You may only read this site if you've purchased Our Kampf from Amazon or Powell's or me
• • •
"Mike and Jon, Jon and Mike—I've known them both for years, and, clearly, one of them is very funny. As for the other: truly one of the great hangers-on of our time."—Steve Bodow, head writer, The Daily Show

"Who can really judge what's funny? If humor is a subjective medium, then can there be something that is really and truly hilarious? Me. This book."—Daniel Handler, author, Adverbs, and personal representative of Lemony Snicket

"The good news: I thought Our Kampf was consistently hilarious. The bad news: I’m the guy who wrote Monkeybone."—Sam Hamm, screenwriter, Batman, Batman Returns, and Homecoming

April 11, 2011

Morbid Fun With Graphs

By: Aaron Datesman

I wrote here about my dissatisfaction with the dose model, although probably without enough scientific support for the part everybody focused on at the end. (I’ll return to it eventually, since I think it’s pretty interesting. There were especially some smart comments which enhanced my understanding and helped to clarify my own thinking.) The post contained one other point which is at least equally significant: the dose model lumps the distinct biological effects of different kinds of emitters together into one number. This isn’t biologically justified; it’s “computationally convenient”.

Although I first encountered the linear dose model in 1996, I didn’t think to question it much until the end of my short career in the nuclear industry. My skepticism emerged during the last several months of 2005, which I spent writing a very long report on radiation effects in (non-biological) materials. As November slipped into December, and the stack of academic papers and government reports on my desk passed twelve, and finally twenty-four inches, I was overwhelmed with conflicting information about defect mechanisms, thermal healing rates, and dose measurements for different radiation spectra. So, what did my colleague and I do?

We did what engineers are trained to do: we picked the papers we understood the best, summarized the results for our bosses, and then constructed a linear model from which we extrapolated recommendations for experimental study. The report is still classified, or I would share it here. We were aware that our report failed to capture very much insight compared to all of the literature we had surveyed; but it’s standard operating procedure, so it’s what we did.

A silly but accurate explanation of the problem we confronted goes like this: the dose model takes apples (alpha), bananas (beta), and grapefruits (gamma), transforms them into pineapples, and then feeds us pineapples in all sorts of forms (in cake, as fermented juice, by throwing them at us, and many others), in amounts small and large, over time periods short and long, in order to determine whether we are allergic to apples, bananas, and grapefruit. If anything, this is a bigger problem for animate biological materials than it is for inanimate materials.

I think about this whenever I hear the claim that, since “background radiation” is safe, the small dose increment which human activities have added to the background (around several percent in most locations) also must be safe. This is wrong. For instance, a commenter pointed me toward the article “Cancer Incidence in an Area Contaminated with Radionuclides Near a Nuclear Installation”, which contains this worrisome piece of information:

The DOE station at the eastern (downwind) boundary of the plant has recorded an average concentration of 2072 attocuries/m^3 (aCi/m^3) of plutonium over the eight year period, compared to 32 aCi/m^3 for New York City….

That 32 aCi/m^3 of Plutonium measured in New York City is a small contribution to the activity of the background radiation, but understand this: there was never one atom of Plutonium on Planet Earth until humans made some. Even if the naturally-occurring background radiation (principally due to radon and cosmic radiation) is safe for humans (and we have no way to know this, since there can be no counterfactual), there is absolutely no scientific basis for lumping any amount of plutonium contamination into the natural background, and concluding that miniscule amounts of plutonium in the environment therefore are also safe.

It is absolutely unjustified. Not only is the conclusion of safety absolutely unjustified for plutonium, but it is also almost completely unjustified for other radioactive materials released by nuclear weapons testing and by releases from nuclear reactors, including strontium, cesium, and iodine. Converted to pineapples, the levels of these radionuclides are indeed small; however, compared to their near-zero natural abundances, the levels of these materials are vastly larger than any sensible definition of a “background” level.

But how to tell whether these low levels of released materials are harmful? Due to confounding factors including smoking, chemical exposure, and not calling your mother often enough, it’s essentially impossible to tell directly. The issue can only be resolved experimentally: for instance, by large-scale epidemiological (medical) studies. As a thought experiment, however, I might suggest shutting off all of the nuclear reactors all over the world while monitoring the rate of cancer mortality. Maybe even just turning them down would be enough, in fact.

I thought of this as I browsed through Annex C of the 2008 Report of the United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR). It is one of the documents upon which George Monbiot bases his new pro-nuclear opinions.

I agree that the report is very interesting, although probably Monbiot and I did not find the same parts interesting. What I found interesting was the section about radioactive effluents from routine operations of nuclear reactors (pages 183-187). Somehow, despite my brush with the nuclear industry in 1996 and my employment in the industry in 2005, until recently I was under the impression that no such thing as releases of radioactive effluents due to the routine operation of nuclear reactors exists.


UNSCEAR also provides a handy graph, which I reproduce below. An interesting feature of the graph is that it provides a line for “Electrical energy generated”. You should note that, inherent to the process of generating power from fission, the more energy generated, the more radioactive byproducts are produced. (They don’t have to be expelled to the environment, but they must be produced. The relationship is proportional.)


Thanks to the internet, which makes many things easy that used to be impossible, it occurred to me to look up the incidence of cancer in the U.S. by year. I didn’t spend long on it; I’m sure I could find a better source. But all I wanted to do was to make two graphs, mash them together, and think about them. So, I did:


I acknowledge all sorts of problems with this scientifically, and myself don’t consider it anything more than a picture to go “Hmmmm…..” at. Therefore, although I’m tempted to write more about it, I think I’ll allow the graph above to speak for itself.

— Aaron Datesman

Posted at April 11, 2011 06:08 PM

The problem with closing all the reactors and looking at cancer rates is that the rates will probably go up. Then you will have to explain how nuclear power was preventing cancer. This is, of course, all too complicated to measure because of our tobacco industry doing such a good job of promoting smoking in the third world, our destruction of the ozone level, and simply by the increase in life expectancy around the world. Cancer is largely an old persons disease, so the longer we live the higher the rates of cancer.

Posted by: John at April 11, 2011 09:20 PM

Stupid question, but is it valid to overlay the last graph like that when the one set of data is plotted on an exponential scale and the other is not? I ask because it seems to me that if the death from cancer rates were done on a similar scale it would be pretty much a horizontal line and would imply no relationship.

Posted by: aric at April 11, 2011 09:43 PM

Aric, it's a good question. Read tomorrow. All will be made terrifyingly clear.

Posted by: Aaron Datesman at April 11, 2011 09:55 PM

Okay, I'm losing all respect for this effort, sorry.

First off: your link reads the incidence of cancer but sends me to a table listing death rates of cancer. Not even close. If you don't know the difference between incidence and death rate, you're out of your element, Donnie.

Part 1A: lots and lots of things factor into incidences (ie. diagnoses, which depend hugely on technology and changing concepts) and death rates (for diseases of aging like cancer, in part reflecting the LACK of death from other disease) that have NOTHING to do with an actual change over time in the actual biology of actual human beings.

Second: you graph the "pre-1970" death rate from cancer as ~160. That number is what the table you link to lists for 1970. Numbers from "pre-1970" actually show a dramatic increase decade by decade from the very beginning of the table, which you don't graph:
1900 = 64
1910 = 76
1920 = 83
1930 = 97
1940 = 120
1950 = 140
1960 = 149
1970 = 160

So it's not like the death rate from cancer was holding steady at 160 for years and years before nuclear power plants came along. The death rate actually DOUBLED from 1900 - 1940, before the atomic bomb was developed, and has failed to double from 1940 until now.

Of course, the fact that it doubled probably has nothing much to do with a change in your personal risk of developing cancer so much as an increase in your risk of living long enough without dying of something else that cancer is listed as your cause of death. See part 1A above.

Numero trois: You plot exponentially the "total collective dose" of radiation, which increases from ~50 to ~1000, roughly a twentyfold increase. You make "10" (10^1) on the y-axis for radiation dose equal to "150" on the y-axis for cancer death rates, and where roughly "5000" would be on the y-axis for radiation (500-fold increase) we find "210" (~1.4-fold increase) for cancer deaths. So you take the increase in cancer death rates from 160 to 200 (increasing by 25%) and make look very comparable to the increase in radiation from ~50 to ~1000 (2000% increase).

I'm not even going to get into the rest of your post, except to pick out one statement:

Even if the naturally-occurring background radiation (principally due to radon and cosmic radiation) is safe for humans (and we have no way to know this, since there can be no counterfactual)

Are you out of your mind? What definition of "safe for humans" involves excluding (inescapable) cosmic radiation? If you went somewhere where the cosmic radiation couldn't reach you, do you think that'd be safer somehow? Are you looking to control for the planet Earth?

By that definition, we don't know that anything is safe, and never will. Which is stupid and trivializes the very useful question: safe compared to what?

Posted by: pmpm at April 12, 2011 04:06 AM

Hmmm, pmpm makes some good points--that cancer dates doesn't seem to me to go anywhere obvious. I'm not surprised that cancer killed more people when TB, pneumonia, and influenza stopped, cuz people are going to die of something. I'm surprised more by the cardiovascular disease data--i assume that smoking boosted that from the 30s to the 80s.

The rest of the post strikes me as making good points.

Posted by: N E at April 12, 2011 06:19 AM

@pmpm - There was a high incidence of toast eaten for breakfast in the Datesman household today, but so far no mortality. Thanks for pointing out the typo. And who's Donny? My name is Aaron.

Yes, I didn't graph the death rates from prior to 1970. So what? The data on the graph show a very close correlation between the death rate from cancer and the logarithm of the number of radioactive fission products generated in commercial nuclear reactors (via the proxy of electrical energy generated) worldwide. The data set covers the years from 1970 to 1997.

Apparently, over those 28 years, it's like there's a big knob in some room, which mankind can turn in order to increase the amount of electrical energy generated from nuclear power. In equal increments around the knob, the label reads "1 10 100 1000" - it's logarithmic. We turned the knob clockwise most years, except that around 1994 we turned it back.

Unfortunately, it appears that Mother Nature has also written another scale, in invisible ink, around the outside of the knob. This scale is linear and reads the death rate from cancer: "160 170 180 190". That is the story the graph seems to tell. Incidentally, this doesn't mean that the cancer scale reads "0" if we turn the knob all the way down.

Regarding #3, the data is binned in five and ten year increments which fail to coincide. So, I took the cancer data from 1970, 1980, 1990, and 2000, and plotted them on 1969, 1979, 1989, and 1997. It would be preferable to have annual data in coincident bins. I agree. There are other technically-educated readers on this site. I welcome their comments regarding how appropriate or inappropriate this presentation of data might be. In my opinion, it is at least approximately correct, and certainly entirely defensible. I have an analagous explanation with puppies and hugs which might be easier to understand. Well, later.....

Look, doesn't it strike you as odd that the cancer rate turns down around the same time that we took some reactors offline? Doesn't that make you say, "Huh?"

About the background radiation, I maintain that what I said is correct: the only way to know with absolute certainty whether background radiation or cosmic rays have adverse health effects would be to have an identical earth where these influences were absent to use as a counterfactual. However, I can say with quite a bit of certainty that cosmic rays harm our health. While I was at Bettis, I spent two days being educated on the topic of radiation fields in space by scientists from the Jet Propulsion Laboratory. We are mostly protected from these effects by the atmosphere, but not completely.

Look, I liked the people I worked with at Bettis, and I could easily have made a career there. Maybe, after 20 or 30 years, I could have been one of the guys on the BEIR XII panel. Or I could have jumped over to Westinghouse to design commercial reactors. I had colleagues who did so. But I do not think I would have gone very far in my career by making simple graphs like the one in this post, and asking the questions which follow from them. Isn't that a significant problem?

It's not my opinion that everybody knows this information and signs a secret oath to cover it up and never talk about it. (In fact, although I can't remember exactly, I think it was somebody at Bettis who first brought this relationship to my attention.) Instead, it's institutional in nature. People who ask these sorts of questions don't make friends, don't get promoted, and eventually grow disillusioned and leave, or are simply fired. It creates groupthink, which is very dangerous.

We should examine our assumptions carefully, including by asking simple questions to which we believe we know the answers. Nearly everybody says levels of background radiation are not dangerous. Is it really true?

If you were alive in 2003, you might remember that nearly everybody said that Saddam Hussein was sitting on huge stockpiles of deadly weapons. Why is this different?

Posted by: Aaron Datesman at April 12, 2011 11:31 AM

I think this is a good presentation of data for a thought experiment, which all scientists and critical thinkers should be able to perform without getting hung up on the details. A general point I am receiving (so far) from this series is that the data is often murky and over-extended at best, so looking at larger trends with an open mind and the ability to see past the conventional wisdom will reveal more worrying statistics than we have ever been led to believe. Looking at it, especially now with the recent re-classification of the scope of the Fukushima meltdown to a 7/7 on the nuclear disaster scale, I think this is definitely more plausible than other commenters are admitting.

Posted by: Gordon at April 12, 2011 01:52 PM

"Incidence" vs "death rates" isn't a typo, it's a glaring conceptual error that suggests you're completely unfamiliar with the subject.

Many cancers commonly attributed to radiation aren't fatal (because they're treatable, some are even curable). Many cancers commonly attributed to radiation only develop after an interval of decades after exposure. Many of those cancers progress gradually, over a number of years, and are not acutely fatal.

So it would in fact strike me as odd that a decrease in the cancer death rate occurring the same year that some reactors were taken offline could be attributed to the change in radiation exposure that same year. You'd expect a delay of years before such a change occurred, even if they were causally connected. And you'd expect a similar delay of years before being able to assess the effects of a specific radiation exposure, which is one reason why Chernobyl epidemiology is a hot topic right now, 25 years later.

It will take years before the effects of Fukushima are known. And it won't be a surprise if the cancer incidence attributable to Fukushima peaks several years from now, even if there's a significant downturn in nuclear energy production in the meantime.

Posted by: pmpm at April 12, 2011 02:34 PM

@pmpm - Now, really. Be nice. I got typed neon when I meant xenon in a different post, and I know the difference between those, too. I have endured unbelievable amounts of radiation effects safety training, courtesy of the Department of Energy. I probably know a few things.

I agree with you about the lag time between cancer initiation and mortality. It puzzles me. So, maybe the relationship in this graph (really, it's just four points) is a coincidence.

Or maybe cancer genesis is more complicated than I understand. I don't know. Do you? My impression is that probably nobody really knows for sure.

Exposure rates and times and intensities all vary over many orders of magnitude, for different emitters, with different disease mechanisms and dose pathways. Yet we roll them all into one single model via the concept of "dose". How much sense does this make really? Is it necessary that only one thing is going on?

If you know, please share. Share scientific references you think I should read, too. I'll look them up at work.

Posted by: Aaron Datesman at April 12, 2011 04:17 PM


FWIW Donny is probably a reference to The Big Lebowski:

The big guy is Walter, the film's asshole.

Posted by: Some other dude at April 12, 2011 06:18 PM

I don't think it's really fair to call Walter an asshole. He has his transcendental wisdom:

Any thread that takes me to the Big Lebowski is okay with me, but I am disappointed that there's no secret oath.

Posted by: N E at April 12, 2011 06:54 PM

My primary concern with the presentation is the existence of only 4 data points. A suggested affine relationship would only leave 2 degrees of freedom for hypothesis testing. I, personally, won't draw any firm conclusions from the data presented. This reminds me of the general problem of optimization: the process of optimizing has costs. Employing optimization techniques to their fullest may sometimes not be optimal because of these costs. I'm sure our industrial society is quite harmful in a variety of ways but it is also quite helpful in a variety of ways. I think, overall, the increase in cancer mortality over the last hundred years is a good thing as we have been better at fighting other methods of death. Sometimes it is better to proceed in ignorance then it is to wait for more information. The problem is telling when it is better to do so and when it is worse and I have no good response to that problem.

Additionally I, in my inexpert opinion, suspect that the natural background radiation (radon and cosmic rays) is safe for humans. We are a species that is extremely well adapted for conditions on Earth a few of tens of thousands of years ago that I would be surprised that if we are not also well adapted to the levels and kinds of radiation that existed then too.

Posted by: Benjamin Arthur Schwab at April 12, 2011 07:22 PM

It also appears that you are comparing the US cancer incidence with the global production of nuclear power. I do not believe they are equivalent. Could you not find global cancer incidence or US power production figures?

Posted by: John at April 12, 2011 10:40 PM

Either way, WE're gonna git some, an' WE're gonna find out.

Posted by: Mike Meyer at April 12, 2011 10:51 PM

Tell it to The Dude:

Posted by: some other dude at April 13, 2011 10:58 AM

I wonder if this simple observation might put this passive aggressive quibbling to rest:

The data is purposefully out of phase and figuring that in supports a correlation that can be interpreted as predictive.

Posted by: LT at April 13, 2011 11:08 AM

Some other dude


Posted by: N E at April 13, 2011 05:59 PM

off topic

I just read the full report by the Northern Kentucky University professor Brad Scharlott, “Palin, the Press and the Fake Pregnancy Rumor: Did a Spiral of Silence Shut down the Story?” Briefly, yes. When you put it all together, it's very unlikely that it's Palin's baby. More likely it's the daughter's. It also was probably born before the stated birth date. The issue is important not because of what it says about the baby, or the baby's real mother, but because of what it says about (a)Gov. Palin, and (b)the way the press has dealt with Gov. Palin. There's a good interview with Prof. Scharlott at

Posted by: mistah charley, ph.d. at April 14, 2011 03:06 PM

"Read tomorrow."

Today is tomorrow.

Posted by: godoggo at April 14, 2011 07:12 PM

@MC: Not just that, but Trig's dad is the same CIA guy who bumped off JFK and then dispatched Lennon, too -- all on Castro's orders, with funding from the Queen of England's drug dealing business. It's a nasty world out there.

Posted by: bobs at April 14, 2011 07:57 PM


Posted by: Aaron Datesman at April 14, 2011 09:50 PM

As a reformed trying-my-best-to-do-better, always-wise-cracking, sarcastic cynic, just like bobs (though I hope upon hope wittier and more clever), I can definitely attest to the fact that there is nothing that will make a person more insufferably opinionated and useless (simultaneously no less!)than that little package of traits he so succinctly displayed with such pinache.

God bless America, we deserve everything we are going to get. Most of us anyway.

Posted by: N E at April 14, 2011 10:08 PM

THAT'S why they call him "Poppy", he absolutely refuses to wear a rubber.

If I may point out, NOW is an ideal time to get those "low dose/cancer ratios, short and long run", especially since the Japanese have yet to get ahold of their reactor leakage issues. I'm sure OUR dosage IS going to increase over the short term, and the long term.

Posted by: Mike Meyer at April 14, 2011 10:09 PM

THAT'S why they call him "Poppy", he absolutely refuses to wear a rubber.

If I may point out, NOW is an ideal time to get those "low dose/cancer ratios, short and long run", especially since the Japanese have yet to get ahold of their reactor leakage issues. I'm sure OUR dosage IS going to increase over the short term, and the long term.

Posted by: Mike Meyer at April 14, 2011 10:10 PM

Wait, I meant 'panache' above!

Aaron, don't blame Bush, that's one of the few things he isn't guilty of.

mistah charley--I haven't read about that in a while, probably because it's just too crazy to get pursued, from a social perspective, but there certainly are many odd questions about little Trig's birth. Imagine Sara Palin being that phony and cynical! How could she do that? How could SHE get away with that?

The thing is, it's incredibly easy to mock and ostracize anyone who claims that someone like Sara Palin has done something that crazy and outrageous and not been called on it by the media or anybody else. "No one could get away with that!" Plus, the fact that she's a certain sort of moron makes it hard for more intelligent cynics to think she could do it. "SHE could never get away with that!"

Au contraire, outrageous acts are extremely difficult to allege when everybody is forced to believe, or at least profess to believe, that they can't be successfully carried out. Cynical and sarcastic people especially perpetuate this folly, because we don't believe a damn thing. That sort of skepticism is easy to use by the users of skepticism, who are professionals with considerable expertise (because bullshit is America's foremost product). Many cynics apparently don't even realize that ads can be targeted at cynics--I think the irony of that is killing.

So anyway, everyone can chuckle their heads off. This stuff happens all the time and has since the dawn of time, which is why ignorance of real history is so important and people have to just crack jokes about Castro and the Queen and JFK, hardy har har. Because if people took the time to learn real history (which is a bit time-consuming) they might quit being so goddamn stupid and spineless.

Sorry, I'm stressed from battle against the greedy evildoers right now, and I'm tired of self-impressed stupidity. Thanks for the link.

Posted by: N E at April 14, 2011 10:57 PM

NE - U & me both. Let's have separate beers tonight in solidarity.

Posted by: Aaron Datesman at April 14, 2011 11:40 PM

Aaron, I will be proud to join you for a separate beer in solidarity anytime! But I have to confess your invitation, though lovely, does remind me of a girl who once suggested that we might try going out separately . . .

Posted by: N E at April 15, 2011 07:13 AM

Indeed, N E, the self-inflicted ignorance of those who doubt the Big Lie theory of the "Trig hoax" is related to the "spiral of silence" journalism prof. Scharlott identifies. Anyone who, like columnist Andrew Sullivan, pays attention to Sarah Palin's own assertions about the circumstances of the birth of Trig (called "The Wild Ride") must find them very implausible. The preponderance of the evidence establishes that Sarah Palin did not give birth to this child. Who did, and when, and where, and impregnated by whom, are matters of conjecture - but that it was not Sarah Palin cannot reasonably be doubted by anyone willing to look at the publicly available evidence (which does NOT include a birth certificate, by the way, or a statement by a hospital where the child was born, or by a physician who attended the delivery).

Posted by: mistah charley, ph.d. at April 15, 2011 07:52 AM

But I do not think I would have gone very far in my career by making simple graphs like the one in this post, and asking the questions which follow from them.

For crying out loud. A couple of posts back you derided the scientists for the very exercise of making simple graphs. If you simply meant that a logarithmic model is preferable to a linear model, you chose a passing strange way to make that point.

Posted by: t.m. at April 15, 2011 08:15 AM

@t.m. -

Thanks for the comment, but your criticism is wrong. I have no problem with making simple graphs. Generally, in fact, there's no other way to go forward.

The big problem I have is with basing judgments of safety on simple graphs which are known to represent assumptions rather than knowledge.

How many people are aware that the announcements that Fukushima fallout is too dilute to be harmful utilize model assumptions which are not strictly known to be correct?

That's where I have a problem.

The graph in this post is just a little something to think about. Actually, since it encouraged me to consider the implications of a logarithmic relationship, I'm quite happy with it! This doesn't mean I think it's very well-supported.

I appreciate the criticism though. Thanks for keeping me somewhat less dishonest.

Posted by: Aaron Datesman at April 15, 2011 11:22 AM

mistah charley

Agreed, but I will note that it's one thing to say that the preponderance of the evidence shows something, quite another to say that no one can reasonably doubt it. I have no problem with people doubting--I believe that's actually good. What I have trouble with is people replacing thought with a reflexive disbelief that really amounts to the opposite of doubt, especially when a large amount of top-down social control is exercised to create this orthodoxy of perception and adherence. I didn't read Professor Sharlott's article closely enough, or pay enough attention to Andrew Sullivan's earlier writings about this a couple of years ago, to form any doubt-free views about whose baby Trig is. You may well have doen so, and your conclusion I tend to trust. I just haven't had time to focus on it that much, and believe it or not, I don't like to reach big conclusions too easily.

That being said, Professor Sharlott is clearly right about how the media has treated the story. It has obviously been dismissed as implausible--basically impossible--without actual evaluation. And I think that is routine. We all have opinions about a lot of things that we don't actually understand well or know much about, basically of necessity, and yet we mostly dismiss views that challenge our preconceptions without any real evaluation. Usually we do that with sarcasm or quips or some other crap, because that's more fun than this comment.

And that's fine--some views are bullshit on their face. Whatever problems he has, Obama really isn't a foreigner raised as a Manchurian Candidate to destroy America. Whatever you think of Israeli behavior, the Jews didn't kill Christian children to use their blood in rituals. Even if OBL really did mastermind 9/11, Muslims aren't trying to bring Sharia law to the U.S. I'd say all sorts of crap obviously isn't true, and I reach that conclusion without much new evaluation, based on what I already know. That's what we all think we do, and maybe it is.

The problem arises when the ideas people dismiss as crazy aren't, and when those people don't dismiss as crazy are. (There really is an objective reality out there.) That seems to me to be where we collectively find ourselves. So nobody is going to be allowed to seriously consider whether Sara Palin is really such a fraud and phoney that she could pass off her grandchild as her child without anyone asking questions, because hell, if she could get away with that, we'd have to question everything.

Better to believe anything than do that.

Posted by: N E at April 15, 2011 07:11 PM