August 4, 2026
Quantum hype meets comment-section chaos
Nobody checked which state IBM's flagship quantum chemistry results compute
IBM’s big quantum win is getting roasted as commenters ask: did it solve the wrong thing?
TLDR: A new audit says IBM’s headline quantum chemistry result may have converged on the wrong answer, undercutting a major bragging point. In the comments, readers piled on with AI-writing accusations, credential checks, and a very public trust test over whether the work is solid or sloppy.
A fresh audit of IBM’s splashy quantum chemistry results has landed like a reality-show reunion episode: messy, dramatic, and full of side-eye. The paper’s core claim is simple enough for non-specialists: IBM’s famous calculation may have reached an answer, just not the right kind of answer it said it was chasing. In plain English, critics are saying the benchmark may have looked impressive while actually locking onto the wrong state of the molecule. Ouch.
But honestly? The real fireworks were in the comments. One of the loudest reactions didn’t even start with the science — it started with the writing style. User gijsleb instantly declared “Claude slop detected,” then came armed with a prosecutorial list of suspiciously AI-sounding phrases like a detective pinning red string to a corkboard. Another commenter, gus_massa, turned the thread into a pop quiz, basically saying: before I trust this, tell me your degree and solve a baby matrix problem without a computer. That set the tone fast: less polite seminar, more “papers, please” for quantum authors.
The hottest opinion in the room is that if these benchmark results really missed something this basic, then the hype around “utility-scale” quantum chemistry takes a serious hit. The funniest reactions came from people treating the whole thing like a prestige-tech version of failing to check whether your GPS sent you to the wrong city. Beneath the jokes, though, the mood is serious: commenters want proof, receipts, and less glossy hype next time.
Key Points
- •The article presents a reproducible data-and-code archive for a preprint auditing sample-based quantum diagonalization benchmarks on iron–sulfur clusters.
- •It argues that convergence in projected energy did not demonstrate convergence to the intended target spin state in the audited executions.
- •Using multiple spin diagnostics, the analysis reports that none of the audited runs returned the named singlet at a competitive energy.
- •At the largest published dimension, 5.625×10⁷ determinants, the reported lowest stable root is about 170 mHa below the published energy and has ⟨S²⟩ = 1.37.
- •The archive includes scripts, result files, reproduction instructions, and SHA-256 hashes, and the article recommends making measured spin moments a required benchmark output.