization dynamics and perfect-size winding. Nevertheless, both properties are central to the holographic correspondence (AdS/CFT) and are known to occur in the SYK model at large system sizes.
Perfect-size winding is a necessary criterion of a traversable (ER = EPR) holographic wormhole. So did Jafferis et al. simulate perfect-size winding with 9 qubits?
Yao et al. argue that perfect-size winding observed by Jafferis et al. seems reliant on the small size, but not on thermalization and scrambling (quantum chaos). I wonder whether this contradiction is correctable or rather, maybe this signifies that we cannot simulate quantum teleportation through a holographic wormhole with 9 qubits (with 164 two-qubit gates circuit).
Gil Kalai wrote that we have difficulties even with 5 qubits: https://cstheory.stackexchange.com/questions/35934/states-and-probability-distributions-that-the-5-qubits-ibm-computer-can-produce?fbclid=IwAR1ntzNZ9df1wZtaIYxBl7uhXoTfRhtPkBo_1e1w4uUgEbIyLHuwUIXfiVM
@Gali Weinstein:
“Gil Kalai wrote that we have difficulties even with 5 qubits: […]”
I don’t know too much about quantum computing, but that question you link has last been updated 5 years ago, which, I believe, in the field of quantum computing on real devices is ancient history.