Most Theoretical Physicists Are Doing Engineering

Most modern theoretical physics, particularly in areas like string theory and supersymmetry, is actually engineering because practitioners design models based on internal mathematical consistency or aesthetics rather than being constrained by empirical data. It contrasts this with the work of a true scientist, who must derive their formalism from experimental results and pursue ontological questions about reality. The author claims that by treating model design as a description of reality, these physicists have abandoned the core scientific pursuit of understanding nature's fundamental structure.

Qian Xuesen, while under house arrest during the McCarthy era, wrote Engineering Cybernetics in 1954, drawing the line between a scientist and an engineer. The book was later incorporated deeply into Chinese missile and aerospace doctrine when he successfully returned to China to head the Fifth Academy of the National Defense Ministry now CASC , and Pakistan’s reported 6:0 air-engagement result against India in May 2025, run on the KJ-500 + J-10C + PL-15 networked kill chain that Chinese systems engineering inherits directly from Qian, is one downstream receipt. The scientist analyses a system that exists. An engineer designs a system that does not. The two are different jobs with different success criteria. The scientist is constrained by what reality returns when you ask it questions. The engineer is constrained by what they want the system to do. The output of one is a description, and the other, an artifact. Most of what is currently called “theoretical physics” is engineering. The people doing it call themselves scientists but they are not, on Qian Xuesen’s definition, doing the scientist’s job. They are designing models and treating the design as a description of reality. The tell is the moment of “choice of formalism”, when the proposal foregrounds the mathematics’ internal consistency or its aesthetic merit as the reason it should be taken seriously. String theory, ahem. A scientist does not have a choice of formalism. The formalism is whatever it has to be to recover what nature returned. An engineer chooses one because the choice is the work. The scientist’s job, in fundamental physics, is harder than this and not negotiable. You take the system the experimental results, the catalogue, all of it and you ask hard questions about what it means. In fundamental physics those questions are ontological by construction. You cannot conjure a model out of thin air and decide post-hoc what its variables refer to. The variables refer to whatever the experiments forced them to refer to. A proposed equation has to be empirically grounded and rationalised on what reality returned, and its ontology has to stay consistent with itself across the catalogue. If a discipline pretends ontology is “philosophy” and not its problem, it has stopped doing science. We ran the experiments. The data came from nature. Nature has some sort of consistent structure or we would not exist to ask about it. The anti-realist move “the formalism predicts, the rest is metaphysics” is defeatism. As one distillation of Jensen Huang’s recent pushback put it, tongue in cheek, I did not wake up a loser. In my previous essay, The Century-Long Pause in Fundamental Physics, I pointed to how fundamental physicists have stopped pursuing ontology as part of science. Most of the BSM and quantum-gravity programs of the last fifty years are engineered systems being presented as descriptions of reality. Below I list several major programmes that one may have heard in podcasts and YouTube channels like Theory of Everything or Institute of Art and Ideas IAI . Instead of pursuing the hard problem of ontology, they offer three moves to "solve" the hard problems in fundamental physics. We either add free parameters and fit them, extend an existing symmetry and wait for an experiment that never arrives, or pick a different substrate and hope physics falls out. Non-local and “plastic” spacetime gravity. These are reformulations of GR with non-local kernels or deformable spacetime structure. The kernel acquires a non-trivial tree of solutions, and the model is then “made to work” by parameter fitting on the kernel, fitting that is never derived but only chosen so the answer comes out. There is no working interface to QM. The engineering tell is that the parameters doing the load-bearing work do not come from anywhere outside the requirement that the output match observation. SUSY and GUT. These are motivated proposals symmetry-completion of the SM, gauge-coupling unification that had no empirical anchor at the time of proposal. They were dead on arrival in the strict empirical sense, and were kept alive sociologically while the field waited for experiments to formally rule out a sequence of specific predictions. People insisted. We waited. The predictions failed. If these proposals had been presented by non-institutional researchers they would have been dismissed as pure fiction. Superstring and M-theory. It was a cute proposal that was never going to recover from a landscape on the order of 10⁵⁰⁰ vacua. Narrowing that landscape is an unlikely task when the theoretical anchor is neither empirical nor ontologically consistent. There is no rule for picking the vacuum because the construction was never designed to have one. AdS/CFT. Crackpot theories and crankery are expected to explain the Standard Model and the cosmological constant before they are read. AdS/CFT is granted institutional standing as a “disciplined, mathematically rigorous” framework while being constructed in an anti-de Sitter background a negative cosmological constant , which is the opposite sign from the one we measure. The double standard is the diagnostic. Mathematical rigour buys an exemption from the empirical bar that a non-institutional researcher cannot obtain. Loop quantum gravity. LQG is symmetric to the above. Crackpots and cranks are expected to clear the Standard Model bar to be taken seriously. LQG does not attempt to recover the Standard Model and is not asked to. The institutional position skips the question rather than facing it. Twistor theory. Twistor theory is Penrose’s variant of the same operation replace strings with twistors and run the same program , except the construction blows up on the chirality of photons. The googly problem is not a technical loose end. It is the construction failing at exactly the place it has to recover the world. The proposal was an alternative formalism without an empirical anchor, and the formalism does not close. Cell-theoretic universes Wolfram . The proposal is a computational substrate of cells, rules, and rewrites. The mathematical model is internally rich, but the physical connection between the model and any observable in this universe is very thin. You cannot trace from “the system updates according to rule R” to anything an experiment measures here. It might be a fine specification for a game in a different universe. It is one pattern run three ways. The proposals were generated by working inside formalism adding parameters, extending symmetry, picking a substrate , and treated as candidate descriptions of reality without ever having been derived from the catalogue of what reality returned. They are engineered objects. Some of them are mathematically elaborate. None of them are science in Qian Xuesen’s sense, and the category error here is intentionally, or unintentionally, obtuse. A discipline can carry on doing engineering and call it science as long as the institutional reflex protects the practice from challenge. The reflex is visible when challenges come from inside the field with credentials intact. Sabine Hossenfelder @skdh has a PhD in physics, a research record, and a sustained body of public criticism of the methodologies above. The institutional treatment is to relabel her as “a YouTuber,” as if she had stopped being a physicist when she stopped agreeing with the field’s working consensus, especially after she tried "defending" non-institutional researchers like Eric Weinstein even though he makes the same mistake as the institutional researchers . The relabelling is not an evaluation of her arguments but a mechanism for not engaging with them. It is time for the scientist’s job to be done as the scientist’s job, to pursue an ontology-first, empirically grounded description of reality. If the diagnosis lands and you want to see a recent attempt to resume the methodology with worked structural derivations of the Standard Model, see Wave Relativity Tan, 2026, CC BY-NC 4.0 .