On telling Andrea Rossi from Eric Drexler
Beginning around 2011, the Italian inventor Andrea Rossi claimed to have built a tabletop cold-fusion reactor — the “E-Cat” — that produced far more energy than it consumed, in apparent violation of accepted physics. Most physicists dismissed it as fraud; a smaller group, including the philosopher Huw Price, thought the claim deserved serious scrutiny. The exchange below picks up after Price has argued that the physics violation Rossi’s device would imply isn’t qualitatively larger than other recent surprises in physics — e.g. the (later-retracted) faster-than-light neutrino measurement at OPERA in 2011 — and that, since no scientific theory is ever exactly correct, residual uncertainty leaves room for Rossi to be right. The reply lays out a general framework for evaluating extraordinary scientific claims, contrasting Rossi with the nanotechnology pioneer Eric Drexler — also unorthodox, also widely doubted, but, Alyssa argues, a categorically different kind of figure.
On the “newness” question: I’d actually agree the degree of newness required for Rossi is much less than would be needed for superluminal neutrinos. If I understand GR correctly, superluminal travel is logically equivalent to having closed timelike curves, which is logically equivalent to having regions of “negative mass”, which is logically equivalent to breaking the second law of thermodynamics, and at that point all Hell breaks loose. However, that seems sort of like comparing the credit score of Bill Gates with the credit score of Michael Bloomberg; it’s sort of a moot point in any case.
On the broader argument that no scientific theory is ever fully correct: technically true, but the implication — that we can’t be super confident about any one aspect of science — is false. E.g., if one took a textbook on evolution, assigned a reasonable probability to each independent statement, and multiplied all the probabilities together, one would get an extremely low number. Thus, in some sense, it’s extremely likely that the textbook is “wrong”. However, that doesn’t mean it’s accurate for a creationist to say that we aren’t sure whether humans descended from earlier primates — that tiny number includes a bunch of less-confident statements, along with the really-confident ones. The odds that humans descended from primates are more like 99.99999999% than they are like 90%. Any uncertainty about, e.g., the contribution that epigenetics makes to inherited characteristics doesn’t “bleed over” and make us uncertain about whether humans and chimps share DNA.
It’s also doubly unfair for a creationist (or Rossi) to say that any residual uncertainty about the basic laws of science strengthens their theories by comparison. The total probability for all theories must sum to 1. Therefore, if we assign (e.g.) a 99.9999% confidence to evolution, the remaining 0.0001% has to be distributed over every other possibility — “humans were created by the Christian God” is just one idea, along with countless other creation myths ( List of creation myths ) and many more non-myth hypotheses, like Sufficiently Advanced Aliens. There are a number of possible “blank spots” in physics that aren’t fully understood yet, just as there are “blank spots” in evolution, but that doesn’t provide real evidence for Rossi for the same reason that it isn’t real evidence for creationism. Proving Rossi correct, or even proving that we should assign his theories odds higher than one in a million, requires specific evidence in favor of his theory, and not in favor of every other alternative that also breaks standard physical laws. (Of course, Rossi has tried to provide this, but there’s very convincing evidence that all of the “tests” were fraudulent.)
There’s a very, very long list of other people who tried to make “unlimited free energy” work — see e.g.
the history of perpetual motion machines , BlackLight Power , Newman’s energy machine , the Patterson Power Cell , free-energy suppression conspiracy theories , Steorn , Methernitha , the Simple Magnetic Overunity Toy , Perepiteia , and the MythBusters free-energy episode for just a few of the most prominent examples. Even if one agreed, in defiance of known physics, that at least one of these devices might work, that wouldn’t say which one it was — one would still have to somehow pick the right device out of all of the hundreds of claims. This is actually my position regarding e.g. cancer research — I think it’s likely that someone somewhere has discovered a cure for cancer, it’s just extraordinarily difficult to find them among the hundreds and hundreds of fraudulent cures for cancer ( Wikipedia alone lists several hundred ). The probability one should assign to any particular person claiming to have a cancer cure is still extremely low, unless they can provide tests which are more conclusive than the tests provided by all of the other such people. Most of the hundreds of cancer quacks each claim to be “scientifically proven”, to be supported by reputable scientists and institutions, to have done studies, to have lots of examples of people who were cured of terminal cases, etc., etc.
Of course, there are still people — I’ll name Eric Drexler, but there are lots of others — who I think are worth listening to despite lots of scientific skepticism. How could one, therefore, tell Drexler apart from people who are almost certainly con artists? One could become an expert in (e.g.) nuclear physics or nanotechnology, and evaluate their claims directly, but that would be really time-consuming. One could also do surveys of scientists, and show that (making up plausible numbers) 75% are skeptical of Drexler while 99% are skeptical of Rossi, but doing surveys is pretty annoying and expensive too. So below, I describe something like the method I use for this myself. Of course, when possible, it’s always better to just evaluate the experimental evidence directly; but it just becomes impractical to do that for each one of the thousands of people making implausible claims.
First, Drexler’s ideas require no new fundamental physics; they are purely about engineering. Of course, new discoveries in fundamental physics do happen sometimes. But the skills needed to engineer commercially practical devices, and the skills needed to make fundamental physics breakthroughs, are both extremely rare, and they don’t correlate very well (hence, the many many discoveries by academics that never managed to hit the market). By itself, new physics is rare, but happens sometimes; new engineering of devices that’s good enough to make them commercially ubiquitous is rare, but happens sometimes; but the conjunction of both happening in the same place at the same time is so rare as to make it extremely improbable.
Second, Drexler is an expert in his field, and before starting his campaign for nanotechnology in the 90s, published a number of papers in top journals and got triple degrees from MIT. Of course, sometimes amateurs do good work too. But again, in Rossi’s case, this requires a conjunction of two unlikely things. For an amateur with no training to do good physics is unlikely, but not impossible; for someone to make the most important physics discovery of the last 50 years is unlikely, but not impossible; but for both of them to happen at the same time is extremely implausible.
Third, Drexler never claimed anything like having a practical device; he only claimed that such devices could eventually be built, likely many decades or even centuries from now. That’s always going to be a much lower bar than claiming to have a commercially practical machine available for sale. There’s a huge huge huge gap between basic lab physics, and production-scale hardware, even for well-understood fields; e.g. the chemistry of redox reactions is well understood, but most of the dozens of startup battery companies are likely to fail, because battery economics is more about efficient production lines than it is about chemistry (indefinite examples available on request).
Fourthly, it’s very clear from Drexler’s writing that he understands the basics of his field; his book Nanosystems is filled with well-understood chemistry equations, and tons of citations to existing literature. One can be pretty sure that Drexler isn’t making elementary mistakes because he doesn’t know Thermodynamics 101. Rossi’s papers simply never mention any of the well-known concepts in fusion, such as the Coulomb barrier, the Lawson criterion, braking radiation, quantum tunnelling, the Gamow factor, etc., etc. In effect, Rossi tries to have his cake and eat it too. On the one hand, he wants to portray his device as an as-yet-unexplained experimental result — “we don’t know how it works, we just know that it does” — so he can avoid answering questions about how it contradicts existing physics. On the other hand, he wants to portray himself as a genius inventor, and guarantees that the device will be cheap, clean, safe, easily mass-producable, reliable over decades, and so on. Needless to say, this is not how one handles as-yet-mysterious experimental results in nuclear physics.
Fifthly, to cover all of the stuff already discussed, Rossi was previously convicted of fraud for free energy scams, cheated the US Army when he tried to sell them thermal devices, said he had already delivered a 1 MW functioning power plant (!) back in 2011 that then failed to show up for unexplained reasons, got a fake chemical engineering degree from a diploma mill, etc. ( full timeline ).
Again, all of this is just rules of thumb, and would be irrelevant if Rossi had produced solid experimental results, or a functioning commercially-available device. But he has not. He’s been predicting both of them imminently ever since at least 2007, but they have never shown up. That’s just not how real product development works. A friend of mine makes an extremely simple piece of hardware — a natural gas tank — and it took them four years to go from prototype to commercial production. This included a months-long intensive certification process for safety reasons, where the tanks were baked, frozen, drilled, dropped from a height, worn through decades of simulated use, burned with acids, burned with bases, attacked with solvents, run over with a Mack truck (yes really), shot with a .50 caliber armor-piercing rifle (yes really)… I don’t even want to think about the approval process for selling a totally new type of nuclear reactor (!). How is this guy going to get insurance for manufacturing unlicensed nuclear reactors? How can he not show his devices to anyone, when thousands of low-paid workers are manufacturing them on some kind of big assembly line? Even if Rossi’s device doesn’t currently produce neutrons, if there’s any possible modification that would allow it to produce lots of neutrons, that would let anyone who bought one create unlimited amounts of cheap nuclear-weapons-grade plutonium. Is Rossi just going to tell the Nuclear Regulatory Commission and the Department of Homeland Security that, hey, just trust me, I’m sure no one could ever use this to make nukes? Realistically, even if all of the physics problems went away, you’d be looking at ten years minimum until commercial scale, and yet Rossi’s company is already taking preorders.