A new preprint on arXiv (2603.28869v1) proposes what it calls a 'totally serious' algorithm to solve NP-complete problems in polynomial time. The catch: it only works if you accept the many-worlds interpretation (MWI) of quantum mechanics and are willing to risk the destruction of all observers in branches where the algorithm fails. This is not an empirical study but a purely theoretical thought experiment with no methodology, no sample size, and no experimental validation. As a preprint, it has not undergone peer review.

The paper revives the quantum suicide concept, first seriously explored by Max Tegmark in his 1998 work on quantum immortality. In this setup, an agent runs a nondeterministic search for a solution (such as a satisfying assignment for 3-SAT). A quantum device is rigged to trigger universal destruction in any branch where verification fails. Under MWI, the observer only experiences branches where a correct solution is found, effectively selecting the right answer through survival bias. The authors argue this constitutes a polynomial-time solution conditional on MWI being true.

What the paper misses, and what much of the coverage ignores, is that this does not constitute a conventional proof that P=NP. P versus NP is a question within classical computational complexity; this approach relies on anthropic selection across unobservable parallel worlds rather than a deterministic algorithm runnable in a single universe. It also fails to address the measure problem in MWI - the exponentially small amplitude of surviving branches could render the procedure useless in practice.

Synthesizing this with related work, David Deutsch's foundational papers on quantum computation (1985) show quantum parallelism can offer speedups, but never for NP-complete problems in the worst case without collapsing to the P=NP question again. Tegmark's earlier writings on quantum suicide highlight its philosophical nature as a test of MWI, not an engineering tool. Conventional complexity theorists, including results from the 1970s by Cook and Karp on NP-completeness, emphasize that any claimed P=NP proof must work within standard Turing machine models without invoking untestable multiverse assumptions.

The preprint taps into a pattern of bold, speculative bridges between quantum foundations and computer science that often generate headlines but rarely survive scrutiny. It resembles past hype cycles around quantum annealing devices claiming to 'solve' optimization problems. Limitations are severe: the method is untestable, requires destroying vast numbers of branches, and only 'works' if MWI is both correct and allows observer continuity across cosmic-scale destruction. Genuine analysis suggests this is more philosophical provocation than computational breakthrough.