News You Can Bruise

In the scant months I've subscribed to it, the arXiv blog has served up many interesting and possibly insane bits of scientific speculation. The longest paper it's caused me to read is Michael Dittmar's four-part "The Future of Nuclear Energy", the most recent part of which was published recently.

On the surface the paper is an overview of the state of nuclear power, but an interesting argument quickly forms that the world doesn't have much exploitable uranium left. Here's my summary of the main argument, with extraneous and historical information removed:

1. Every estimate of how much uranium is available uses data from the IAEA "Red Book".
2. The Red Book is not a scientific document. The data is self-reported by uranium-mining countries, and is given to an unrealistic degree of precision with no margin of error. There's no way to check it.
3. Trends in the Red Book year-over-year indicate that certain countries are fudging their data, for instance by pumping up the huge "uranium reserves that probably exist" number. This number being the main underpinning of the claims that there's enough uranium to supply our cravings for nuclear-derived electricity for however many years.
4. Even if the Red Book is accurate, "however many years" is not as long as it sounds, since nuclear power plants are built to amortize their cost over 40-60 years. And right now nuclear power only generates 15% of the world's electricity. If you want to double that, "however many years" gets cut in half, and you still haven't replaced fossil fuels.
5. What about high tech like breeder reactors? According to Dittmar, there are only a couple commercial breeder reactors in operation and they don't really produce a net increase in radioactive material. It's more like regenerative braking in an electric car, where you get some of the energy back.

   The technical problems with breeder reactors can theoretically be overcome with more research. (Dittmar has no hope for commercial fusion.) But, given the long development lead times and the fact that the necessary research isn't being funded right now, by the time they're commercially viable there won't be any tritium to run them on, because of the uranium problem. (Tritium can be generated in a conventional uranium-burning nuclear reactor, but very slowly--and it has a half-life of 12 years.)

I was intrigued by this analysis, because I always think a claim that some data set is wrong or has been misinterpreted is interesting. But because it's such a seductive narrative I also wanted to look at a high-quality response to Dittmar's claims. Any problems with the logic aside, there are a couple problems with Dittmar's presentation that give bad smells:

1. Dittmar has an obvious bias against nuclear weapons, a bias that seems quite sensible to a civilian but which you should probably not write into your papers if you're trying to convince other nuclear scientists of something. Seriously, don't tell your readers to watch Dr. Strangelove. Anyone interested enough to read your paper has already seen it.
2. Dittmar also believes that certain recovery techniques like reprocessing fuel rods for plutonium will inevitably drive proliferation of nuclear weapons. I've heard this argument before but I don't understand the physics of it, so I can't judge. Anyway, this really has nothing to do with his main argument, except insofar as he's attacking a pie-in-the-sky idea where the whole world uses nuclear energy for exclusively peaceful purposes.
3. I don't think English is Dittmar's first language, because his grammar is not great. This is nothing that couldn't be fixed by an editor, but it makes his writing look crackpot-ish.

So I was on the market, and happy this morning when I saw that Paul Raven posted about Dittmar's paper on Futurismic today, and linked to this 2008 post on the Wall Street Journal environmental weblog. That weblog post dismisses contemporaneous "peak uranium" concerns by... citing the IAEA Red Book data. But the main point of Dittmar's paper is that the Red Book data is unreliable!

In desperation, I turned to the comment section on the original arXiv blog posting, hoping that there would at least be people arguing over the paper instead of just quoting the Red Book. Well, there was a lot of irrelevant arguing (fair enough, since there's a lot of irrelevant points in Dittmar's paper), but below is a rough classification of the relevant rebuttals:

• "Dittmar is an obnoxious jerk." This appears to be true, and it might also be relevant. "The data set you've been implicitly trusting for years doesn't look right" is the kind of argument that could give you a reputation as an obnoxious jerk, but it's also the kind of thing an obnoxious jerk would say just to make trouble.
• "Critiquing the Red Book data doesn't prove anything." Given that everyone who argues against peak uranium references a document that has its ultimate source in the Red Book data, it seems legitimate to critique that data. BUT, a critique only establishes that we don't know something. You can't build a positive argument on this.
• "There's lots of uranium we just haven't looked for yet." But any attempt at quantifying this means relying on Red Book data, which Dittmars claims makes things look too good. Also, getting the uranium into a reactor is a complicated process with a long lead time, which starts with looking for the damn uranium.
• "We won't run out of uranium, it'll just get more expensive." Sure, but any attempt at quantifying this means bringing in Red Book data. Part III of Dittmar's paper claims that the Red Book gives an incoherent estimate of the total cost of the world's uranium reserves. BUT, that just means the Red Book is wrong about the price classification. It doesn't make the incoherently-priced uranium nonexistent or infinitely expensive.
• "We don't have to get fissile material from the ground." Indeed, there are a lot of interesting things we could do to get more fissile material: decommissioning nuclear weapons, reprocessing uranium or fuel rods. But these things are accounted for in the Red Book and covered in Dittmar's paper. We're already doing all of those things to some extent, and the only one that gives significant material is decommissioned nukes (currently providing 25% of world uranium). We can ramp up these processes, but there's a long lead time.

OK, so after colliding thesis and antithesis in my Hegelian particle accelerator, I've come to the following tenative conclusions:

• The Red Book data is unreliable. It's based on self-reported data by interested parties. There's fudging and uranium reserves are priced in an incoherent way. For political reasons, the Red Book itself can't mention this fact. The same problem probably exists with, eg. reported estimates of coal reserves, but it's worse with uranium because 1) there's more geopolitical incentive to fudge, 2) the numbers are smaller because uranium is much scarcer than coal.
• This doesn't change the basic rules of natural resources. If uranium gets scarcer the price will go up and people will try to extract marginally recoverable uranium. This would be true even if the Red Book with its incoherent pricing didn't exist. We just don't have an accurate estimate of the prices.
• If the Red Book data is bad, you can't use it. As Daniel Davies says, "If you have doubts about the integrity of a forecaster, you can't use their forecasts at all. Not even as a 'starting point'." By showing that the Red Book data is bad, you can drive demand for a better set of numbers. If you can prove that the numbers are systematically manipulated, you might be able to cancel the systematic error out, though you'll never know for sure. But you can't demonstrate that the numbers don't look right and then make recommendations based on what you think they really are.
• Breeder reactors are not as cool as I'd thought, and progress on them slower than I thought. This is where the "obnoxious jerk" factor comes in. It sounds like Dittmar has had the following conversation with other physicists:

  "Even if you solve problem K, how are you going to solve X, Y, and Z to make your reactor capable of 90% uptime in a commercial setting?"

  "Why do I have to solve X, Y, and Z? It's hard enough getting funding for K. Let someone else do it. You jerk."

  The problem according to Dittmar is that these problems will get solved serially over the course of 30 years and then we'll spend 15 years building a power plant and by that time the tritium will have decayed. The timing here depends on the question of how much uranium there really is, but the general argument is that with more funding and more physicists we could solve X, Y, and Z in parallel with K.

• The big problem with any fissile-material-related "we'll be fine because of X" argument is lead time. It takes longer to get from uranium-in-the-ground to uranium-in-a-reactor than it takes to get from coal-in-the-ground to coal-in-a-furnace. It takes a long time to ramp up uranium reprocessing or any other exotic solution. It takes a long time to build a nuclear power plant or to get an experimental result to commercial viability.

So there's something here, but it doesn't seem like a bombshell. Basically, the Red Book numbers are inaccurate and vague in many different ways, but used in policy arguments as though they were very accurate, because they're the only game in town. It's possible the numbers are so far off the mark that there will be a uranium price shock in 2013, but it seems pretty unlikely.

I'm interested in hearing the opinions of those who know more about these topics.

Here's a story that seems unbelievable in the Internet age, but it actually happened. For my science fair project in fourth or fifth grade, I decided to explore an idea that had occured to me in a flash of insight: that the ancestors of today's whales lived on land, and were dinosaurs. I saw this project through to completion without ever discovering the truth: that whales are mammals and so their land-dwelling ancestors were also mammals. That while the dinosaurs have living descendants, they're not mammals but birds.

I can explain in a couple sentences what's wrong with my old science project because I now have a basic grasp of evolution and the history of life. That information was not available to me in 1988. We had a World Book encyclopedia that probably had this information somewhere but not under any heading I would think to look under, and certainly didn't discuss the evolution of whales under "Whales". I think I was able to wrangle a trip to the college library to get a better picture of a whale skeleton so that I could point out artifacts of the whales' land-locked history. Nowadays, if I go to Wikipedia, the first and last reference for grade-school science projects, I see on the second screen: "All cetaceans... are descendants of land-living mammals." Case closed.

I'm not going to blame my tools. I must have blocked out contradictory evidence--I'm pretty sure I knew about Archaeopteryx. And I knew at the time that my basic hypothesis, that dinosaurs didn't go extinct, was based on unscientific wishful thinking. But the tools made the difference between this being a passing fancy that was validated in an unexpected and fruitful way ("Aha! Birds! More generally, evolution!"), and something I did a lot of work on without managing to notice it was totally wrong.

If you're like me, you probably lack the attention span neccessary to read through seventeen pages of the World Question Center finding out what eggheads have changed their minds about. You're stymied by the fact that there's no consistency whether the headline of an entry is the proposition the author changed their mind to, the proposition they changed their mind from, or just the topic on which they changed their mind.

But you're probably not enough like me to decide to go through the whole thing anyway and post a best-of, the entries that stuck in my mind as telling me something new and interesting. Which I just did. Enjoy!

• If you only read one entry, you're probably a very impatient person, so I recommend you read the shortest entry, which is Alan Krueger's.
• There are many, many entries about religion and atheism. The only one with new ideas is Clay Shirky's. I'd also like to point out Patrick Bateson's and Alan Alda's, because they're right next to each other on the same page, and although they agree with each other, their changes of mind involved exchanging labels.
• Also many entries on how to do science. I think the best one is George Church's. Related is falsifiability, and for that the entry to read is Rebecca Goldstein's. Robert Shapiro has the flip side, on "Smothering Science with Silence". Geoffrey Miller is good too.
• The "awwwww" prize goes to Daniel Gilbert. His is also the most interesting of the many meta-entries: entries about the act of changing or ability to change one's mind.
• Also a lot of ones about climate change, but the best one is probably Stephen H Schneider's. If you just want to hear how screwed we are, try Steve Connor.
• Daniel Dennett has the best out of many entries on the origin of consciousness.
• Oliver Morton's entry on manned spaceflight had more effect on my own opinions than any other entry. Linda Stone's entry on breathing exercises was a close second.
• Most interesting answer from the Dyson family: George Dyson on Russian America.
• People I've met: John McCarthy, Tim O'Reilly, Chris DiBona.
• People mentioned in Future Stuff: Jaron Lanier
• There were several entries on the mind-body problem, and John Horgan's is the one to read.
• Denis Dutton has the most interesting entry on evolutionary biology: a discussion of teleology.
• Miscellaneous hit parade:
  • Joseph LeDoux on memory formation and maintenance.
  • James O'Donnell on taking sides in history.
  • Timothy Taylor reveals the Incas trying to cheat their gods.
  • Paul Steinhardt speaks out against inflationary theory.
  • Daniel Engber on animal research and institutional incentives.
  • Stephen Kosslyn on the interaction between genes and environment in the programming of the brain.
  • Sherry Turkle has a creepy story of people choosing the ELIZA effect over real emotional interaction.

A decent heuristic for seeing whether something is real science is to check whether other places have it too. This is probably the flip side of Jared Diamond's idea about why China's early lead in science and exploration died out (more on this later, actually).

Everyone has relativity and genetics and evolutionary biology and quantum physics and plate tectonics, but only the US has creationism (similarly for intelligent design, or as I call it, c******nism). Only the USSR under Stalin has Lysenko-style evolutionary biology. Everyone has astrology and alchemy, but everyone's astrology and alchemy is different. And only the UK has this special newspaper-based "formula for the perfect x" branch of science.

This is where you take something vague, dissect it into a bunch of vague contributing factors, assemble them back into a vague equation, and claim that it represents perfection. This gets you in all the papers and nowadays the weblogs as well. As far as I know only one article two years ago has ever dared to point out the ridiculousness, and it's still going on.

Oddly, NYCB has previously encountered Len Fisher, unwitting founder of this particular branch of pseudoscience. He's the one who started it with his measurements of liquid absorption in biscuits, and his use of a real, preexisting equation to describe it. I think he deserves some kind of meta-Ignobel for the monster he's created.