Open threads at the Open Thread tab every Sunday and Wednesday

Meat Your Doom

[Epistemic status: Very dirty and approximate, but I think roughly correct. Check my calculations and tell me if I’m wrong.]

A recent formative experience: a seriously ill patient came in and I recommended a strong psychiatric drug. She looked it up online and told me she wouldn’t take it because was associated with an X% increase in mortality.

“But,” I pointed out, “you’re really miserable.”

“But I don’t want to die!”

So I looked it up, did the calculations, and found that it would on average take a couple of months off her life. And I asked her, “Which would you prefer – living 80 years severely ill, or living 79.5 years feeling mostly okay?”

She still wasn’t convinced, so I asked her if she ate cookies. She said yes, almost every day. I told her that the cookies were probably taking more time off her life than the medication would, and I assured her the medication would probably add more value to her life than cookies.

She took the drug.

I thought of this the other day when everyone started sharing that study about meat causing colon cancer. A lot of people used headlines like Processed Meats Rank Alongside Smoking As Cancer Causes. This was very correctly debunked by infographics like this one:

But I feel like this leaves something to be desired. Eating meat is not as bad as smoking. But that’s still a lot of room for it to be bad. Can we quantify the risk better?

From the BBC article: “‘[There would be] one extra case of bowel cancer in 100 lifetime bacon-eaters,’ argues Sir David Spiegelhalter, a risk professor from the University of Cambridge.”

This teaches us something important: “risk professor” is an awesome job title and “David Spiegelhalter, Risk Professor” ought to be a BBC television show starring Harrison Ford.

But also: use absolute risk instead of relative risk! “21% of bowel cancers are caused by meat” doesn’t give you a really good handle on how worried you should be. “One extra case of bowel-cancer in 100 lifetime bacon-eaters” is better.

But let me try to give even more perspective. A bit less than half of colon cancers are fatal. So one extra case per hundred means if you eat bacon daily then there’s an 0.4% chance you will die from a cancer you would not otherwise have gotten.

The average age at diagnosis of colon cancer is 69; the average life expectancy is 79. Sweeping a lot of complexity under the rug and taking a very liberal estimate, the average death from colon cancer costs you ten years of your life.

Multiply out and an 0.4% chance of losing 10 years means that you lose on average two weeks.

Suppose that every case of cancer, fatal and non-fatal alike, causes you additional non-death-related distress equal to two years of your life. That’s about another week.

So overall, if you eat processed meat every day your entire life, you’ll lose about three weeks of life expectancy from colon cancer. That means each serving of meat costs you a minute of your life. You probably lose twenty times that amount just cooking and preparing it.


Note that I am not saying “eating meat will only decrease your lifespan by three weeks”. That is the amount that we have clear evidence for, from this study. It is an example of why this study needs to be put in context so that you don’t worry about it too much.

There are nevertheless a lot of other studies that suggest greater risks, mostly cardiovascular or metabolic. For example, as per this article, some studies suggest that a serving of red meat per day increases mortality 13%, and a serving of processed meat per day increases it 20%. But it also quotes another study of half a million people that finds meat to be slightly protective (sigh) and finds a higher all-cause mortality in the non-meat-eaters.

Whatever. Forget the object-level question for a minute. What are we to make of a claim like “processed meat increases mortality 20%”?

If you’re like me, you want to think “Okay, average life expectancy is eighty years, subtract 20% off of that, and you get 64 years. I’ll live 64 years if I eat bacon every day.” WRONG. Mortality rates are much more complicated, but the key insight is that very few people die when they’re young. If you have approximately a 0% chance of dying at age 30, then adding 20% to 0 is still 0. Chance of mortality creeps upward very slowly and so even large changes in mortality barely affect the underlying distribution. The only good presentation of this I have ever seen anywhere is on Josh Mitteldorf’s blog, which includes the following chart:

This is decrease and we’re talking increase, but but it shouldn’t make much difference here. A 20% increase in mortality isn’t going to bring you from 80 to 64. It’ll probably just bring you from 80 to 78.

Indeed, later in the BBC article, they bring in David Spiegelhalter (RISK PROFESSOR!) who explains that:

If the studies are right…you would expect someone who eats a bacon sandwich every day to live, on average, two years less than someone who does not. Pro rata, this is like losing an hour of your life for every bacon sandwich you eat. To put this into context, every time you smoke 20 cigarettes, this will take about five hours off your life.

That’s for processed meat. Red meat is safer. Also, we still don’t know if these studies are right.

This is why it’s important to distinguish between absolute and relative risk. You hear all of these scary numbers – 21% increase in bowel cancers! 20% increase in all-cause mortality! – and it sounds like you’re going to drop dead the moment you take a bite of a hot dog. And there’s always that chance. Being healthy is good. Being unhealthy is bad. But is life so dear or peace so sweet, that you’re never going to want to sacrifice an hour to have a bacon sandwich?

All these hours do add up. I’m not saying dietary recommendations aren’t important. But the recommendations are important in aggregate. If you stick to the spirit of not eating in a horribly unhealthy way, you have a lot of leeway to continue to eat specific things you like even if you know they’re not the best for you. And meat falls firmly within that category.

(though you might also want to consider how to manage the moral issues)

This entry was posted in Uncategorized and tagged , . Bookmark the permalink.

201 Responses to Meat Your Doom

  1. Dave Orr says:

    One question I always have: how much should I adjust by beliefs based on a study like this? How does that affect my estimate of how much life I’m giving up for delicious, delicious bacon?

    If I think that half the time or more that a study gets press will later fail to replicate, can I just multiple that in? 50% chance that this is true times increased chance of cancer times…

    No, obviously not, because we don’t know which direction the study is wrong in. It could be that it’s an underestimate. It’s much less likely to be an underestimate because of publication bias, but we still need to account somehow for the fact that it *could* be an underestimate.

    Pulling numbers out of thin air, we could say that 90% of the time that it’s wrong it’s an overestimate, so now we have (1-(.9 * .5)) * .004 * 10 years = .022 years, or 1.15 weeks. 8 days.

    Is this a remotely sensible way of trying to take into account that the study might be wrong?

    • Nornagest says:

      I trust nutrition studies just slightly more than I trust the guy who hangs around the Powell Street BART station carrying a sign about Sodom and Gomorrah.

      • onyomi says:

        Just assume that whatever tastes good is probably bad for you.

        • It’s probably more accurate to say that anything which tastes good will be claimed to be bad for you.

          • chaosmage says:

            Well sure. If you say something that tastes good is actually good, or if you say something that tastes bad is actually bad, nobody cares.

          • onyomi says:

            Actually, I find there is a huge market for telling people good news about their bad habits: how often do you see a news story saying, in effect: “hey, it turns out chocolate is really good for you!” or “drink more wine!”

            In contrast how often do you see a news story like, “scientists confirm: broccoli is good for you”?

          • onyomi says:

            And I am, of course, sort of joking, but also sort of not. As Dave Barry notes, “a calorie is a measure of how good a food tastes.” I think for most Americans, worrying exactly where they get their calories is premature: they need to first worry that they are just getting too many of them. I recall experiments where all of people’s test results improved after losing a bunch of weight on a diet of nothing but twinkies.

            Last night Mike Huckabee kept going on about how we need to declare war on heart disease, cancer, etc. And I kept wanting to say, “you mean, declare war on McDonalds and Krispy Kreme?”

        • Sokka says:

          That seems like the opposite of the evolutionary perspective.

          • Oscar_Cunningham says:

            If something isn’t on the taste-health optimality frontier then there’s little reason for people to eat it at all. But among the things people actually eat health anticorrelates with taste. Like eating shampoo would be bad for you and taste horrible.

            Someone did a post on this kind of situation on LessWrong.Ah, here it is


          • onyomi says:

            Agree with Oscar, though the simplest explanation is just that the food environment first-worlders are living in is radically different from the one we evolved to survive in. For 99% of our history, sugar, fat, and calories in general were scarce, fiber and sinew plentiful. In those circumstances, if you came upon a dense source of calories, like an unguarded honeycomb, it was adaptive to gorge yourself on it in a way that it clearly isn’t now.

            Now we’ve concentrated all the fat and sugar and calories and removed the fiber and sinew that get in the way (both of the eating and the digesting), causing us to be massively overnourished.

      • Saul Degraw says:

        I prefer the person who once hanged out at Powell or Montgomery BART with a sign that said “Buddha loves everyone.”

    • AnonymousCoward says:

      I think you do have some idea in which direction it is more likely to be wrong. The bias that leads the media to report on it, as well as to the research being published in journals in the first place, is in the direction of stronger results. I think you can assume the truth will regress to the mean a decent fraction of the time, depending on how often studies like this don’t replicate.

    • gwern says:

      I would argue that the correct approach here would be subjective Bayesian decision theory served with a side of meta-analysis. Something along the lines of:

      1. meta-analyze the studies in a multilevel model, including at least an attempt to correct for publication bias; extract the posterior predictive interval to get a distribution of values for the correlation between a dose of bacon and bowel cancer. (Note that prediction intervals are not the same as the mean of the risk parameter, since there will be variation from population to population and study to study, so the credible interval around a parameter will be different from the risks realized in various populations, including yours.)
      2. now that you have a distribution over the correlation, adjust for correlation!=causation; my own suggestion, based on papers systematically comparing RCTs against the correlations they were based on, is that around a third of correlations are causal. You can implement this as a mixture model of two groups where one group is the correlation distribution (P=0.33) and the other is just 0 (P=0.66), giving you a spike at zero causal effect and spread out.
      3. with that distribution over causal effects, you can bring in the QALY loss of getting bowel cancer and turn it into a distribution of loss. (I’m sure someone has already calculated the QALY for bowel cancer, so you don’t have to.)
      4. you can then compare mean expected loss per bacon units with your expected utility per bacon units; due to diminishing returns, your bacon curve will look like a parabola, and you can find the pair which maximizes the sum. It might be 0 units of bacon, it might be more.

      Of course, that’s all a lot of work, but that’s what we have Risk and Decision Professors for, right?

      No, obviously not, because we don’t know which direction the study is wrong in.

      Overly-large. Publication bias & the significance filter, p-hacking, small numbers, lack of regularization/informative-priors, lower methodological quality (eg. lack of blinding and randomization are associated with consistent increases in effect size, not simply increasing variability with no net effect), all lead to overestimates. Underestimates will be much rarer. And to the extent that you are asking about the causal conclusion, you can be that much more confident that it’s an overestimate simply because in the common case that it’s noncausal, the causal effect is ~0… (And so the mean estimate will be much lower when all those 0s get averaged in, although to the extent supported by the data, my suggested mixture group approach does allow for occasional causal estimates to be higher than the mean correlation estimate.)

      • Sam says:


        “We don’t know which direction the study is wrong in” is an important case of misapplying the principle of indifference.

      • chaosmage says:

        > my own suggestion, based on papers systematically comparing RCTs against the correlations they were based on, is that around a third of correlations are causal

        You just blew my mind. This seems epistemologically legal, but it would never have occurred to me, and it is super useful.

        Thank you.

  2. Grant Gould says:

    David Spiegelhalter (RISK PROFESSOR!) is also a regular guest on the BBC’s podcast “More or Less” which is a pretty awesome podcast devoted to just this sort of thing — taking numbers from the news and explaining which ones are big and which ones are small and what they are going on about.

    So if you want more RISK PROFESSOR then that is the place to get it.

  3. Alicorn says:

    I don’t like the “it costs a minute extra of your life to make the vegetarian meal” formulation compared directly with earlier death. I have no reason to quibble with your math per se and don’t even have a reason to object to the underlying argument, but I don’t think that’s how life works unless cooking is worse than oblivion for you. I’mma quote my own novel:

    “Just because you would have gotten money to pay for the vase by working servant jobs for a few weeks doesn’t mean that is all you would have gotten for those few weeks,” said Meea, picking her way slowly and distinctly through the sentences. The other students were listening in too, their exercises forgotten. “You would have gotten to live them. Time you lose pausing is not time you get to spend living, not even living through your least favorite chores. At the end of those weeks of work, however wasted you think they might have been, you would have learned new things and gotten closer to your friends and family and accomplished something valuable enough for the Temple-Guild that they’d have traded money for the work. Instead you destroyed a few hours, maybe a couple of days, of your life, and during those lost hours or days you will do nothing, not even cook or care for horses or attend your relatives.”

    (Context: recipient of this speech has used magic that manipulates time at the cost of total lifespan in an ill-advised way to save some money.)

    • Eric Rall says:

      That’s pretty much the point I came here to make. It’d have to be at least two extra minutes to cook a vegetarian meal, since I expect I enjoy cooking at least twice as much as I’d enjoy being dead.

    • another_anonymous says:

      I’d just like to say that when I read the above comparison of the times “lost”, I thought of exactly that passage from your (Alicorn’s) novel 🙂 (I obviously agree with your point).

    • anon says:

      My life has immediate-term negative value when I’m doing something I really don’t like. I can expect the future consequences of this to be worth it, or do it to avoid greater pain, but it’s not valuable. I would absolutely take a few hours off my total lifespan over a few weeks of unpleasant work. The inherent value of living in general isn’t *that* high.

      • Richard says:

        An hour on a bicycle gives an average person an hour extra life span. (Disregarding diminishing returns and a lot of other complexities.) If you enjoy riding a bicycle that’s two hours of enjoyment, if you hate bicycling, it’s essentially zero so you might as well stay on the couch.

        (This is a part of my standard spiel pro-exercise, though taken out of context it can be interpreted either way depending on your prior attitude towards said exercise)

        • anon says:

          How is this a pro-exercise spiel? It sounds like an anti-exercise spiel.

          Probably the good way to think about this is “if you could die right now, or you could get an hour to ride a bicycle in an entirely normal manner unaffected by your impending death and then die, which do you pick?” In my case, I would take riding the bicycle, but I wouldn’t take a few weeks of unpleasant work over a few weeks minus a few hours of not doing unpleasant work.

          • Richard says:

            The full spiel is strongly focused on finding some sort of physical activity that you enjoy because if you enjoy the activity in its own right, you are less likely to stop doing it. (case in point, the thousands, possibly millions of gym membership cards stuffed away in drawers. Lifting iron or running on a treadmill is boooring)
            I use bicycling as an example because I think it’s an awsomecool way to experience the world. The number of people I have personally “converted” is somewhere in the low triple digits, so it seems to work.

        • Max says:

          How about the risk of dying in the accident. AFAIR riding bicycle has a very high death ratio probably among “common activities”. Not quite as high as motorcycle, but way higher than driving a car

    • Seth says:

      But the speaker implicitly presumes there’s positive associated benefits to doing any job. That’s not, let us say, an obviously correct assumption. Indeed, it’s a rather typical moralism, that work _per se_, however demeaning and burdensome, is somehow virtuous in itself. Perhaps this is meant to be a statement of viewpoint of the character? However, it’s hardly universally held.

      That is – “unless cooking is worse than oblivion for you” – Well, am I allowed to say it’s *comparable*? That is, I don’t like cooking, it’s boring wasted time to me, it is basically the equivalent of destroying the associated time of my life. If you do like cooking, if you derive enjoyment from it – great. But someone like the character above comes off to me as one of a long line of annoying preachers, usually directing sermons at the poor who are not sold on the character-building aspect of being servants to the rich.

      • I, too, find the choice between performing chores and spending the equivalent time in oblivion to be hard and non-obvious.

        Alicorn does indeed enjoy cooking very much, from what I’ve heard.

        Whether cooking is worse than oblivion depends on what you are able to multitask at while doing it. If you live with other people who hang about near the kitchen, you can chat while you cook, and the time is better. If you live alone, the time is worth less. Other factors include whether you can listen to music at the same time, and whether the work is complex enough that you must focus completely on it, or simple enough that you can think about other things while you’re doing it. And there’s the question of whether you think your undirected, distracted thoughts are sufficiently entertaining to be worth having.

  4. Ben Finkel says:

    Wait, so according to David Spiegelhalter – RISK PROFESSOR – eating a bacon sandwich is as bad as smoking 4 cigarettes? Or is it non-linear? Still, that’s actually worse than I expected, not better!

    • Megaburst says:

      Yeah I agree. I’m confused because Scott writes

      So overall, if you eat processed meat every day your entire life, you’ll lose about three weeks of life expectancy from colon cancer. That means each serving of meat costs you a minute of your life.

      But the risk professor says eating a bacon sandwich causes you to lose an hour of life. What gives?

      • Jbay says:

        Is it the difference between increased risk of death due to cancer only, and increased risk of death due to all causes?

        I agree this was confusing.

      • Saal says:

        Perhaps a bacon sandwich contains more than one recommended serving of red meat?

        No, no that can’t be it….60 servings of meat in a bacon sandwich?

        • Pete Alex Harris (@ScavengerEthic) says:

          Though that would be one awesome bacon sandwich. You’ve got to die of something, sooner or later. Could do worse.

          • Anthony says:

            I think you missed a decimal point. But *60* (not 0.60) servings of meat in a bacon sandwich would be awesome in a “can I make my jaw unhinge like a snake’s?” kind of way.

      • Nornagest says:

        This confused me at first too. But one’s looking at bowel cancer, the other’s looking at all-cause mortality. The latter’s a lot sketchier, so it’s probably best to treat it as an upper bound.

    • Randy M says:

      Maybe they both rounded differently? I wasn’t exactly following along at home with my own calculations, but remember–many people who smoke smoke about a pack a day, which is, what, 12? 20?
      I’m sure there’s plenty of fast-food lovers who eat a bacon sandwhich for breakfast and a burger for lunch and dinner, but I expect for most the red meat is once a day or a bit less.

      Also, smoking has little good for you (unless you meditate while smoking, I guess) and red meat does have plenty good as well–that, yes, can be obtained other ways, but it’s not like eating meat is in every way self-destructive–it’s quite likely that in moderation, it’s healthy on net for most people.

      • Anthony says:

        Nicotine relaxes people. When California banned smoking in all medical facilities, the staff at the state mental hospitals protested very loudly, predicting impending disaster.

        • Marc Whipple says:

          It seems to be more or less inarguable that quitting smoking is very likely to cause people to gain weight. Depending on how much they smoked versus how much weight they gained, multiplied by the increase in stress (and stress also causes weight gain, hello positive feedback loop) it seems very likely that quitting smoking has caused some statistically significant number of people to die earlier than they would have had they continued to smoke. Not to mention how it alleviates stress for some people independent of weight control. (Doctors used to tell really nervous people to take up smoking. I assure you this was not entirely due to the large sacks of money R.J. Reynolds used to send them every week.)

          I’m reasonably certain that smoking, net, is still a pretty bad thing, but it is not universally negative.

    • Mary says:

      Does that factor in the anti-depressant effect of smoking?

  5. Douglas Knight says:

    That’s a great graph.

    Note that you can run this forwards and backwards. A 50% reduction in mortality adds 7 years to expected life. The opposite, a doubling of mortality, cuts expectancy by 7 years. That’s the number for type 2 diabetes. It’s also a nice round number that’s worth remembering.

  6. NZ says:

    As with all dietary advice, it seems like you should first consult your genetic background. People whose ancestors are mostly from place A probably digest red meat a lot differently than people whose ancestors are mostly from place B.

    Dietary advice given to all humans based on findings from one sample of humans strikes me as advice that’s not worth putting much stock in.

    Also, I was going to write that my great grandfather lived to 104 and ate red meat all the time and that my maternal grandfather and grandmother are both 91 and both still going strong and they also eat red meat all the time. But now I wonder, for those of us with an abundance of such anecdotes in our families, how much should we allow them to influence our diets?

    • Matt H says:

      I think this is doubtful. It’s unlikely there is much selection on improving one’s ability to tolerate something unlikely to cause serious consequences until you’re 70. From an evolutionary stand-point their is no problem.

      • Cerastes says:

        Matt H – It could simply be correlated. Increased digestive or absorptive efficiency might have the immediate benefit of “get more calories and/or amino acids from each meal of protein”, with the positive side effect of “incur less cellular metabolic damage per unit of protein consumed, resulting in lower mutation rates and less eventual chance of cancer”. With cumulative damage, tiny improvements could have big eventual consequences.

        Or it could just be good old genetic drift, with no selective value whatsoever, but still have a functional consequence at age 70.

      • Anonymous says:

        >It’s unlikely there is much selection on improving one’s ability to tolerate something unlikely to cause serious consequences until you’re 70.

        That doesn’t mean different groups can’t process it differently. Or is every group the same once they pass breeding age?

  7. Simon says:

    One thing I feel is also missing from studies like this: who is the person out of one hundred who gets the bowel cancer? This is related to Kahneman System 1-nudges and Scott’s old LW-article on names as primers for city and occupation.

    Kahneman (I’m typing from memory here, all numbers made up) gives the example of a school with a voting booth, and 3% more people voted for extra money for schools than in voting booths not located in schools.
    Scott gave the example that people with a name that sounds like a city are 4% more likely to move there.
    Kahneman has another example of different pictures prompting people to give money to the donations-cup next to the coffee machine, and an aggressive picture leads to 5% more givers.

    But who are these people?

    I think we can safely assume that the people who went to the voting booth and just reallly hate schools didn’t change their mind based on location.
    It seems likely someone named Ernie and got an offer to work in Chicago for MILLIONS OF DOLLARS PER DAY and an offer to work in Indie, Indiana for a dead rabbit and eternal damnation took the Chicago offer.
    It’s not that weird to assume the person who took a coffee from the coffee machine on the day he had 2 bucks for dinner didn’t give to the donation pot, even if the picture was very aggressive that day.

    In the Rationalist community, there’s a lot of talk about how or thoughts are shaped by nudges and primers. And it’s understandable why: it’s a great example of how we’re often not conscious of why we make our decisions. But if 3% more people take a certain choice when offered certain primers, that doesn’t mean we’re all more likely. It might just shift the middle.

    The same goes for this study. That one person who now gets bowel cancer from eating bacon, who is he? Did he already have an inclination towards bowel cancer? Eating a lot of bacon is surely linked to other unhealthy habits, how does that figure in?

    This can of course lead to complete immobility if you take it too far. Well, half of smokers die from smoking-related diseases, but which ones exactly? As long as we don’t know that, let’s not judge etc. And it seems obvious that not eating processed meats in large quantities every day is probably a good idea. But I feel the same thing Scott is critiquing here can also be applied to Rationalist analysis of the ‘gullibility of the Masses’.

    e. Other way to put this: do primers influence 3% of all people, or do they strongly influence 20% of people who don’t have a strong opinion?

    Does eating processed meat give you 1% extra chance for cancer, or 15% extra chance of cancer if you are a typical person with associated habits for those who eats a lot of processed meat?

    • Randy M says:

      This is a good point, but for some issues it is hard to tell if you are one who is susceptible to the marginal influence or one secure in one’s position. In the case of diet, cancers, etc., family history could likely inform one there.
      In the cases where “irrational” factors are influencing one’s decision, if one truely doesn’t care “I can move to city x or y and there is no real reason to select between them” a random deciding factor is as good as any, I suppose.

  8. matt says:

    One other key difference is that we have to eat something. As you mentioned with cookies, other food choices also come with associated risks. We can’t just stop eating.

  9. Matt H says:

    Bacon only really became bad for you in the 1960s. Before then life expectancy was lower than 69 in the US, so no negative effects. Same things for cigarettes. Average age of lung cancer diagnosis is 70.

    I’ve been making this argument for a long time. If we push life expectancy above 120, and find out there is a cancer resulting from 120 years of broccoli consumption then we can declare broccoli bad for you?

    I think as we live longer our knowledge of what it takes to live that long will change our ideas of what a healthy life is. Their might even be trade-offs, like its way healthier up to 70 to be a vegan, but then becomes healthier after 70 to eat meat with every meal.

    It’s a weird world.

    • Vox Imperatoris says:

      Same things for cigarettes. Average age of lung cancer diagnosis is 70.

      Wow, I never thought of it that way! I mean, it’s obvious in hindsight, but it’s a great way to formulate it.

      It explains why “Big Tobacco” and their evil paid shills didn’t find any serious evidence that cigarettes are harmful to your life expectancy until around this time. Because they were right! It wasn’t.

      (Though, as you formulate it, it’s not strictly true. Obviously, cancer due to smoking was one of the things pushing down life expectancy at that time. It just was small in comparison to the other things.)

      • baconbacon says:

        Life expectancy doesn’t work that way. For every child that dies at one you need 68 people to live to 70 to have a LE of 69 (or 6 people to live to their 80s- or whatever). Life expectancy of 69 pobably means 75% of people (or more, just ball parking) live past 69. Median age is probably mid to high 70s.

    • Scott Alexander says:

      Not sure this is quite right – if people were only living to 60, it means the whole mortality curve was shifted to the left, which means that increasing the mortality rate by 20% or whatever would still cost you the same number of years, just at a different time.

      I guess it depends on whether everyone was dying at sixty because of generalized accelerated aging or for some specific reason like the monster who shoots you on your sixtieth birthday. The truth is probably in between.

      • Matt H says:

        Not sure that the mortality curve is a good way to think about it. The current mortality curve is based on data from people alive today. You probably want to do something like if the average diagnosis age is 70, what is the standard deviation? Let’s say it’s 3 years for bowel cancer. If the life expectancy is only 64 2sigma below our average, expect only 5% of the current effect on mortality. Which is only 1/20th as bad in year X than today.

        In terms of what was killing people: In 1900 cancer caused 1/3 as many deaths as tuburcolosis or the Flu, or Pnumonia. We may have slayed most of the monsters, but they were there. By 1950 the rate of cancer near doubled, and it probably had little to do with changes in diet. By 2010 it nearly doubles again reaching the same rate as TB in 1900. It could be the crap in our environment, or it could be that we are just living longer, and more likely to discover cancer.

        • Nornagest says:

          My money’s on living longer. The crap in our environment is a lot more varied and less “organic” (in the colloquial, not the chemical, sense), but there’s a lot less of it. Forget bacon and sausage; in 1900 people heated their homes with flickering coal grates or poorly sealed wood-burning stoves or fireplaces, insulated with asbestos, ate and drank from lead-based pewter, took quack cures involving radium and cocaine. Victorian London suffered smog that made Beijing look like Yosemite. And the idea of occupational safety was pretty much nonexistent; big chunks of the woods around my hometown are off limits because they’re contaminated by the mercury and cyanide used in turn-of-the-century gold extraction.

          And, of course, people living in 1900 were exposed to a lot more literal crap.

      • Richard says:

        A lot of things that causes cancer is dose and time dependant. Example:

        Around here, the national allowed radiation dose for a regular certified worker is 20mSV/year, the allowed dose for a certified worker over 60 is 200mSV/year because at that dose it will take somewhat over 30 years to develop cancer.

        I would expect tobacco and other things to work in a similar fashion which is why I am utterly unconcerned about cigars at my age, given that I’ve not been a lifelong smoker.

    • NN says:

      Is that average life expectancy and, if so, how much does infant/childhood mortality figure into it? I ask because it’s common to read things like “the average lifespan in Medieval Europe was about 35.” While this is technically true, this was largely caused by enormous early childhood mortality rates, and in fact anyone who survived until adulthood back then had a pretty good chance of living into their 50s, 60s, or maybe even 70s. I want to make sure that a similar bias isn’t at work here.

      • Nornagest says:

        According to this table life expectancy at age 10 in the mid-1800s was about 50 additional years, for a total age of 60. If memory serves it was about the same for the medieval European aristocracy, but would have been a bit lower for the peasantry (and a lot lower for everyone during the Black Death).

        By 1961, it had reached 60 (for an age at death of 70). By that time infant mortality seems to be making a relatively small contribution; the gap between life expectancy at birth and life expectancy + age at 10 is only 3 years, compared to 20 years in 1850.

      • RCF says:

        And if life begins at conception, then life expectancy today might be as low as five years!

  10. keranih says:

    For me, the major issue is the substitution problem.

    The “13% increase” in the BBC article comes from this study,, and the abstract there says

    We estimated that substitutions of 1 serving per day of other foods (including fish, poultry, nuts, legumes, low-fat dairy, and whole grains) for 1 serving per day of red meat were associated with a 7% to 19% lower mortality risk.

    If you could figure out a way for even half of the population to replace their bacon with one of those things (and oh, btw, we’re not so sure about the lowfat dairy anymore either) instead of substituting a cheese danish or a double fat cappuccino and a chocolate donut, then heck, we’d be golden.

    But as the experience with disapproving fats has shown, we aren’t going to drop bacon for kale. At best, we’ll eat more grits.

  11. walpolo says:

    Do the same risks apply to cold-cut chicken and turkey? I eat almost no bacon, but plenty of lunch meat.

    • Sokka says:

      The bacon-related risks may, if the meat is cured with nitrates (as far as I’m aware, that’s one of the big concerns with these “processed meats”. That includes “nitrate-free” lunchmeats that instead include “cultured celery extract”, which is a fancy way of saying “nitrates”.) However, there’s also evidence that antioxidant substances in the stomach at the same time as nitrates reduces their absorption into the body, so if your sandwich has a nice amount of bright fruits and/or vegetables on it, you may not have to worry so much. The “bacon sandwich” in the article is misleading, since I imagine most people eat their bacon with eggs and toast, not a dietary serving of vegetables, but the things you eat alongside a food do change its nutritional profile greatly, which studies like this do not take into account.

      You can solve and/or reduce a lot of problems by having a healthy, balanced and varied diet, regardless of its individual contents.

  12. Jeremy says:

    I would be a little more careful of using “hours of life” as a currency for health tradeoffs. For example, would you rather die now, or feel moderately sick for another year? I think most people would prefer to live longer for some moderate level of discomfort. But now if we said, “eating bacon causes you to feel moderately sick for an hour”, suddenly the tradeoff doesn’t seem so good. It’s not such a good idea to convert everything into “Quality Adjusted Hours of Life” because that’s not how people actually reason about these things. People can be utility pumped by hypotheticals like this, so all you’re going to prove is which direction you ran the pump in.

    It’s not that I think that the tradeoffs are a slam-dunk in either direction, but I do think that if one of the side effects of eating a bacon sandwich is “feel sick for an hour years later”, people would definitely consume significantly less meat. It’s only because the consequences are pushed towards the end of your life that framing it as QAHoLs seems to take the kick out of it.

    Furthermore, you’re not mentioning that there are plenty of processed meat substitutes, and they certainly provide some percentage of the enjoyment of eating bacon, without the risk. I can say as a “born and raised” vegetarian, that if you’ve never eaten meat, you don’t develop a taste for it. So it seems like you could probably raise your kids on a none-low processed-meat diet and get all the benefits without most of the downside.

    (Disclaimer: I am a vegetarian for ethical/status quo reasons, so obviously the risk of motivated reasoning, although I did assume the hullabaloo about “meat causes cancer” was stupid at first, because I assumed the lifetime risk would be like 0.001 percent, not 0.5 percent.)

    (Note: Also, I recognize that “eating something -> makes you feel sick ->AVOID THAT THING” is a strongly evolutionarily reinforced pattern, and so framing cancer that way may be unfairly priming you ( However, the “feeling sick” effect here is not a false positive, but really /is/ the food causing disease. I mean, in general, the aversion to being sick and dying early is evolutionarily reinforced. So maybe humans just value being healthy more than is proportionate to how much being healthy improves their life. But at this point our model of a human as a rational agent starts to break down.)

    • Tracy W says:

      Feeling sick 12 hours later is far from universally effective at putting people off alcohol.

      • Jeremy says:

        Wouldn’t you say that people drink less and less frequently than they would if hangovers didn’t exist? I’m not claiming that it would stop people from eating processed meat, I’m saying they’d think twice about it, just as most people think twice about getting drunk every day.

    • Deiseach says:

      If feeling sick put you off indulging in things that are bad for you, nobody would ever drink again after their first hangover 🙂

      • Urstoff says:

        I can’t really stomach the smell of liquor anymore from drinking heavily one too many times in college. I don’t know why this doesn’t happen to more people.

        Beer and wine are still good, though.

      • Jeremy says:

        I think it does put a lot of people off drinking excessively? If you always got a hangover regardless of how moderately you drank, I probably wouldn’t drink at all.

      • Nornagest says:

        I have a theory that every drinker has at least one species of hard liquor that they can’t stomach after an episode of overindulgence early in their drinking career.

        Tequila seems to be a common one, at least here in California. Mine is cheap spiced rum, like Captain Morgan’s.

        • ivvenalis says:

          Yeah, I have a fondness for good tequila, but nobody will ever drink any of my stuff because of their experiences with Jose Cuervo.

          Mine is probably soju, but to be fair after some suspicious reactions to it, I drank about 130ml one night and then nothing but water as an experiment, and was still mildly hung over/under the weather the next morning (that’s less than two “glasses”).

          I doubt anyone young enough for this to be a problem cares, but overindulging on bad booze in your youth probably significantly damages your long-term hedonic utility. As I get older, more and more people either have adverse conditioned reactions as discussed, or they’ve had to swear it off altogether. Meanwhile I derive quite a bit of pleasure from a dram of whatever every couple of days. I’m not limited to what a college student can afford either.

  13. alexp says:

    “This teaches us something important: “risk professor” is an awesome job title and “David Spiegelhalter, Risk Professor” ought to be a BBC television show starring Harrison Ford. ”

    Never tell a Corellian the odds!

  14. Thursday says:

    Honestly, the common sense here is common sense: don’t eat processed meat every day, but it’s fine to have a bit of it now and then.

    • Sokka says:

      I’m finding this, too. I’m quite the carnivore and even I can rarely relate to these studies about the doom and gloom of meat. Really, a serving of red meat per day? There’s no way I can afford that. Much of my meat consumption comes from buying a chicken or two at the beginning of the week, carving it up into parts to freeze, and turning the remaining carcass into stock. I buy pork when I can get the free-range ethical hippie stuff, and a pack of sausage to make chickpea-and-sausage mash, and if I’m lucky I’ll have the money for a beef roast or a pair of steaks for my boyfriend and myself.

      Though, admittedly, beef is far more expensive here, relatively speaking, because we have more restrictions on it with regards to hormones and antibiotics. I suddenly realize this may also have a confounding effect.

  15. Cerastes says:

    The flip size of effect size is dose size. This study prompted a blowup on a friend’s FB, and, upon reading the study in detail, I found that the “high risk group” in the bacon study was eating bacon/smoked meat *every day*. Another study, quoted by a vegan in that thread, included people eating multiple servings of red meat per day at one end of the scale (to their credit, they did show a dose-response curve).

    I’m about as close to purely carnivorous as my wife will allow, and I’m still wondering who the hell eats like that? Some of it is that I’m an impoverished post-doc, but even if I wasn’t, I cannot imagine having time for a bacon & sausage breakfast every morning before work. And as much as I love burgers and steak, I can’t fathom having it for every meal – I like chicken, turkey, fish, shrimp, etc. too.

    • Scott Alexander says:

      A lot of Irish people have bacon every day.

      And I used to have bacon every day when I had to be at work really early and eat at the cafeteria where there was a limited selection of good breakfast food.

      • Cerastes says:

        Really? Wow, now I feel like my life is positively spartan.

        Still, I wonder if they’d be better off using a dietary condition that resembles the mean consumption levels of the target population.

      • keranih says:

        A “traditional” breakfast in the South includes fried/scrambled eggs, fried bacon (or sausage), toast (or biscuits) with butter and jam, and grits. At least one of the later in the day meals would include a pork entree, of which probably 50% was ham or sausage.

    • whateverfor says:

      These probably aren’t people eating bacon every morning, but people eating bacon a couple times a week, ham sandwiches for lunch a couple times a week, and sausages for dinner fairly often. It’s not hard for that to add up.

  16. ButYouDisagree says:

    It’s easily to be sloppy when talking about averages and expectations.

    Multiply out and an 0.4% chance of losing 10 years means that you lose on average two weeks.

    That means each serving of meat costs you a minute of your life.

    It doesn’t make sense to talk about averages for a particular person. Each person has one actual life, which either includes GI cancer or doesn’t. We might say that, in a large group of people, the average life lost is 2 weeks per person. We might say that, for a particular person, we expect them to lose 2 weeks of life. But this uses a technical sense of expectation. The real decision facing people isn’t 2 weeks of extra life vs. meat, it’s a small chance to avoid a painful and seriously life-shortening disease vs. meat.

    This is important because people’s preferences about length of life don’t follow expected value in the way your post suggests. People have preferences about risk, often in complicated ways (see e.g. the Allais Paradox). Utility may not be constant throughout your life. And people have preferences about the shape of their lives, e.g. a bad event may be particularly bad because it occurs at the very end of life.

    • baconbacon says:

      People only have one life, but they have tons of choices in that life. How many days on average will driving take off your life? Social drinking? Roller coasters? Sitting at work and not standing? Talking about expected loss is the only sensible way to go about it.

      • Jacob says:

        If you had to pick one strategy to apply to all decisions I think that would be true. But why do we need to use the same strategy? When gambling small amounts of money I use expected value. But I still buy insurance even though the expected value is negative (I would do so even without it being legally mandated).

    • Jacob says:

      Came here to say this. It seems like people use more of a minimax approach.

      Consider a devil offering the following bet: Roll a million-sided die. If it comes up 2 or greater, the devil pays you $10. If it comes up 1, you pay the devil $1,000,000. If you don’t have that, the devil garnishes your wages until paid, charging interest exactly keeping up with inflation.

      This bet has positive expected value, about $9.00. But is the near certainty of winning a small sum worth even the tiny risk of such a catastrophic loss? (edit: I wrote this before reading the Allais paradox; an economist might consider my formulation equivalent but a normal person might not)

      Maybe this is just a question of utility being non-linear in money; but it could be that utility is non-linear in quantity of life lost as well.

      • Peter says:

        Don’t gamble with the devil, it never rarely ends well.

        (Or, to put it another way, the “gamble with the devil” framing suggests that we can’t do the usual “reduce it to a maths problem and solve that” trick.)

        • Marc Whipple says:

          Gambling with the Adversary includes, in most instances, at least one outcome where your immortal soul suffers terribly for all eternity. Mathematically, it’s hard to make that a rational choice under any circumstances.

          Hard. Not impossible. For instance, for Billy Markham, raiding Hell to save three souls from eternal torment was a rational mathematical choice even though it had a nonzero risk of causing his own soul to suffer eternal torment. (The other three souls were already in Hell. The possible net changes were one more soul in Hell or three less souls in Hell.)

    • Tom Womack says:

      Bowel cancer statistics are odd, because it’s a disease for which the treatment if caught early is reasonably effective if unpleasant, but is often caught untreatably late; 75% survive one year from diagnosis, 75% of those survive four more years, 90% of those survive five more years)

      I think that the delicious sausages I was eating last night corresponded to about a one-in-thirty-million additional chance of needing my bowel resected at age eighty.

  17. Gary Jones says:

    “What makes sausage-like meat more dangerous are the chemicals released in processing them, including nitrates and nitrites, the cancer agency said.”

    These are preservatives, and are not needed except to extend shelf life and buffer possible poor handling by shippers and retailers.

    It is possible to buy sausage-like meats that do not use preservatives. I sell them, and I’m not unique. Nearly every producer of “natural” meats – grass fed, yada yada – has such products. They are either very fresh or frozen, which eliminates need for preservatives.

    Preservatives are ancient methods, but more developed societies more often use refrigeration to store foods. Even vegetables are flash frozen in the field moments after harvest to retain as much nutritional virtue as possible. Given that something like 40% of world food never even makes it to market due to spoilage it seems like the WHO would do well to nudge in the direction of refrigeration.

    The supposed harms from red meat are actually harms from high heat cooking. When temperatures rise above 400 F some mildly carcinogenic compounds are produced, along with perhaps a zillion yummy flavor compounds (the Maillard reaction). There are also some carcinogens that can be introduced to the meat by the burning fuel that provides the heat.

    You can avoid the fumes from burning fuel in several ways, and you can even avoid the Maillard reaction, though at some considerable aesthetic price. Note that the same thing happens when other meats are roasted too. It isn’t just red meat.

    There are also things one can do to inhibit the formation of HCAs when cooking. The use, or misuse, of marinades, oils, carbohydrates (sugars) etc. can raise or lower production. It’s complicated.

    • walpolo says:

      Aren’t there much greater amounts of nitrates in leafy greens than there are in processed meats?

      • Gary Jones says:

        Over 90% of the nitrates consumed, on average, come from leafy greens. The amounts vary depending on the leaf type, its age, and growing conditions including soil fertility and even temperature.

        Vegetable powders such as celery are sometimes used in sausage-like meats instead of nitrates/nitrites precisely because they are so high in nitrates. It sneaks the nitrates into the product with a green halo.

        It is also claimed that nitrites are beneficial for immune and cardiovascular function; they are being studied as a potential treatment for hypertension, heart attacks, sickle cell and circulatory disorders. Nitrates are metabolized to nitrites, so eat your veggies.

        It all seems pretty muddled doesn’t it?

        Recall that this is a post-rational issue in a post-rational age. It’s all about eye balls and attention, marketing, rather than some sort of rational or objective analysis.

        I can’t usefully educate my customers about the issue. I can’t refute false claims and expect that to undo the intellectual damage done by the false claim. What I can do is to produce foods with no added nitrates/nitrites. The virtuous word there is “added”.

        • Winter Shaker says:

          It’s all about eye balls

          That’s quite disturbing in a comment about the contents of sausage meat 😛

    • Arthur Bludgeford says:

      What actually are the nitrites used to preserve food? Is the problem caused by the nitrite group itself? Because I thought that ammonium nitrite was commonly used as fertilizer and occasionally as a heart medication.

      • Marcel Müller says:

        Mostly sodium, yes, the nitrate or rather nitrite (which can be produced from nitrate by certain bacteria or reductive conditions) is the problem, mostly because under high heat or catalysis by certain enzymes they react with amino acids to form nirtosamines, which in turn are carcinogenic.

      • Gary Jones says:

        Sodium Nitrite, Sodium Nitrate, and Potassium Nitrate (deprecated).

        Ammonium Nitrate is fertilizer, though there are many other nitrogenous compounds used to supply nitrogen in plant available forms. They can’t just breathe nitrogen out of air like they do CO2.

        Amyl nitrite is a heart medication, a vasodilator. Also used as a recreational drug.

  18. mobile says:

    People who don’t eat bacon don’t really live longer lives; their lives just seem longer.

  19. US says:

    I can’t help myself from adding a few quotes about risk communication and health information here by a guy who might also be called a Risk Professor (Gerd Gigerenzer):

    “Over the last decade, I have taught about 1,000 physicians in their continuing medical education (CME) in risk literacy. These sessions covered natural frequencies versus conditional probabilities, absolute versus relative risks, frequency versus single-event statements, and mortality rates versus five-year survival rates. In each of these pairs, the first representation fosters insight, whereas the second supports innumeracy […] Unclear presentation of medical statistics such as the risks associated with a particular intervention can lead to patients making poor decisions about treatment. […] Good communication of risk requires a clear statement of what a probability refers to. Although necessary, this step alone is not sufficient, given that some patients misinterpret risks even when a reference class is given […] Frequency statements can help reduce potential confusion because they always refer to a class and are easily understood.”

    “Many doctors, patients, journalists, and politicians alike do not understand what health statistics mean, or they draw wrong conclusions without noticing. The causes of statistical illiteracy should not be attributed to cognitive biases alone […] We argue that [the] causes [of statistical illiteracy] are not simply inside the minds of patients and physicians — such as the lack of a math gene or a tendency to make hard-wired cognitive biases. Rather, we show that statistical literacy is largely a function of the outside world and that it can be fostered by education and, even more simply, by representing numbers in ways that are transparent for the human mind.”

    Some of the examples included in the book in my opinion really add some perspective in terms of how to think about a topic such as the bacon/cancer link (…and how to think about what ‘people in general’ are likely to think about that link):

    “Screening is intended to detect existing cancers at an early stage. So it does not reduce the risk of getting breast cancer; it increases the number of positive diagnoses. Nevertheless, 57%, 65%, 69%, and 81% of […] random sample[s] of women in the United States, Switzerland, the United Kingdom, and Italy, respectively, believed that screening reduces or prevents the risk of getting breast cancer (Domenighetti et al., 2003). An equally astounding 75% of a representative sample of German women who participated in mammography screening wrongly believed that screening reduces the risk of developing breast cancer […] Patients at Auckland Hospital, New Zealand, were asked: “What do you feel is the likelihood of you having a heart attack over the next 12 months?” This likelihood depends on individual risk factors, such as age, sex, a previous cardiac event, a family history of coronary heart disease, diabetes, smoking, and other known factors. Yet patients’ risk estimates showed no correlation with any of these factors (Broadbent et al., 2006) […] In a study in Switzerland, people were shown to lack even minimum medical knowledge of the risk factors for stroke, heart attack, chronic obstructive pulmonary disease, and HIV/AIDS. No participant was able to answer all questions correctly — on average, they got only one third right. The number correct was only moderately higher for people with personal illness experience (Bachmann et al., 2007). […] A review of experimental studies showed that many patients do not understand the difference between relative and absolute risk reduction and that they evaluate a treatment alternative more favorably if benefits are expressed in terms of relative risk reduction (Covey, 2007).
    In summary, the available studies indicate that very few patients have skills that correspond to minimum statistical literacy in health […]. Many seek certainty in tests or treatments, benefits of screening are wildly overestimated and harms comparatively unknown, early detection is confused with prevention, and basic health statistics such as the differences between sensitivity and specificity and between absolute and relative risks are not understood.”

    The quotes are from the book Simply Rational, which is a great resource on these and related topics. On the basis of these observations it does not seem crazy to assume for example that some of the people who’ll stop eating bacon because of this information will also assume that they’ll have by doing that eliminated their risk of getting bowel cancer.

    One thing I find slightly amusing is that despite the fact that some people will probably conclude from the reported risk that if they eat processed meat every day they’ll face an almost one-in-five risk of getting cancer (I’m sure that’s how some people will interpret this statement: “The IARC’s experts concluded that each 50-gram (1.8-ounce) portion of processed meat eaten daily increased the risk of colorectal cancer by 18%”), these people will probably still be unlikely to change their habits and cut down on the bacon; there’s a related psychology of lifestyle literature indicating that there is in e.g. the context of diet and exercise at best only a weak link between people’s knowledge about risk factors and their actual behaviours. For example: “Despite considerable efforts over a number of years, there is limited evidence to suggest that educational approaches to dietary change (that is providing basic information about what constitutes a ‘healthy’ diet) alter children’s eating habits” […] “Many early drinking interventions are educational in nature. In essence these are risk communication messages and the evidence from psychological research is that improving risk perceptions will have little impact on levels of drinking. Unsurprisingly then, there is little evidence that alcohol education and health promotion have any positive effect on drinking habits in Britain […] These campaigns are heard and understood because knowledge increases in targeted populations […] so it is not that the message is failing to reach the designated audience, rather the message has no impact on behaviour.” […] “Extensive evaluation of social cognition models’ ability to predict uptake of physical activity leads to the conclusion that a perception of the risks of non-activity and the benefits of activity for health has at best a small impact of overall variation in physical activity behaviour.” (Quotes from the book The Psychology of Lifestyle: Promoting Healthy Behaviour, by Thirlaway & Upton).

    • Jiro says:

      Nevertheless, 57%, 65%, 69%, and 81% of […] random sample[s] of women in the United States, Switzerland, the United Kingdom, and Italy, respectively, believed that screening reduces or prevents the risk of getting breast cancer (Domenighetti et al., 2003).

      More likely most people didn’t act like a stereotypical literal Internet geek, and responded like that because screening reduces the risk of *dying from* breast cancer.

      • US says:

        The responses are still wrong, and to me it seems important for someone to at least be able to distinguish between the risk of getting cancer and the risk of dying from cancer. Participation in breast cancer screening also increases your risk of getting diagnosed with cancer, but that’s not (necessarily) a strong argument against screening..

      • Sokka says:

        I’m a stereotypical literal internet geek who’s well aware the statement is false and I still thought it was true simply because I misread it the first time around. Your brain is very good at deciding what a statement is meant to mean even if you think you read it properly. It’s a “toasters toast toast” thing.

    • Urstoff says:

      Is Gigerenzer incredibly underrated among the rationalist community, or does it just seem that way?

      • Peter says:

        I think there’s the issue of the Gigerenzer/Kahnemann dispute[1], and lots of people seem keen on the Kahnemann side. Also there’s the fact that Gigerenzer really doesn’t like single-event probabilities, which gives problems for the sorts of Bayesianism popular in the LW-exodus rationalist community. He has a certain “lots of stuff is rational if you think about it the right way” vibe which jars with the (OK, I’m going to be uncharitable) “Oh noes, cognitive biases! Irrationality!” vibe on LW and elsewhere.

        [1] For a microcosm, look at all the stuff about Linda the possibly feminist possibly bank-teller.

  20. anon says:

    1% chance of losing 10 years of life is not equivalent to losing 1/10 years.

    Also, lifespan being equal, living in poor health is not equivalent to living in good health.

    • Jeremy says:

      That’s a good point, I didn’t think very much about the marginal utility of years, but I would definitely choose an X% shorter life over an X% of death any day. Likewise, I would choose to live X% of my life sick over an X% chance of being sick for the rest of my life.

  21. Sam says:

    I admire your rhetoric. An excellently understated allusion to an excellent quote, the first part of which I did not know:

  22. Linch says:

    Do people actually like bacon more than other forms of meat? It always felt really bland and tasteless to me. Heck, if you tell me you have this vegetarian substitute of bacon that tastes *exactly the same*, I probably wouldn’t eat it of my free will unless it’s free or you pay me to.

    • The Anonymouse says:

      You’re weird, but we love you anyway.

      • Linch says:

        Assuming that nobody else is around to hear you talk about it, do you prefer bacon to eg, pork chops or steak or salmon or fried chicken or sushi?

        My theory for the extreme popularity of bacon is that it’s the same as the distaste for “moist”: People like/dislike it because they’re socially expected to.

        The problem with that theory is that it’s very close to mind-reading: since I have no way to gauge internal mental states other than statements and behavioral cues, it seems epistemically and morally problematic to not take what people say at face value.

        • jaimeastorga2000 says:

          Assuming that nobody else is around to hear you talk about it, do you prefer bacon to eg, pork chops or steak or salmon or fried chicken or sushi?

          I don’t eat sushi or salmon, and I like bacon about as much as I do fried chicken and pork chops, all of which I prefer to steak.

          My theory for the extreme popularity of bacon is that it’s the same as the distaste for “moist”: People like/dislike it because they’re socially expected to.

          I liked bacon long before I knew about its canonical role as one of the most delicious foods in American (internet?) culture.

        • I don’t need social pressure to tell me that bacon is delicious. The preference for bacon is not socially manufactured. And I’m a little perplexed as to how you can find it bland. Are you eating the right kind of bacon?

          That said, it is a socially mediated fact that Internet loves bacon, and bacon is the canonical example of “delicious meat”.

        • The Anonymouse says:

          Assuming that nobody else is around to hear you talk about it, do you prefer bacon to eg, pork chops or steak or salmon or fried chicken or sushi?

          Good bacon or a nice medium-rare steak I would like equally (and in preference to any other protein). This is only complicated by the fact that, for me, they occupy different taste-spaces; that is, when I want a steak, a pork chop will do, but bacon will not, and when I want bacon, steak won’t do.

          Either of those I will take in preference to salmon, sushi, or fried chicken, in that order. Were I to name a meat that is, to me, as bland and tasteless as bacon is to you, it would be the ubiquitous boneless/skinless chicken breast.

        • sweeneyrod says:

          As a vegetarian, the smell of bacon appeals to me much more than that of any other meat.

          • Linch says:

            As somebody who made the switch to reduce eggs/meat fairly recently (but not dogmatically), I miss fried chicken (and chicken in general, though not chicken breasts or nuggets, which are ridiculously bland) A LOT more than basically any other type of Western cooking. Quitting pork was ridiculously easy for me emotionally, the only issue was socially sometimes it gets a little awkward (On the other hand questions like “Why do you eat chicken but not pork?” might have helped push other people to be more considerate in their own ethical choices, hopefully). On the other hand, the substitute good there was probably chicken around 60% of the time, so now I have a coupla more sad birds on my conscience. Still, I do have cravings for chicken now in a way that I never have (or had) for any other meat.

            There are certain Chinese dishes involving pork that I enjoyed immensely and in a strict-ordering sense I would rather eat them than fried chicken or stewed drumsticks etc. But I don’t have any particular desire to seek them out, nor do I consider it a loss to never eat them again.

            Although “craving” in general might be an exaggeration; the compulsion to eat has never been nearly as strong for me as the compulsion to surf the internet, videogames or finishing (though almost never starting) a good novel.

        • John Schilling says:

          Assuming that nobody else is around to hear you talk about it, do you prefer bacon to eg, pork chops or steak or salmon or fried chicken or sushi?

          That’s like asking if you prefer salsa to pizza.

          Pork chops, steak, salmon, fried chicken, and sushi are main-course dishes. If you’re eating any of those, they probably constitute the majority of the food on your plate. Or at least the meat plus a neutral starch; steak and potatoes, salmon and rice, burger including bun, etc.

          Almost nobody makes a meal of just bacon, or even mostly-bacon. Bacon is used in small quantities to enhance other foods, mostly by way of its smell and its fat content. If the main course is deficient in those qualities but otherwise desirable, bacon is indeed a most wonderful addition. If you’re offering me a plate of just bacon, that’s like offering me a bowl of salsa and a spoon, and I’m going to rank it very low on my preference list.

          Anonymous’s comment of a “meat-based condiment” is about right.

          • Nornagest says:

            Well, there’s bacon and eggs. It’s usually not going to be the only thing on your plate there — I’ve made just bacon, but mainly while camping — but it’s not being used as a condiment either.

    • Anonymous says:

      Bacon can enhance a sandwich or hamburger but it’s more a sort of meat-based condiment than a main meal in itself imo.

      • Chris Conner says:

        In the semantic hierarchy of sandwich nomenclature, bacon is a vegetable.

        Consider a sandwich made with bread, mayonnaise, lettuce, tomato, cheese, and turkey. It is a turkey sandwich, because the sandwich is named after its most salient ingredient. Meats stand at the top of the semantic hierarchy for sandwiches.

        Take away the turkey, and it’s a cheese sandwich, because now cheese is the most salient ingredient.

        Take away the cheese, and it’s a lettuce and tomato sandwich.

        Take away the lettuce and tomato, and it’s a mayonnaise sandwich. Thus condiments rank lowest in sandwich semantics.

        Put the tomato and lettuce back, and add bacon. If bacon had the same salience as turkey, or any other meat, you would call it a bacon sandwich and the vegetables would not merit a mention. But you don’t — you call it a BLT. Bacon ranks with the tomato and lettuce. Bacon is a vegetable.

        • The Anonymouse says:

          Clever, but not quite true.

          Consider the BLT. Sure, there’s equal billing in the name, but imagine a person eating one: were the lettuce to suddenly disappear, that person would likely continue eating the sandwich. Were the bacon to disappear, I suspect the sandwich would be thrown away.

          • Chris Conner says:

            I’m making a semantic argument. Your scenario goes to taste, and what you would eat versus what you would throw away does not affect nomenclature. You might eat a hot dog with mustard, but throw away a hot dog with ketchup. Either way, you call it a hot dog.

        • Deiseach says:

          Not quite; there are ham salad/chicken salad/turkey/cheese salad sandwiches because they include the lettuce, tomato and mayonnaise with the meat. They are so named because they are distinct from the ordinary ham/chicken/cheese sandwich.

          You have ham and cheese sandwiches. You can have salad sandwiches with no meat or cheese included but as many types of vegetable as you want. You can have jam sandwiches and you can even have banana sandwiches. (I don’t know if any brave soul has yet tried to have a jam and bacon sandwich).

          You have the bacon sandwich, and you have the BLT which needs a different name because it includes lettuce and tomato and is not bacon on its own.

          Bacon is not a vegetable but bacon can be accompanied by vegetables in a sandwich. Consider the cheeseburger: who would say that the cheese is the main or major constituent of the burger, or that the meat is now a vegetable because the name of the comestible has changed? It is so named to distinguish it from an ordinary hamburger.

          • Peter says:

            Cheeseburger – it started life as a regional delicacy from the little-known town of Cheeseburg.

            Or maybe not. Burger nomenclature is a funny thing. A just-plain burger is a hamburger – i.e. a beefburger. If you wanted a burger made of ham, or with ham on top, good luck with naming that. A chicken burger or a bean burger is a burger made of chicken or beans instead of beef; a bacon burger is a burger with additional bacon. Furthermore “burger” can refer to either just the beef patty or the whole sandwich[1]. A cheeseburger is a beefburger with cheese, however a halloumi burger has halloumi cheese instead of beef.

            There you go – we never say a “lettuce burger” or an “onion burger” or “gherkin burger”, bacon, like cheese, merits a mention in your burger name.

            I think the BLT issue is that a “bacon sandwich” generally refers to just bread, butter/margarine/other spread, and bacon, so if you want garnish you have to ask for a BLT, whereas a burger comes with variable garnish sort-of by default, but if I make a burger at home then I don’t call it a “burger no salad no mayo no relish” I call it a burger.

            I think the problem is not that there are no patterns in how comestibles are named, but that there are too many.

            [1] It feels really odd calling it a sandwich.

          • Peter says:

            Oh, and people who want to see a really silly bacon sandwich should see

          • Chris Conner says:

            The semantic properties of burgers are irrelevant to sandwiches, because a hamburger is not a sandwich. So while the cheeseburger phenomenon is an interesting subject, I’m going to disregard it.

            In the case of salad sandwiches, I believe the salad is handled as semantically distinct from the base ingredient and it seems to me that they rank with meats. Make an egg salad with egg, mayo, mustard, and celery. Put it between slices of bread and you have an egg salad sandwich. Add a slice of cheese and it’s still an egg salad sandwich. Remove the egg salad and replace it with mayo, mustard, and celery, and it’s a cheese sandwich. Add a sliced hard boiled egg and you have either a cheese sandwich with egg or an egg and cheese sandwich, but either way the egg and the cheese are nearly equal in salience. So despite being made of exactly the same things, egg salad is more salient than its ingredients are separately. The two cases are semantically distinct.

            (Eggs are weird. If you put a sliced hard boiled egg in a cheese sandwich, then the egg and cheese have nearly equal salience and they appear to rank the same in the hierarchy. But if you fry the egg, then I think most people would call that a fried egg sandwich. A fried egg is more like meat and a hard boiled egg is more like cheese. Weird.)

            The ham and cheese sandwich is a bit of an oddity. The meat should outrank the cheese and thus make it not worth mentioning, as happens with other meats, but it doesn’t. Some idiosyncrasy related to the history of this particular sandwich? I don’t know.

            Peanut butter, jelly, jam, and bananas are all at the same rank. They’re also not things you normally mix with the fixings that make up, say, a roast beef sandwich. If you try mixing them with bacon, or lettuce, or pastrami, then the novelty of mixing makes EVERYTHING salient, and it all has to be named to describe the sandwich.

            “Salad sandwich,” isn’t a term that I recognize. I think in the US that would just be called a vegetable sandwich, though there could be more variation than I’m aware of. Using either terminology, the name functions to group together the salient ingredients, which are salient because there’s nothing higher-ranking around.

          • Peter says:

            I think this debate might be complicated by different varieties of language. I’m from the UK, Deiseach is from Ireland, for both of us salad sandwiches are totally a thing, in fact I didn’t know they weren’t ubiquitous. Although I don’t remember seeing them in shops or cafes or whatever – they seem to be a thing you make at home with whatever’s on hand. As for burgers – I think in plain everyday English they aren’t sandwiches, but in the trade, or in some parts of the trade, they are. Possibly in obscure legalese. Certainly on wikipedia.

            (If you cut a bread roll in half, and stick butter and cheese in it, I wouldn’t call that a sandwich, it’s a roll, and that’s at least as sandwichy as a burger. Weird.)

            I think the problem is that there are strange exceptions and it’s difficult to tell which things are examples of rules and which are irregular. Also I think there’s a difference between bacon having the same rank as a vegetable and bacon being a vegetable.

            Also, cheese and bacon sandwiches, that google assures me are a thing. If cheese outranked vegetables and bacon ranks as vegetable then it should just be a cheese sandwich. Also, bacon, egg and cheese.

            I’m also not sure about this “you don’t have to mention the cheese” thing. But possibly I’ve been veggie for too long and my knowledge of meaty sandwiches has atrophied.

            (Seeing as this is a culinary matter, has anyone mentioned the hysterical raisins yet?)

          • Chris Conner says:


            Certainly a sandwich made with whatever vegetables you find on hand is a thing in the US; we just don’t call it a salad sandwich. At least, not in my part of the US. In the northeast they have quite a well-developed vocabulary for describing all kinds of sandwiches, so who knows.

            I think that the idea that hamburgers are sandwiches stems from the fact that if you attempt to naively define a sandwich, a hamburger meets that definition. The lesson here is not to rely on naive definitions. Are hamburgers what people mean then they talk about sandwiches?

            Chris: Hey, I’m going down to the cafeteria. You want anything?

            David: Sure. Could you get me a sandwich?

            Chris: What kind?

            David: Whatever, just any sandwich.

            If I return with a hamburger, hot dog, burrito, pirozhki, gyro, or two slices of pizza stuck together face to face, you can legitimately say that I didn’t bring you what you asked for. If I return with sliced beef and lettuce held between two pieces of bread that have been spread with mustard, then I have done what you’ve asked. If I return with a hero, po’boy, or Reuben, then I have done what you’ve asked. If I return with ghost peppers and peanut butter held between two halves of a pumpernickel bagel, then I’ve done what you’ve asked but I’ve been an asshole about it. What people mean when they say “any kind of sandwich” includes a lot, but it doesn’t include hamburgers.

            Also I think there’s a difference between bacon having the same rank as a vegetable and bacon being a vegetable.

            What, you don’t think you can change one thing into another just by manipulating semantic associations? I do that all the time. Is it just me? Maybe it’s my superpower.

            Also, cheese and bacon sandwiches, that google assures me are a thing. If cheese outranked vegetables and bacon ranks as vegetable then it should just be a cheese sandwich. Also, bacon, egg and cheese.

            I’m also not sure about this “you don’t have to mention the cheese” thing. But possibly I’ve been veggie for too long and my knowledge of meaty sandwiches has atrophied.

            Yeah, I think bacon is more complicated that I previously realized. How’s this: if I say “turkey sandwich,” the implication is that there is no other meat on the sandwich (or much less of it than there is turkey), and ordinary sandwich fixing that occur with turkey might or might not be present. That would include cheese, lettuce, tomatoes, condiments, and so on. If you ask for a turkey sandwich without being more specific, then you can’t complain about the presence or absence of any one of those fixings in particular. The same applies to other meats and to things like egg salad. Ham is an exception — “ham sandwich” seems distinct from “ham and cheese sandwich” — and I don’t know why.

            “Cheese sandwich” carries the implication that there is no meat and that other fixings that ordinarily occur with cheese might nor not be there. If there is both meat and cheese on the sandwich, you can justifiably expect me to mention the meat, and object if there is meat on your “cheese sandwich.” Similarly for “cucumber sandwich” — the implication is that there is no cheese, and if there is someone ought to have mentioned it.

            Bacon, on the other hand, doesn’t seem to carry any particular implications on its own, because it occurs in too many distinct contexts. Bacon and cheese sandwich, BLT, peanut butter, banana, and bacon sandwich, club. Its versatility means that it has no typical role, so it mentioning it is generally mandatory. Does that make sense?

            I’m also not sure about this “you don’t have to mention the cheese” thing. But possibly I’ve been veggie for too long and my knowledge of meaty sandwiches has atrophied.

            I am also a vegetarian, but my wife and most of my friends aren’t. I have plenty of opportunities to appreciate bacon as a social phenomenon, even if I don’t consume it.

    • I’ve heard a theory that after decades of low-fat propaganda– long enough for animals to be bred to have less fat– bacon was a way of bringing flavor back to meat.

    • Tom Womack says:

      I found that, when I had bacon in the fridge, often it would sit there until it went green and then I’d throw it away; so given the choice between the really not very great inconvenience of cooking bacon and then having a bacon sandwich, or making a cheese sandwich with the cheese that was there, I’d always pick the cheese. So I effectively don’t buy bacon.

      I think the problem there is that cooked-and-left-to-cool bacon is not as good a foodstuff as cooked-and-left-to-cool chicken or sausages, so if I tend to cook a batch of food to have as finger-food for the rest of the week it will be sausages or mini-chicken-fillets.

      But I eat a fair amount of bacon because it’s so ubiquitous in prepared food; if there’s not the extra inconvenience of cooking it, I will take a chicken-and-bacon sandwich over a chicken-and-tomato sandwich.

      • David N says:

        10-12 minutes in a 400 oven. It’s possibly one of the easiest things to cook.

        • Tom Womack says:

          It’s one of the easiest sandwich fillings to cook, which makes it one of the more inconvenient sandwich fillings to work with at home in comparison to cheese, cold sausage, tuna-from-a-tin, smoked salmon, cold roast chicken, cold ham …

    • tcd says:

      I am not really for or against bacon, I will enjoy it when it is around but I do not go out of my way to buy it.

      A simple pet-theory explanation for bacon rising to sponsored meat of the internet status: it is trivial to cook and near impossible to mess up.

      • RCF says:

        Bacon is “near[ly] impossible to mess up”? I was going to respond with the opposite point: Linch may have never had proper bacon. “Bacon” varies quite considerably in content, preparation, and cooking. I’ve had delicious bacon, meh bacon, and disgusting bacon.

  23. Linch says:

    Hmm…mortality is one thing, but morbidity is something to consider too. I expect to enjoy 1 additional hour of life significantly more in my twenties than past the age of eighty-five.

    That said, assuming the entire curve goes down evenly, 1 hour is HUGE. How many people will be willing to eat a bacon sandwich if it means they’ll drop unconscious for an hour afterwards?

    • Scott Alexander says:

      When I have a really big meal, I’m pretty useless for an hour afterwards, but I still do it sometimes.

      • Linch says:

        Yeah, for a huge family meal (or one with friends) I get that, but for a SANDWICH? Also, “useless for an hour afterwards” presumably applies to productivity, which is super-important and all, but I assume you’re still capable of generating fuzzies for yourself in that hour, like your internal mental state isn’t completely neutral.

        Unless you feel so guilty for that lost hour that the hedonic effects wash out, which I certainly hope isn’t the case.

  24. bacon - No! says:

    If someone told you and 99 other people to line up against a wall, and then said, “I am going to shoot one of you randomly. Your only way to avoid this 1% chance of being shot is to [do something kind of difficult but actually good for you].” Would you do it? My guess is colon cancer is at least as bad as getting shot, and the fatal kind of colon cancer a lot worse than the fatal kind of getting shot.

    I understand I am glossing over a lot of statistical nuance here, but it seems the simple conclusion ought to be: “Eat less meat.”

    • chaosmage says:

      Your analogy is invalid. Being lined up against the wall and shot right now is far, far worse than dying from cancer, with time to put things in order and say goodbyes, around age 69.

    • Sokka says:

      Your argument is intuitive and pithy, but I think it proves too much. I imagine people would feel the same way about having a million people lined up against a wall. How many one-in-a-million chances in your life do you live with? How does this argument play out against driving vehicles, riding a bicycle, preparing your own food with knives, drinking alcohol, going to clubs, swimming in a swimming pool, swimming in the ocean near the beach, going to foreign countries to see the sights, …

  25. Ishaan says:

    Not that this isn’t nice to know, but for this sort of thing there’s a rather high premium on the life year reduction analysis being quality adjusted. Cigarettes are case-in-point: 15 minutes of life lost per cigarette doesn’t sound that bad – but the reality of cardiovascular/breathing problems and the potential for mood imbalances and reductions in cognitive function can potentially lower the quality of entire decades. Life to years > years to life.

    So I guess essentially we need word on processed meat’s impact on bowel function / physical and mental capacity / a mechanistic understanding of it in general.

    • stargirl says:

      15 minutes per cigarette sounds extremely bad to me. For such a high price to be manageable you would need a large payoff. I do not think that the average cig provides such a payoff. Though for example an lsd trip certainly would (many times over)

    • My impression is that cigarettes aren’t exactly a pure pleasure– the major reason for smoking is not feeling the pain of withdrawal. I’ve never smoked, so I don’t know how much the cessation of that pain counts as a pleasure.

  26. Ilya Shpitser says:

    “In the case of red meat, the classification is based on limited evidence from epidemiological
    studies showing positive associations between eating red meat and developing colorectal cancer
    as well as strong mechanistic evidence.
    Limited evidence means that a positive association has been observed between exposure to the
    agent and cancer but that other explanations for the observations (technically termed chance,
    bias, or confounding) could not be ruled out.”


    Have to bake structural uncertainty into this also. This is pretty indirect. Usually if the correlational effect is super strong, then somehow confounders aren’t very believable because they have to be incredibly strong also for there not to be a causal link. But doesn’t seem that strong here. Processed meat gets a strong classification, but this is based on associations also.

    I love how that infographic tacitly moves from correlational evidence to making bold statements about counterfactuals.

    I think actual cited studies generally use very cautious language.

    Personally, for effects this weak, I am more worried about incentive structures that reward finding positive effects.

  27. Tracy W says:

    I love your use of examples and illustrations for complex statistical ideas and plan to steal them.

  28. Marcel Müller says:

    “I told her that the cookies were probably taking more time off her life than the medication would, and I assured her the medication would probably add more value to her life than cookies.”

    Upon reading this my thought process was something like this:
    Huh, no just no, nutrition effects are, severe deficiencies aside, never that big (citation obviously needed but see bacon). Did he just add up calories, convert to fat tissue, convert to BMI and look up in a table?Wow, wait a minute who wrote this, yes, Scott “debunking three bullshit studies before breakfast with statistics” Alexander. No, no way Scott would do something that irrespnsible and use the result to convince a patient. Wow, this is really confusing. Does Scott know about a very good study about badness of cookies? No, nutrition scientists just do not do randomised intervention studies on cookies with a lifetime followup (and if they did, they wouldn’t replicate 🙂 ) and correlational studies just don’t cut it to make such claims and Scott knows that. Now I am really confused.

    • Scott Alexander says:

      Eating bacon every day lowers your lifespan two years, and you don’t think eating cookies every day can lower it six months? Note that eg there’s some suggestion that 25% more sugar per day in mice can double mortality; not sure if this is the sugar itself or the extra calories. I can’t prove what I said but I still find it very plausible.

  29. Ral says:

    I think you might be confused about relative and absolute here.
    to demonstrate, i have switched the articulation in this paragraph:

    “But also: use absolute risk instead of relative risk! ‘[21 extra cases of bowel cancer in 100 cases of bowel cancers caused by meat]’ doesn’t give you a really good handle on how worried you should be. ‘[1% of lifetime bacon-eaters get bowel-cancer]’ is better.”

    Both are relative statements. They just have a different relation in the form of a different target group (people who will get colon cancer vs. all people). If I’d be able to – and knew, that i will get colon cancer, the 21% figure is what is interesting to me. Since i don’t, the 1% figure is important to me.

    Also, you may be tripping over repeatability here. Since i can’t repeat my life when i die, i must not calculate risk*impact in low probability, high impact situations. It’s the same problem with x-risk prevention. you can’t ignore it because the probability is very low.
    The absolute outcome is undefined in both situations.

    So the idea to tackle this problem is trying to balance effort and risk. Lower the risk as efficient as possible.
    For some, that might well be ignoring the risk completely because it is already low and it would take too much effort to get to be able to lower it at all.

    It is also important to focus on high risk, low avoidance effort conditions first. Like smoking. or driving more careful. or not working as a climbing boy before 1875.

  30. I’ll make roughly the same comment I made on tumblr: I don’t give a shit about living past seventy unless some major technological innovation happens that makes being seventy not suck. In which case I probably won’t have to worry about cancer anyway. Not all years of your life have equal value, and every time I read a study like this all I can think about is how little I value the time I’ll spend in a nursing home.

    • I don’t think most seventy year olds are in nursing homes, but does anyone have actual numbers?

      • Jaskologist says:

        In this case, it is probably more instructive to look at the fates of your own grandparents. If they lived long, independent lives, you probably will too. If they keeled over from heart attacks at 50… well, at least you don’t have to worry much about the nursing home

    • chaosmage says:

      The happiest seniors I know taught me not to go to a nursing home. Go to a quiet, cheap place in the country, with a big garden, books, wine and maybe some weed or LSD. Have visitors, but mostly enjoy solitude. Write memoirs or letters or a blog, listen to all your favorite music again, live every day as if it were your last. See a doctor occasionally for the painkillers, but not often enough for her to actually prevent the preventable medical issue that’ll eventually kill you.

      Old age ain’t for sissies, as they say, but it does come with a remarkable freedom from responsibility that can be lived and enjoyed – provided you’re not in a nursing home where your entire environment wants you to have a pulse, and hence pay them, for as long as possible.

    • The Anonymouse says:

      Therein lies the problem, though. A cavalier attitude about old age (and the consequent live-for-now fuck-everything mindset) creates exactly the lifestyle that will require being in a home at seventy, and for it to likely suck. If you live like your older years are going to suck, you’re going to make your older years suck.

      It’s /fun/ to be a hot-shit 20 year-old living like there’s no tomorrow. But give it a few years, and you’ll likely make the acquaintance of plenty of 55 year-old men who also did so, and have been brought low by diabetes, emphysema, cirrhosis, or crippling accident. (If you want proof, just head over to your local VA hospital and hang out in the lobby for an hour.)

      Young people are such amazing time-discounters that it frequently overwhelms their ability to do basic math. Everyone over fifty feels ‘old.’ A guy who keels over at 55 falls into the ‘huh, younger than average, but died old’ category, while someone who kicks it at 105 falls into a similar ‘huh, pretty darn old’ category. But the difference between the two is fifty years of extra life! If you were born in 1910, would you have preferred to have died today, or during the Johnson administration?

      • Sokka says:

        I don’t think this argument is relevant to the issue at hand. We aren’t talking about being a “hot-shit 20 year-old living like there’s no tomorrow”, we’re talking about eating meat. From anecdotal experience, I think most of the people keeling over dead at 55 AND most of the people living grand lives at the age of 55 spent their lives eating meat. Perhaps in different quantities or qualities, but we’re not talking a diet of krispy kreme and ecstasy, here.

        • The Anonymouse says:

          The comment was a reply to Wirehead Wannabe, not the broader discussion of processed meats.

          • Sokka says:

            Yes, but I don’t actually think Wirehead Wannabe spends all their time coked-out and clubbing in anticipation of dying at the age of 55. I think it is meant to be a statement taken in the context of the broader discussion. For an example, imagine, I dunno, if someone made a post explaining how impoverished African children were smashing the aid being offered to them because it was the wrong shade of green and their culture said it was associated with homosexuality and outgroups, or something else hopefully fictional like that — a commenter saying “I don’t give a crap about these African kids’ preferences” probably does not mean “I don’t give a crap about what African kids’ preferences are at all, including their preferences not to become horribly ill and die all the time”, but rather “I don’t give a crap about what African kids’ preferences are with respect to the colour of the packaging on their vaccine boxes”. Wirehead may correct me, but I would imagine this is more a case of the latter.

      • tcd says:

        I do not want to speak for the OP, but I am not sure that he was suggesting the cavalier attitude.

        Within my own extended family I have witnessed both extremes. I had a great grandmother live a wonderful active life into her late nineties. When she was finally hospitalized she was dead in two (!) days. Minimal suffering and no complications, this is the death after a full life I would wish for every individual. I had a different close relative show signs of Alzheimer’s in her seventies and live well into her eighties. These years were not only traumatizing to her, with both increased mental and by-product physical suffering, but tragic for everyone involved.

        Again, I will not speak for the OP, but there is a consistency in their position which you might be overlooking. I have no plans for living past the expiration of either the physical or mental capacity of my body. I take very good care of myself now (but far from perfect), and I suspect this will pay dividends down the road, but of course it may not. Will there be medical innovations between now and then? Certainly, but unlike your cavalier type I would not live my life in a way that supposes their existence when I would need them. Desiring to not live past the usefulness of your own body or mind does not imply their mistreatment.

        It took me a while to accept this position, though recently my parents both revealed to me the same preferences (this is a common position for doctors to personally hold).

    • I am currently seventy and life is as enjoyable as in the past, which is to say very.

      I quit sword and shield fighting a few years back, in part because recovery from strained muscles and such was slower than in the past, in part more general worries about increasing fragility, in part concern with having a metal plate in my skull. That’s the only thing I can think of that I stopped doing largely because of age.

      I’m pretty sure my father was still playing tennis past seventy. My wife’s grandmother was still functional and coherent, although with some memory problems, and pretty clearly enjoying life, past a hundred.

  31. So while we’re talking about food, does anyone have any unusual alternative meats that they like to eat? Specifically, does anyone here eat insects? I’m kind of intrigued by the idea though I’m not sure if I’m game to actually go for it as a thing…

    • Berna says:

      I’ll definitely try them if I get the chance.

      • simon says:

        For edible insects google Thailand Unique. I have tried their grasshoppers (tasty) Dung Beetles (also good) and Ants (fell to pieces in transit. Not worth the bother)

    • hlynkacg says:

      Deep-fried grasshoppers can pretty be damn tasty if you get them fresh. They remind me of a weird cross between pop-corn and crab meat. A lot of entertainment venues in southern Asia serve them as a concession.

      Personally I’m fond of Alligator-tail, and of old school Creole Barbecue in general.

  32. Nero tol Scaeva says:

    Bacon fact #786:

    Bacon is healthier for you than crystal meth.

  33. onyomi says:

    Though it sounds unimpressive to say eating bacon every day lowers your lifespan by two years, isn’t that true of most health advice, other than “don’t ride a motorcycle without a helmet”? I mean, at any given moment there are so, so many different factors contributing that I, and probably most people, just have to do our best to do more of the good things and fewer of the bad things, hopefully resulting in a significant improvement in actual lifespan and health on net.

    Like, on any given day I am generally trying, if not always succeeding at: eating more fruits and vegetables, eating fewer animal products and fat, drinking fewer sugary drinks, exercise more, sit down less, and, of course, not smoking (though that I actually never do). It could be I’m wrong about one or more of these things, but I think on net they probably make me a little healthier than I otherwise would be, if not as healthy as I optimally could be, even if the effect of each one individually seems negligible.

    That said, it does seem like, if one happens to just have one vice–like you exercise a lot, eat lots of fruits and vegetables, and just happen to love bacon, then the negative effects of that one vice, so long as it’s not helmetless motorcycle stunt riding, are probably not as bad as a sensational news report may make it seem.

    That said, it doesn’t mean those reports are useless: 100 years ago I think people genuinely didn’t know smoking was bad for you.

    • Randy M says:

      “probably not as bad as a sensational news report may make it seem.”
      Really going out on a limb there.

  34. I wonder if being terribly health-conscious is yet another kind of signalling, but OTOH my background contains lots of dying-at-60-due-to-cigarettes-and-alcohol-is-totally-okay kind of influences which are probably just as extreme.

    Nevertheless I smoke and drink daily, and well over recommended units. Could something like diet matter upon that, or would that be more like penny wise, pound foolish? I have followed the paleo vs. others debates and basically what I figured is that fast food, refined carbs, refined sugar and suchlike is probably bad, but whether your breakfast sandwich contains bread or not, contains two slices of ham or not probably doesn’t matter much. Certainly not even close to as much mattering as cigarettes or a few bottles of beer.

    So I have to think that e.g. when paleo types say rice is basically poison or this meat causes colon cancer thing as something like virtue signalling.

    Then again I must consider that these stuff are probably not written by people whose dad died at 62 due to bladder cancer from all those cigars. And his dad at 56 from all the booze. What looks like signalling to me, to someone far more health conscious than me could look like just a next step of optimization.

    This may be a general rule. Doing something one step better than you do is optimization. Five step better, and you look like a snob. I work out sorta hard enough but inconsistently. Consistent workout is an improvement. People who debate about whether squats are better than leg presses look like virtue signalling snobs to me, my view if anything made my thighs sore this week and preferably twice then it was a good week for the legs and I don’t care what, but perhaps if you work out consistently, that sort of thing is the next step of improvement.

    After all, there was a time, in recent memory, when not smoking was a bit of a virtue signalling too. Now we all see, even we smokers, that there is more to this.

    Anyway, I still don’t know if you are a smoker, drinker and general bad boy, does it matter at that point if you eat salmon salad for lunch or sausage. Whisky, cigar, salmon salad? Does that make any sort of sense beyond whisky, cigar and bacon?

    Does history now of smokers and drinkers who were nevertheless fairly healthy food conscious and what was the outcome?

    Another health related thing. I plan a HUGE article on Dividuals about stiff painful upper traps / poor posture / Upper Crossed Syndrome, with everything from recognition, posture, stretch, exercise and so on. But since everybody who spends a lot of time on a computer has some amount of this problem, I figure Rationalists already attacked it. It’s literally a huge pain in the neck. Can someone give me a pointer? Quality control: if it mentions lower traps, it is getting good, if it also mentions the serratus anterior then the author probably really researched it properly. Another quality control: it is about posture, but what it is really about is gravity. Gravity causes the pain, through posture. Did anyone write that article already?

    PS I have a wonderfully non-bullshit doctor and I asked him to prioritize the unhealthy things people do and healthy things people not do, what to optimize first and he gave this

    1) Exercise is first. Exercise is more important than even food or obesity.

    2) Food, obesity comes second.

    3) Alcohol is third. Unless you are emptying a bottle of whisky a day, according to my doc it can actually be a smaller problem than a poor diet. I don’t understand how.

    4) Cigarettes are fourth. In the longer run it kills you, but if you are like 30 you don’t necessarily have to stop tomorrow, it makes no difference about how you feel at 45.

    5) Coffee is fifth. Too much coffee is a problem only if your blood pressure is already fucked, otherwise OK.

    PPS If you live in a private healthcare country like US all doctors work private including the bad ones. If you live in a public healthcare country, EU, only good ones work private, the bad ones work for the government (broadly speaking, so basically they accept whatever the lowest level of health insurance there is). Hence, an EU private doctor is usually someone really worth listening to.

    One wonderfully perverse and useful consequence of socialist healthcare is that it is super easy to identify good doctors: they refuse work in it, they only or mostly do private. I mean, I obviously hate the thing but sometimes bad has good side-effects.

    • keranih says:

      Regarding your physician’s advice…

      1) Exercise – this is strongly supported. “Pandering to Big Exercise” has been mocked in this space before, but physical activity does more for improving/stabilizing physical and mental health than anything else tried.

      2) Obesity – I think the evidence of this fairly weak. Under weight people have a high death rate than over weight, and joint issues are about as prevalent in highly active thin people as in overweight people. This is an area highly influenced by social aesthetics. Where obesity really starts kicking in is at the higher edges of the spectrum where it interfers with exercise. If one is talking strictly about healthy diet, then I agree that the impact is much larger than just being overweight. But it would be awesome if we had better data on what is a “healthy diet.”

      3) Alcohol – in moderate amounts, alcohol improves health. The key is sticking to ‘moderate’ levels. (And it’s not just the hammering blows to the liver, but also the impacts on social interaction and higher loss due to accidents and violence.)

      4) Smoking – this is my most serious disagreement. Smoking makes almost everything it touches worse – if you have a choice between working in an asbestos mine or smoking, work the mine. Between heart attacks, strokes, various cancers and the impact on wealth, smoking is a very bad deal. While it is not worse than all the others put together, it comes close.

      5) Coffee – largely a non-issue. Sexually transmitted diseases would probably be a better fit here, particularly for women (who would also have to include the pregnancy risk.)

      • Anonymous says:

        >5) Coffee – largely a non-issue.

        TheDividualist said that. It’s only bad if you have cardiovascular problems.

      • Accidents and violence wrt alcohol are actually an excellent example why personal advice should not be gleaned from statistics and why the Rationalist habit of over-reliance on statistics is not ideal. There is a R. habit to think statistics equals probability but it doesn’t.

        I mean, as for accidents, it takes a very special person to ever even think about driving drunk. I mean, basically a gigantic asshole and very high risk taker. Of course living somewhere with good public transport and not good, non zone and zoned bars (so bars where you live) and bar drinker or home drinker also matters. Same about violence .People react extremely differently to alcohol, some indeed get violent, others go the opposite angle and write incredibly elaborate pseudophilosophical ramblings.

        In other words, the only good way to predict how alcohol actually affects you in these senses is to check the evidence of how it actually affects you. This is a good example of why I don’t really like this Rationalist habit of relying on statistics.

        Now as for liver and everything else, you can rely on stats, because human bodies are far more similar than human personalities and habits.

      • Linch says:

        “Under weight people have a high death rate than over weight” This point seems to contradict evidence on calorie restriction.

        • Nornagest says:

          Well, there’s an obvious confounder in that a lot of unhealthy people are a lot more likely to be underweight, such as the elderly and people suffering from certain chronic illnesses. I don’t know how well keranih’s sources have been correcting for this, but my guess, judging from the average quality of research around diet and weight, would be “not very”.

          That said, even after that there are nasty complicated correlates going on here — lifestyle and class and geography and genetics, to name a few — and they’re next to impossible to untangle. I’d be deeply surprised if obesity made you live longer, all else equal, or if exercise (below obsessive levels) didn’t, but we shouldn’t be surprised to find weird relations in some of our data.

          • Cadie says:

            There’s also the fact that overweight and obese refer to different weight categories, and people marked overweight, especially men since it doesn’t control for lean weight and isn’t very accurate for tall people, are much more likely not to actually have a serious weight problem than those in the obese category. And health problems that do result from excess weight are dependent on how much excess weight there is – 15 lb does much less harm than 100.

            Underweight is a category that counts everyone below a certain weight for height, but overweight is only the slice of the high-weight population that has the least amount of too much weight. So it’s not a very fair comparison.

        • keranih says:

          Study here.

          Obvious disclaimers are obvious: we separate out ‘overweight’ and ‘obese’ but not ‘underweight’ and ‘cachexic’. Other confounders are not so obvious.

          I don’t know how well keranih’s sources have been correcting for this, but my guess, judging from the average quality of research around diet and weight, would be “not very”.

          *busts up laughing* Nutrition research is…difficult, given the lifespan of the target species, difficulty in accurately recording intake, and the frankly appallingly anti-science attitude of IRBs toward destructive testing and terminal experiments.

    • Anonymous says:

      >People who debate about whether squats are better than leg presses look like virtue signalling snobs to me

      I immediately thought of several reasons they are better, so thank you for picking what was for me a very effective example!

    • Paul says:

      Coffee is (arguably) good for you. It’s the sugar, cream, milk and so on that we add to it that’s likely to be problematic.

      I must admit to bias though. As someone that drinks a fairly large amount of unsweetened black coffee a day, I’m hopeful that the study results that show coffee consumption is related to lower dementia risk (among other things) are not just accurate, but causal.

  35. Kevin C. says:

    This sort of diet-lifespan things always remind me of a bit from the 1950’s song “Black Strap Molasses“:
    “Black Strap Molasses and the wheat germ bread / Makes you live so long you wish you were dead.”

    • Sokka says:

      I have never had wheat germ bread, but it sounds delicious to me. Particularly if the wheat germ is toasted. Is there something horrifying about it I am missing? The recipe I found on google even calls for molasses, but it sounds lovely, and the author raves over it.

  36. Andrew M. Farrell says:

    How does Risk Aversion play into this? If someone is highly risk averse, doesn’t an 0.4% chance of losing 10 years have a great deal more negative utils than a loss of two weeks?

  37. Tom Scharf says:

    What has happened to science journalism? I think most people with half a brain read headlines on this study and instantly thought “this is being overstated”. The media plus science have gone down a black hole and credibility is really suffering. Everything is now formulated for “maximum impact” instead of “maximum accuracy”.

    It’s pretty easy to perceive that food science groupies have declared a war on meat, they seem quite emotionally entangled. Being a vegetarian seems to be as much about twisted morals as anything else. In fact there seems to be some really perverse thing going on that anything that actually tastes good requires immediate studying to determine why it shouldn’t be allowed. Meat, butter, salt, sugar, saturated fats, and of course…..The Big Gulp. It’s not like they haven’t looked into caffeine a thousand times and basically found nothing, but by God, it requires much further study.

    I suppose after the low hanging fruit of tobacco and obesity have been plucked, things get much more nuanced and harder to define. And the food research industry must continue on with their research and fill their days. But I wonder, is there really much left here in a cost/benefit analysis? If the entire food research complex disappeared overnight, would anyone really notice?

  38. Josiah says:

    The big unaddressed issue here is what kind of death is eating meat likely to give you. Is it likely to be a long painful process, or you just going to keel over one day while tending the garden? And of course you have to compare the unpleasantness of death by meat to the unpleasantness of the death you are likely to have otherwise.

    At this point those sorts of issues loom larger than wether I will live to be 78 or 80 (obviously by the time I turn 77 I may have a different perspective).

  39. baconbacon says:

    This seems like the right place to share this.

    What almost everyone is ignoring is the entire point of life is to have grandkids. Never, ever do you hear grandparents badmouth their pre-teen grandkids, and never, ever have you seen them happier than having them. I worked with a woman in her mid-late 40s that was working back 2 back bakery shifts starting at 3am 4 days a week to make ends meet. When her son told her his girlfriend was pregnant she cried….. out of joy.

    The optimal life is to make enough money to get through raising your own kids (enough to give a high probability of having grandkids before you die), have grandkids, and then die before the youngest turns 12.

  40. Phil Goetz says:

    I tried once to estimate what the odds were that smoking for most of your life would give you cancer. It turned out that I couldn’t find any data directly addressing that, because smoking declined over time while lifespan (and cancer incidence) increased, and while the population of smokers became more enriched with people with other confounding demographics and habits (like maybe, just guessing here, alcoholism).

    So, for instance, that picture above says smoking cause 19% of cancer, and 19% of the British smoke. But most of that cancer is in the older population that had much higher rates of smoking. So figuring out what fraction of people who smoke get cancer requires integrating fraction of cancers caused by smoking over the cancer cases by age in a way that hasn’t been done.

    I did some interpolation between different data sets, and came up with the estimate that, given today’s lifespans, smoking regularly has a 50% chance of eventually killing you. I don’t remember which datasets, though.

  41. Vijay says:

    “risk professor” is an awesome job title and “David Spiegelhalter, Risk Professor”

    “Sir David Spiegelhalter, Risk Professor”, you peasant!

    Mirror Holder? Is that a profession?

  42. RCF says:

    It seems to me that there’s a rather important point that you didn’t mention and which, from what I can see, no one else noticed: The subtitle is “UN health body says bacon, sausages and ham among most carcinogenic substances along with cigarettes, alcohol, asbestos and arsenic”. I think most people understand “most carcinogenic” to mean “cancer per unit”, not “total cancer”. Space travel is more dangerous than air travel, even though orders of magnitude more people have died in the latter compared to the former. If people ate asbestos every day, I think that we’d see more asbestos-caused cancer than bacon-caused cancer. And eating uranium would cause even more cancer; how is that not on the list?