JOSH: Hello, and welcome to Clearer Thinking with Spencer Greenberg, the podcast about ideas that matter. I'm Josh Castle, the producer of the podcast, and I'm so glad you've joined us today. In this episode, Spencer speaks with Richard Bruns about cost-benefit analysis, improving air quality, and drug regulation.

SPENCER: Richard, welcome.

RICHARD: Thank you.

SPENCER: Today we're gonna talk about some important and spicy topics like: Is the air killing us? How do we know? What can we do about it? How do we make big mega systems like the FDA and bureaucracies in general work better? And also cost-benefit analysis. I'm excited to dig into these topics with you.

RICHARD: All right, good.

SPENCER: Let's start with air quality. Can you just open this topic for us? Why does air quality matter?

RICHARD: Basically, the stuff that you get breathing into your lungs can be very bad for you. In the extreme, you have the London smog situation that kills thousands of people from acute exposure. But even stuff that you wouldn't even notice, stuff that you wouldn't even see, can get into your lungs and kill you with chronic exposure.

SPENCER: And how do we know that? Because presumably, it's not the sort of thing where you can do randomized control trials where we put bad quality air into some people's homes and not into others' or that kind of thing. So where does the evidence base come from?

RICHARD: Epidemiology mostly. There's a lot of various kinds of controlled experiments that have been done, often in China. For example, back a long time ago in China, they did a program where all houses north of a certain geographical feature got free heating coal in the winter and the houses south of it did not get it. So you have this kind of natural experiment discontinuity and the indoor air quality north of that was a lot worse because they were burning coal in their houses, and you started to see a lot of health effects from that. And there's just a large, large pile of other epidemiological evidence usually based from outdoor air sensors, like when the particulate matter fine particulates (or PM 2.5) is high, you have a lot more people showing up to the emergency room, people dying of strokes, heart attacks. That's the more obvious and you do have longer-term chronic studies. But there's been a lot of that kind of strong epidemiological data coming in over the past few decades really showing the harm of this stuff.

SPENCER: Let's get our terminology straight here. What does PM 2.5 mean?

RICHARD: That means particles with a diameter of 2.5 microns or smaller. That's much smaller than the diameter of a human hair. There's a graphic that's on the CDC website and other places that you might have seen before where, if you basically string a bunch of PM 2.5 particles around a human hair like it was beads on a necklace — these are super, super fine particles that don't get caught by your mouth and the protective mechanisms in your lungs that evolved for larger things — they'll shoot straight into your lungs, straight through the cells, and into your body.

SPENCER: So it's the little particles that are actually the dangerous ones?

RICHARD: Absolutely, yes. PM 2.5 or fine particulate matter, those are basically synonyms.

SPENCER: Presumably, there's some size where if it gets small enough, it no longer matters. Is that true?

RICHARD: That is a very good question and I never thought about it. The problem is, it's very hard to study. In the past, people had trouble weighing and counting these things. And they've recently had some NIST standards that make it easier to actually have a consistency of what you're talking about. Because you have a bunch of particles in the air, you'll have big things: the soot, PM 10, and almost all of the weight is from the really huge things. So counting by weight, it looks like almost everything in the air is the heavy stuff, but your lungs will filter that out. So I honestly don't know if the really tiny stuff would be mostly harmless or if it's just a matter of, we don't actually know how to filter out stuff that's smaller than half a micron and measure its effects separately from the stuff in the half a micron to 2.5 range. I'm not a technical expert there.

SPENCER: And how much does it matter what the stuff is? Are essentially all random particles of these sizes bad for us? Or are there some that are harmless and others that are harmful?

RICHARD: I'm not aware of the type of particle mattering a lot. I would suspect that heavy metals are a lot worse than other things because you've got the heavy metal dose in addition to that. But as far as I know, it's just the size of the stuff getting into your system; the composition doesn't matter that much.

JOSH: After the episode was recorded, Richard reached out to us with this note:

"Since recording this, I have learned from talking to technical experts that there is ongoing research into the effects of the chemical composition of the PM2.5 particles. The current working assumption is that the composition does not matter much; but some people suspect that it does and are investigating."

SPENCER: Yeah, I mean, we definitely know there's some particles that are really, really bad for you, like poisons and so on. I'm sure there are some that are worse than others. I was just wondering if there are ones that are benign, or just everything that size is bad, at least to some degree. What do we know about how these accumulate? Is it more like mercury, where you build them up over your whole life and then eventually it causes problems because your body can't get rid of it? Or is it more like your body recovers and so it's really about avoiding having too much in (let's say) a one-week or one-month period?

RICHARD: My impression: while it doesn't bioaccumulate, it's always stressful. I think a good analogy might be UV radiation or sunburn. Ultraviolet radiation doesn't bioaccumulate in your body. You don't get radioactive from spending too much time in the sun. But every exposure hurts you. Obviously, massive exposure will hurt you, a little bit will hurt you and accumulate, and it's the same kind of thing.

SPENCER: But doesn't DNA damage accumulate with light exposure?

RICHARD: Yes, it's a similar kind of effect. Damage to — not necessarily DNA, although there might be a bit of that — but damage to your blood vessels, your lungs, the blood vessels in your brain, these kinds of things cause stroke as well. So yeah, it's damage to your body that's accumulating (not like mercury) or the concentration of the stuff increases in your body. So bioaccumulation and long-term chronic damage are different things, and it does the latter, not the former.

SPENCER: Okay, got it. So somehow these small particles are actually damaging yourselves. And then that can accumulate if maybe the repair mechanisms can't repair the cells fast enough, or if it's a type of cell that doesn't get easily repaired or something like that?

RICHARD: Exactly. I'm not aware of there being any kind of hormesis effect for PM 2.5. I would assume a linear no-threshold dose response for this. Are you familiar with those terms?

SPENCER: Why don't you just define those really quickly?

RICHARD: Okay. For some kind of harm or toxin, a small dose deals no damage or, for some weird reason, sometimes possibly even negative damage, like sometimes the damage is so small, but it activates the body repair mechanism and your body burns calories to fix the thing, and you wind up being a little healthier than before. That's called hormesis where a tiny dose seems to slightly improve outcomes. But with PM 2.5, I would basically assume that it's similar to lead and heavy metals kinds of issues where a little bit hurts, more hurts and, if you do a graph, with exposure on one axis and harm on the other, it would basically be a line with nothing weird going on, except extremely high doses can actually start to have acute effects and start to kill you. But for most modern situations, you're going to be in the area of the graph where it's pretty much a line: a small amount will hurt you, twice as much will hurt you twice as much.

SPENCER: Okay, that makes sense. What do people need to know about air quality and how it might be affecting their own health?

RICHARD: The most important thing you need to know is, get a better filter for your house. That is something that anyone can do. It's very cheap. It's very easy. You get a MERV 13 or better filter. They're basically going to be ten or 20 bucks compared to two or three bucks for the cheap, crappy filters. Change them out three or four times a year. It's a very small expense, a very small inconvenience, for a big health boost of cleaning a lot of crap out of your lungs that might kill you.

SPENCER: Now, MERV 13. What does that refer to? Is that the particle size that can be trapped?

RICHARD: It's a complicated multifactorial thing. It doesn't have a simple technical explanation, but basically, higher MERV numbers are better and MERV 13 is the minimum number that actually traps a significant number of these fine particulate matters. MERV 8 is almost useless. MERV 11 will get a little bit of the stuff. But for a couple bucks more, a MERV 13 will start to trap 95% plus of these fine particles.

SPENCER: If people go on Amazon, and they just search 'air filter,' and they buy one of the first few things that shows up, is that likely to be good enough? Or do you actually have to specifically seek it out?

RICHARD: Probably not. You need to type in MERV 13.

SPENCER: Got it. Do a lot of people buy air filters that essentially don't filter particles?

RICHARD: Yeah, and part of that is because they're getting bad advice from HVAC people. Basic cheap air filters are not designed to protect your health because it's only been in recent years that we realized just how bad it was. They're just designed to protect the equipment so these cheap filters that are like this see-through mesh thing, they're basically just like lint traps in a dryer that are only designed to protect the equipment. What often happens if you're lazy and you don't change them out a lot, if you use a good filter, it will get completely clogged up, and then your system won't work as well, might sometimes even suck the filter into the fan. So because of that some people said, "Oh, just use the cheap thing and change it out once per year. That's good enough to protect the equipment." And yeah, it is, but it's not filtering out the stuff that's killing you by getting into your lungs. That's why this older stuff is on the market, because people didn't realize or think about protecting their health instead of just convenience.

SPENCER: If you look into HEPA filters, which is what seems to come up when you do searches for filters, they'll say things like — I'm looking at a HEPA website right now — it says, "This type of air filter can theoretically remove at least 99.97% of dust, pollen and mold bacteria and any airborne particles with a size of 0.3 microns.

RICHARD: Yeah, HEPA filters are better than MERV 13. The problem is, when you're getting the stand-alone filter unit, the quality of the filter matters less than how much air you're pushing through it. So if you can get a HEPA filter that's sized for your HVAC system, and it's a reasonable price, and you might want to change it out a bit more than, yeah, sure, go the extra mile. But MERV 13 is the minimum. MERV 15 is a bit better. But yeah, HEPA is slightly better, but you should just be getting, at minimum, MERV 13. And when it comes to the indoor air filter things, it's probably useful to use those in the winter times, in respiratory virus season. The type of filter in your unit doesn't matter as much as the Clean Air Delivery Rate or just pushing more air through. Basically think about visualizing, imagine that you've got a really good filter on just a tiny fan that's only moving a little bit of air. Now imagine a slightly worse filter on a giant box fan that's moving a lot of air. That worse filter on the larger fan is going to suck more crap out of the air, and actually do more to protect you. There's this thing called a Corsi-Rosenthal box that actually performs in lab tests much better than HEPA filters, because it's moving a lot more air. Of course, The Corsi-Rosenthal box is basically, you get four of these MERV 13 or better — you can go up to 15 or HEPA, if you want to, but MERV 13 is kind of the sweet spot of cheap but effective — you make a box where you've got the box fan blowing upward, and four of these filters around the sides. So the box fan is sucking air through four of these filters and that will do a really effective job of circulating the air in your room, clearing out the viruses and other stuff, so you can have a safe gathering.

SPENCER: Yes, let's talk about capturing viruses and bacteria as well. It might be surprising to people because viruses are so ridiculously small. How on earth can a filter actually catch a virus?

RICHARD: Again, PM 2.5, I've said, is extremely small. These are just a couple of microns across. You basically have this thinly woven mesh that will suck out anything; even at one micron, the tiny dust and particle things, it just goes through a big net. Some of them also have a bit of an electrostatic effect, especially when you first buy it and they're charged, so that charge actually causes the particles in the air to cling to the net, being attracted to it, in addition to the filtration.

SPENCER: Let's see, I think if the COVID virus is 100 nanometers, that's 0.1 microns, that seems way smaller than PM 2.5.

RICHARD: MERV 13 will still filter out virus particles. It's not just the stuff in 2.5; it's the 2.5 and smaller. MERV 13 or better will filter out most things that aren't gases.

SPENCER: I've heard there's a funny phenomenon where the reasons that they filter larger particles are different physically than the reasons they filter really tiny particles. There's a different mechanism by which they capture the really, really tiny stuff from the stuff that's around the 2.5 that it's actually rated for. Okay, so basically, if these capture virus particles, bacteria, does this mean that people using air filters are actually less likely to get sick, and do we have direct evidence of this?

RICHARD: Oh, absolutely, yes. They've done a lot of tests of these. They can't do controlled trials on humans, but they basically have rooms full of, imagine things like CPR dummies that have little pumps in them to breathe in and out. And they can look at how many virus particles are trapped in the mouth covering of the dummies. Yeah, there's a lot of room tests. EPA has them, NIST NIOSH. There are a lot of very carefully done environmental testing where they run lots of different air cleaning technologies, release a virus into a room and then see what happens, how much of it's sucked out of the air. One of the best real-world pieces of evidence that we have from this is the Mendell et al. study about California schools. Basically, it was a natural experiment. They measured the ventilation rates in a bunch of different schools, and they found that schools that had a higher ventilation rate had noticeably lower rates of sickness and absence.

SPENCER: That's really interesting. Yeah, I was gonna ask if you know if there's ever been a study where they actually randomized to give people filters for their homes and see if that makes a difference for how often they get sick.

RICHARD: Actually, there was one that was profiled in Vox where, because of some industrial accident, they gave a bunch of air filters to everyone in the vicinity. And the air filters that they gave didn't really do much for what was released from the accident, but it had really noticeable impacts on the health of the people that they gave them to. Where we keep trying to make it happen — and it's always hard to convince people to randomize over a large area — but there have been a lot of natural experiments or other kinds of things where, for plausibly random reasons, filters get installed in one place or another. Or you just look at the differences in filtration in different buildings that happen because of different corporate investments or randomly different policies, and buildings with better filtration have less sickness.

SPENCER: Let's talk about effect sizes. Maybe these kinds of filters, they can help prevent negative health effects from these small particles, and maybe they can help capture viruses and bacteria. But are these effects likely to be big enough that for someone (let's say) living in New York or San Francisco or LA or London, that it's actually going to make a difference, that it's actually a worthwhile investment?

RICHARD: Worthwhile investment, yes. Different people have different ideas of what it matters to make a difference. In public health, when you're talking population-wide effects, a five or ten percent difference is huge for us. If you can reduce the number of people who catch COVID by five or ten percent in a virus season (COVID, flu, etc.), you're saving thousands of lives. But from an individual's standpoint, ten percent might not even be noticeable if all you're doing is looking at your personal experience. So yeah, if everybody installs these filtration systems, the effect size, depending on the study, you're looking at five to maybe 20%, depending on how good the filters are, and how much air is moving between them. So it's certainly not like you install this thing and you personally will never get sick. There's a lot of stuff going on. But on a population kind of scale, it definitely matters, and the overall benefits, the expected value, is a lot higher than the cost.

SPENCER: If we're talking about five to 20%, what are the kinds of outcomes that are being affected here? Like how frequently someone gets a cold, or what kind of things do you mean?

RICHARD: Yeah, basically, how frequent your chances of getting sick, so the chance of catching cold, influenza, for the virus things. With PM 2.5, you're looking at asthma; with vulnerable populations, cardiovascular events, heart attacks, strokes, lung-based events. It's the kind of thing where, if the PM 2.5 levels in a city are increased, then you'll have two or three percent more people showing up in the hospital that day, which really matters at a population level.

SPENCER: Yeah, it's interesting to consider. Is it just moving forward? People getting sick when they were gonna get sick anyway? Or is it actually causing events that might not have happened for many years otherwise? Maybe it's hard to know.

RICHARD: There's a little bit of both, yes, especially with a PM 2.5, almost anything except for a few random influenza viruses that kill you with the cytokine storm from your immune system overreacting. Basically, any insult or punishment to the human body is always going to be much more likely to wound or kill someone who is more frail, more vulnerable, etc. That is taken into account when we look at how many life years are lost. A PM 2.5, the particulate matter causing someone to have a stroke, it's usually people who are more vulnerable, people who didn't have as long to live, but you're still chopping five to ten years off their life on average, and occasionally it will get someone young and healthy.

SPENCER: Wait, five to 10 years off their life? That's an enormous amount! Is that a real number?

RICHARD: That's not a population average. That's saying a lot of the individuals who are affected by this, are people who are in an age cohort that you would expect them to live at least five or ten more years.

SPENCER: Right. Okay, so someone who might have a stroke or something.

RICHARD: In terms of the population average, changes in PM 2.5 of a scale that we see in the modern world will still take at least half a year of life off of the population on average. It's pretty big.

SPENCER: Interesting. And I assume that that's not just, oh, instead of dying at 90, you die at 89.5. I imagine that it also affects how healthy you are at different ages, too.

RICHARD: Yes, the health effects are included. And with a lot of these things, it's not an 'everybody loses exactly half a year of life.' Some people who were blessed with good genes might not be affected at all until they die of something completely unrelated. And maybe one out of every ten people in the population gets killed five years earlier than they would have been.

SPENCER: Now how much of this is a big city thing versus it doesn't really matter where you live? Like if someone lives in the suburbs or rural area, is it going to have much less effect?

RICHARD: Well, you're gonna have some effect. The particulate matter is higher in a big city. There's a lot of reasons. Air quality isn't the only reason. But I think the statistic is overall, holding socio-economic status and all of the other confounders that we know how to measure constant, an observably identical person living in a city will expect to die about five years earlier than someone living in a pristine country area. Some of that is air, some of that is other effects, some of that is other kinds of pollution. But yeah, the problem is biggest in big cities; although as we saw with the recent wildfire, if something bad happens, there can be air quality events that affect anyone.

SPENCER: Do you think that some of that five-year gap — which is a very substantial gap — might be just failing to control for all the different confounding variables about what differs between people in cities and non-cities?

RICHARD: Partly, yeah. Although, in general, more highly-educated people with better job prospects tend to move to cities. So you could argue for the confounding going one way or another. But if you don't do any of the corrections, then you don't see an obvious effect. And in some cases, urban areas — especially the rich, high-income urban clusters — have higher life expectancies than rural areas, but again, correcting for confounders. Yeah, it is always hard to know exactly how that matters. So yeah, there's big error bars on that five years.

SPENCER: Interesting. And what about children? Because my intuition at least is that children are more affected by things like this. Like if they're growing up in bad air pollution, that could cause lifelong effects. Do we know anything about that?

RICHARD: Asthma tends to be caused more by mold than PM 2.5, although good air filtration, good air quality is extremely important for mold. And we really need better air quality in our schools instead of just the cheap crap. A lot of our listeners probably remember going to a school where they have this loud, cheap air conditioner somewhere off there that's making this horrible noise, and the teacher doesn't want to turn it on unless they absolutely have to. We really need to invest in better HVAC systems in our schools for that reason: developing bodies, mold, PM 2.5. With children with particulate matter, you're less likely to see heart attack and stroke, because obviously, the base rate of kids getting those kinds of cardiovascular diseases are so small. In terms of who shows up at the hospital when there's a pollution morning, it's going to be older people, people working out. Kids are more robust to the acute effects. But yeah, we do believe that the chronic effects will just cause damage to developing brains and mess them up a little bit for the rest of their lives, sometimes in ways that don't even show up in measurement. I do believe in the linear no-threshold effect where, if we see some people are noticeably messed up, I would generally take as my prior that even the kids where you don't see anything happening, I assume there's a little bit of harm, even if it's below the level of detection.

SPENCER: Have there been studies that specifically look at effects on IQ in terms of particulate levels in the air?

RICHARD: Yes and no, because IQ is a lot of complicated things. We do have a lot of evidence that bad air quality lowers performance on all kinds of tests. Office studies, for example, if there's a bad carpet that has a bunch of crap in it, and it's out-gassing particulates, gaseous contaminants, etc., people will be worse on a variety of cognitive tasks, from standardized tests to IQ tests to stuff that matches real-world performance. So there definitely is an acute effect on IQ as measured by any kind of test or task performance. I would assume, based on what we know from other effects, that long-term chronic effects would also cause that, but it's hard to know exactly what the mechanisms are. I remember a study, they looked at, no, sorry, that was heat stress in schools, not necessarily the particulate matter in terms of the HVAC system messing up that wrecked people's test scores and end-of-year stuff. So I suspect that it happens. I suspect that almost everybody is going to get some fraction of IQ points getting whacked off of them. But it's below the limit of detection in our tests, and people haven't specifically looked for that.

[promo]

SPENCER: I just want to mention that, if listeners are wondering why you don't actually know about all the biological aspects of this, it's because you're not a biologist; you're an economist. You're looking at this from a broader health point of view. Do you want to just comment on that?

RICHARD: As I've said, I don't know a lot of the technical details. My role in the knowledge ecosystem for this is generally a consumer rather than a producer of technical knowledge. I will look at the quality of a study. I'm generally pretty good at knowing, is this a well-done study or not? I know a lot of the standard things that you look for, to see if things are legit or not, to get a good feel of how good the literature is. But once — and I admit this is kind of a weakness because my brain can't hold everything — once I've decided that something is legit and looked into the technical details, I often forget the chain of evidence that led up to that. I just cached the output of it and moved on with other stuff. Whereas an actual biologist would always have at the top of their head, "This is my research specialty. I could go on for five or ten minutes about the exact chain of evidence that led to this." I'm doing much more of a meta analysis of, does this research seem legit? Yeah, it does. Okay, natural experiment, quasi-experimental data, they seem to know what they're doing with respect to causality, decent confidence intervals. There's a stream of literature there. Okay, yeah, seems legit. Now I take all of that stuff — without remembering too many of the technical details — and do meta analysis. Cost-benefit analysis is like a meta analysis where you're combining the effect sizes from a lot of different papers, except then you go on to monetize it and put everything into the dollars and cents bottom-line figure.

SPENCER: Your specialty, as I understand it, is cost-benefit analysis for these big questions like, what is the value of this intervention? Is that right?

RICHARD: Exactly, yes.

SPENCER: Taking this topic in particular, with regard to your speciality would be like, what is the cost-benefit analysis around interventions in terms of improving air quality?

RICHARD: Correct.

SPENCER: Let's talk about where the rubber meets the road. [laughs] What is your personal setup in terms of air filtration? Do you have one?

RICHARD: Basically, the apartment that we're staying in, I upgraded the air filters, got the best air filters you could get at Home Depot. And when the weather's decent and there's not an air quality event, I just keep the windows open as much as possible, even in temperatures that some people might feel uncomfortably hot. It's more of a psyche benefit of being close to nature, hearing the birds, feeling the breeze. I'll just keep the windows open a lot, enjoy the cross-breeze and have good filters. But, no, I don't actually have a separate setup, partly because we don't host gatherings. There's not a lot of people in here. So, no, I don't have any air filters. Just open the windows if things are nice, make sure that the normal apartment HVAC system has the best filter installed and I keep the fan on "on" instead of auto so it's recirculating more even when it's not heating or cooling the air.

SPENCER: Opening the window, that's an interesting one because people might think, "Well, if I'm living in a big city, the outdoor air quality may not be very good. So does opening the window actually make the air quality better or worse?"

RICHARD: Well, this is a suburb. It's one of those four-storey apartment buildings that's surrounded by a lot of trees. I'm looking at a crepe myrtle outside my window now. Outside air does tend to have a bit more particulate matter; inside is more sink than sources for particulates. But when you have the outside air, it's more CO2. You get cleaner air coming in, the carbon dioxide level in your apartment goes down. And also if you've been cooking, or if you have anything like the carpets or chemicals or any kind of plastics or stuff inside, those are generating stuff. Especially if you're doing a lot of cooking, stuff inside your apartment will be generating air, putting off particles. In general, unless there's an outdoor air quality event, it's good to flush things out again, unless you're living next to a big highway or something. In which case, yeah, keep things closed and filter a lot.

SPENCER: You mentioned carbon dioxide, because that's a whole other topic. Have you looked into that? And what are the conclusions around the health effects of carbon dioxide?

RICHARD: It doesn't really affect you that much compared to the particulate matter. So if CO2 levels get really high, like above 2000, then that'll start to affect cognition or performance. But in terms of health effects, it's the particulate matter and the viruses that you catch that are much more harm to you than slightly higher levels of CO2.

SPENCER: And what usually causes those especially high levels of CO2? Would it be like a malfunctioning boiler or something?

RICHARD: It's just a lot of people in the room exhaling CO2 and the system can't keep up. If you've got a lot of people in a small room, and they are breathing out air faster than the HVAC system is moving it around, then CO2 will build up. It almost always comes from people.

SPENCER: Oh, okay. So you could have office environments where you systematically have overly high levels of CO2 that actually hurt people's performance.

RICHARD: Yeah, you can. And any old office, if it's not up to modern code, will often have that as a problem. If people are trying for extreme energy efficiency, and they're not getting enough outside air and filtration, you often have bad air sick building syndrome, etc., where they're just not filtering it enough and not getting in enough outside air.

SPENCER: You'd think that large companies that have many office workers would have a real incentive to make the air quality good for all these reasons: it might reduce sick days, it might improve cognitive performance to some extent, at least relative to having it be overcrowded or having bad air. Is it something that companies tend to be aware of? Do companies invest in high quality air?

RICHARD: The good companies do. Like your large Fortune 500 kind of companies in class-A commercial real estate, they tend to be very sophisticated customers. They demand good air quality. They get good air quality, and their people perform better as a result. But a lot of smaller companies where they don't have an industrial hygienist on staff, they don't have policies and procedures, if it's just like 20 or 30 people, they're concerned with other things, and they don't even know what's going on, then they're just not aware of the effect. So they'll buy cheap office buildings, just show people in and not realize how much it's hurting them.

SPENCER: I'll just tell my story with air quality briefly. I had an interesting experience when I moved to a new, new apartment and I found myself getting sick. I found that I started having trouble breathing. It felt heavy to breathe and I started getting a cough. And just every week, this got worse. And I started thinking, "Well, this seems like it might be related to the new apartment I moved into." But on the other hand, my partner's not having any symptoms, so that's interesting. And I ended up hiring an air quality inspector who came and inspected the apartment. And interestingly enough, he identified two different air quality problems. One was with a vent that was blowing instead of sucking air. And then there was an issue with the air conditioner where it hadn't been cleaned in a long time and there was paint dust in it from when the painting had occurred. It was supposed to be cleaned, but it wasn't. So we corrected those two issues. I was like, "Okay, great," and then I continued to get sicker. And I was like, "What the hell?" And so I started thinking, "Do I have some serious medical problem? Maybe I have lung cancer? Who knows?" I went to the doctors, get checked out. Finally, I was like, "Could it be mold?" So I had an inspector who specializes in mold come in and I asked them, "Do you think it's mold?" And they're like, "99% chance, it's not mold. I looked over everything. But if you want, we can do this test where we suck the air into a vial and actually measure the amount of mold," and I was like, "Okay, let's do it. I'm going crazy. I've got to figure out what's making me sick." They found the air was full of mold, but there wasn't a single visible sign in the whole apartment even an inspector could find. And so then we knew that there was mold in the air, but we had no idea where it was. So then they literally had to start bashing holes in different parts of the wall, making guesses. And after a few holes bashed, we eventually found it. It was really hidden, growing behind some wood — it was completely not visible inside the wall — about five feet from where I sleep. And I guess that I have an allergy to it of some sort. And I'm not allergic to anything, so it was quite strange. And my partner, I guess, is immune to it. That's why she wasn't having any of the problems. But yeah, so then we got the mold removed. We had to move out for a few weeks. And then I totally recovered and it was totally fine. But that really got me thinking, "Wow, the air quality can be a huge determinant of how healthy you are." And so yeah, now in our apartment, I think we have three different filters running at all times, just in case.

RICHARD: Yeah, I've heard a lot of stories like that. This actually relates to some of the work I did in the Food and Drug Administration on the gluten-free rule. There is a huge variation among people in sensitivities or allergies to different chemicals, a lot more than most people realize. So yes, something like you said: one person can be completely unaffected, somebody else has their life ruined. In general, the more you filter, the more things you have clean, less opportunity for stuff to grow. You never know exactly who is going to benefit. And it's not like everybody's a little bit better; it's some people prevent their life from getting wrecked, and other people may not notice the difference.

SPENCER: Yeah, so now my rule of thumb is, if you have a mysterious illness, definitely investigate whether it could possibly be your home that's suddenly poisoning you in some way. Because I've actually heard of a bunch of other people that similar things have happened to, not necessarily mold also. There are other contaminants: paint dust, it could be gas leaks that can cause crazy things, slow gas leaks over a long time.

RICHARD: Yeah, mold, dust mites, allergies. I was also reading... Watch out for fast fashion kind of things. A lot of the chemicals that they put in fabrics nowadays, some people may not be affected, other people might be allergic to them. So yeah, the stuff that you're buying, anything that is paint or fabric or chemicals or coatings, there's a lot of stuff that can really mess up some people in the population.

SPENCER: You mentioned the FDA, and I know that you worked there. What was your role at the FDA?

RICHARD: Doing cost-benefit analysis. I was hired in the food part and they merged us all a few years after I was in there. So near the end of my time there, I was doing a bit of stuff on tobacco and devices. But mostly I was doing the Regulatory Impact Analysis, which is the special kind of cost-benefit analysis for when the government is putting in a new regulation on FDA Food Policy. There was a gluten-free labeling rule, there's the trans fat ban, and there was a rule on intentional adulteration of food, which is basically terrorists using food to poison people.

SPENCER: Right, because trans fats are now banned. Is that correct?

RICHARD: Correct. Technically, trans fats as a whole are not banned. Partially hydrogenated oils, which are the artificial trans fats that are created in a factory, are banned. But all meat has a little bit of trans fat in it and that's not banned.

SPENCER: Is that just a clear win? I mean, the way that that was sort of circulated in the media was just like, "This is clearly bad for us. It's definitely good to ban it." Is that your view as well?

RICHARD: Yes, it was a clear win. I was asked to look into it and I didn't know much about it. And I was, "Okay, what are the trade-offs?" I was just looking into things. But after I dug into the research, I realized, yeah, this stuff is much more harmful than saturated fat. The reason it took so long to ban is people were concerned, and this is exactly what does happen. If you banned trans fats from the food supply, people are going to use saturated fats instead. And for a long time, saturated fats were considered like, "Oh, this is the horrible thing that you have to avoid at all costs," because people weren't doing the cost-benefit analysis. They weren't looking at the relative amounts. They had this idea of, "Well, we can't do anything that would increase the consumption of saturated fats." And it took a lot of work, a lot of evidence to actually show people, "No, trans fats are even worse than saturated fats." They're like six to ten times as bad. So even if people substitute out and use more saturated fat than trans fats, it's still going to be a huge win. But it took a lot of time to convince people in and outside the agency. You have to have this heightened standard of proof before you make something illegal, so it can survive a lawsuit, etc. So it took a bit longer than maybe it should have because they had to accumulate all of that proof. But yeah, once the proof was accumulated, once we actually learned what was going on with this stuff, it was a clear win.

SPENCER: Now with regard to saturated fats, it seems like the tide has come back a little where people were really afraid of them and now, there seems to be an increasingly large number of people saying, "Well, maybe they're actually okay." Do you have a view on the effects of saturated fats?

RICHARD: Again, with the caveat that I'm not a technical expert here, 100 calories worth of sugar can mess you up. 100 calories of saturated fat can mess you up. The amount at which it messes you up is going to depend a lot on what's going on with you: your diet, your life cycle, etc. So, if you are at risk of heart disease or cardiovascular diseases, then 100 calories of saturated fats will mess you up more than 100 calories of empty carbs. But conversely, if you're much more at risk for obesity or diabetes, and you're consuming a lot of that kind of high glycemic index food, then 100 calories of sugar might mess you up more than 100 calories of saturated fat [sic].

JOSH: Richard accidentally said, "100 calories of empty carbs could be worse than 100 calories of trans fat," when what he meant was, "100 calories of empty carbs could be worse than 100 calories of saturated fats."

RICHARD (continued): So remember before, when I was saying about the linear no-threshold model when it comes to air pollution? Well, diet is going to be a lot more complicated than that. Your body evolved to handle a certain mix of foods. So if you have zero saturated fat, you will die because, at some amount, it's an essential nutrient that your body can't produce. You have to have some proportion of saturated fats, some proportion of unsaturated protein, carbs, etc. Once you start deviating from the ideal proportion, which could be significantly different based on your genetics, your family history, etc... People from different populations, depending on your genes, you'll probably have a different tolerance or a different requirement for different kinds of things, like some people can eat more carbs and not get messed up. Other people will get messed up if they eat fewer amounts of carbs. Depending on what's going on with your body, it could be that the harm starts to increase more in an exponential way as you accumulate more of a particular thing. Once you go beyond the threshold of what you need, it messes you up more. For some people, saturated fat is worse than sugar. For other people, sugar will be worse than sat fat. I think that at the time that saturated fat was demonized, probably most people in the population were in that category of, they're eating way too much saturated fat. Because think about the 60s and the 70s. You didn't have as much of the modern sugary junk foods in the diet, and a lot of people were into meat and potatoes, eating lots of deli meat all the time. So the population then would probably be more harmed by that. So they strip the saturated fat out and people eat a different kind of processed junk food that has sugar instead of sat fat, and now... When people really reduced saturated fat, there were fewer heart attacks, there were fewer strokes. You look at a graph of that, it's gone down a lot. And then diabetes, obesity, that kind of stuff due to too much sugar, goes up. So that was a bit of a long-winded thing but hopefully that gives a better perspective.

SPENCER: Oh, no, that's fine. So the people who go around recommending everyone eat lots of butter and steak, you think they're making a mistake to recommend those kinds of diets?

RICHARD: Yes, basically, any one size fits all dietary requirement is going to be bad. There are some people out there who absolutely should eat more butter and steak. If your diet is 80% potato chips, then eating less potato chips and more butter and steak will absolutely help you. But if your diet is already high in meat and saturated fats, etc, then eating more of that will really mess you up.

SPENCER: Let's take a step back and talk about the FDA more broadly. Now, I think the FDA gets a lot of flack because, on the one hand, if a drug or food ends up being harmful, they get blamed for it. And on the other hand, a lot of people say, "Well, why don't you put things through faster? People are dying because you're blocking stuff. Why won't you let me use the medicine that I want to use? Why do you make it so difficult?" So I'm wondering, where do you fall on the 'the FDA is too fast' or 'the FDA is too slow'? Because obviously they get criticized for both.

RICHARD: Okay, so first, a few clarifications. The food and drug parts of the FDA are totally different. Different legal authorities, different cultures, they're even operated and housed in different buildings. So the food part of the FDA is much more like a beat cop. All food is basically, by default, legal, and the FDA only starts showing up if there are bodies on the ground and they're trying to figure out, okay, who killed these people? A lot of what the food part of FDA does is foodborne illnesses. And there's been a huge success story. If you think, look at how many people were killed by foodborne illnesses now compared to 100 years ago, it's been a huge success and they can basically genetically sequence any time somebody dies of listeria, store that in a database and connect that to their inspection of plants to figure out which particular thing caused this, and shut down outbreaks quickly. So there's a lot of behind-the-scenes, technical things that are working really well that people don't understand. And then you have the nutrition part of food, which is always going to be difficult because food is tied to people's culture and identity. And with something like that, we demand a very large burden of proof before we actually start to ban things, to say it's not generally recognized as safe. And in general, they're trying to solve it as much as possible with nudges and tweaks and slight encouragement and putting out good advice. So that's the food part. I don't think there's been that much serious criticism of them, especially among the more intelligent people. The things that people are talking about the invisible graveyard comes from the drug part of FDA. That's a completely different regulatory regime. All drugs are automatically illegal unless you prove that they are legal for you. And with the drugs, I definitely am sympathetic to the invisible graveyard kind of arguments. That's the idea — I think popularized by the economist Alex Tabarrok of Marginal Revolution (great blog) — that basically talks about, if FDA is slow to approve something that is a good drug, then the people who could have taken that drug after it was invented before it was approved, could have had their lives saved. So depending on some calculations, you can have tens of thousands of people who, if the system had worked perfectly and all good drugs were approved immediately after being invented, those lives could have been saved. And in general, I think that they probably could approve things a bit faster. But there's always the standard problem that you have with bureaucratic incentives where, if you approve something that is found to be harmful, you get in much more trouble from the public, the politicians, the media, than if you take too much time and are slow approving something that's good. So there, they're responding to the incentives of the system that they live in. And there are some tweaks that I would like to make to try to either move things a bit faster, or use more cost-benefit analysis in the procedure. So instead of using more fuzzy and hard-to-define language, just do a cost-benefit analysis of, what's the monetized benefit of approving this? What are the side effects? What are the uncertainties there and have a more mathematical decision-making procedure. I think that it would improve things. Also unbundling the approval, basically having tiers of, "Okay, this is past phase two clinical trials. We are conditionally allowing you to get it from a specialist, but we're not declaring that it's safe and effective." And I would say, you should have even more evidence than they're currently demanding. So right now, it's an all or nothing effect of, once FDA says that it's safe and effective, it becomes legal and it also kind of becomes mandatory as well: Medicare and Medicaid cover it. Private insurances are starting to almost be required to cover it. I would want to have more space in between required and mandatory. So after it finishes phase two clinical trials, it'll say, "Okay, yeah, you can buy it with your own money if a specialist approves it. But we're not going to recommend that any insurance company or Medicaid approve it." Then you have the current approval which, "Okay, we've declared it safe and effective. Insurance companies can do it if they want to." But I would require a fourth level, like stage four clinical trials, even more evidence of safety and efficacy and being a cost-effective drug before Medicare and Medicaid start to cover it. And I would want even a fifth level of approval, even more evidence than that, of it being extremely good before you actually mandated all private insurance companies to cover it.

SPENCER: Okay, so let me make sure I understand. Right now in the current system, phase two is safety, and phase three is efficacy (like does it work). Is that correct?

RICHARD: No. Phase one is it's not going to kill you. You have your animal studies, that's pre-phase one. And phase one is, "Alright, we're giving it to 100 people and looking for any nasty side effects. Phase two is a smaller-scale clinical study that's both safety and efficacy. And if it shows safety and efficacy in phase two, then you have a much larger-scale phase three that basically weeds out false positives that you might get from a smaller sample size in phase two.

SPENCER: That's a little strange. I mean, sorry. Obviously, that's correct. It's a little strange because, with a small study, it's not so much that you expect false positives, though you'd expect false negatives. You expect not to have the power to find real effects.

RICHARD: There's a little bit of both but, anytime you have a small sample size, it's going to increase the variance in both directions. So if you're only testing something on 40 people, you have 20 people taking the placebo and 20 people taking the drug, well, just from random effects alone, you might have a case where five more of the people taking the drug get better, even if the drug doesn't do anything, because of just the individual randomness.

SPENCER: Sure, but if you said (let's say) P equals 0.05 as your threshold for what passes muster, then regardless of the sample size, that means you're gonna have a 5% chance of getting that result, if there's no effect. That's by definition. That's just by the P equals 0.05 cut-off, that's gonna be true at every sample size; whereas as the sample size gets bigger, really, the big effect is that you're going to have fewer and fewer false negatives. You're going to feel the missed effects. Now, I totally agree with you that there do tend to be more false positives for smaller studies for a variety of reasons. One is (I think) that with small amounts of data, it's easier to fudge things or like, if you throw away an outlier, sometimes the result changes, this kind of thing.

RICHARD: Yeah, exactly. Phase three tends to be multicenter, more independent people doing it.

SPENCER: Yeah, I just wonder if a lot of good drugs are missed because the small trial doesn't find anything, but it could be a false negative. But anyways, okay, we've got these three phases. So you're saying phase two is both safety and efficacy. And then phase three is the really big efficacy trial, you would want to say, should really have different levels of evidence, not just, "Okay, after phase three, everyone has to cover it. It can be prescribed by anyone," but adding these extra layers (like phase four) of even more evidence in saying, "Oh, it's actually cost-effective." And then phase five is like, "It's so good, it should be mandatory. Everyone should have to cover it." That's what you're saying, right?

RICHARD: Exactly.

SPENCER: Interesting. Why do you think it is that it flips in such a binary way, from 'nobody can get it' to 'everyone has to cover it'?

RICHARD: I think that just comes down to a quirk in human psychology. Robin Hanson... I don't have nearly enough time to go into more details about this but one of the big ideas that is worth thinking about is, medicine is deeply tied to human psychology, of showing that you're caring about someone, showing that somebody is valued, that everyone in the tribe is being treated the right way, according to the right rituals. And that kind of ritualistic communal thinking tends to infect a lot of medical decisions. Even without that, humans do tend to have a thinking in terms of social categories and all or nothing where this is what we do; if this is good, everybody should do it. If this is bad, nobody should do it. It takes a lot of training in statistics and rational thinking to realize just how valuable it is for there to be a large amount of space between what is forbidden and what is mandatory. Totalitarian kind of thinking — everything is either forbidden or mandatory — is pretty much the default throughout human history, unless you have a specific force pushing against it. I think that's just human psychology. Unless it's specifically trained otherwise, we'll always default to, 'don't let this thing happen unless it gets approval. And once it gets approval, it has our stamp, and it's good, and everyone should do it.'

SPENCER: I think what resonates with me a little more than the way you describe that, but it's probably related is like, people tend to think of things as good or not good. So once it's in the good category, it's like, "Okay, it's good, so we should use it." And if it's not in the good category, why would we use it? It's pretty similar to what you said, but maybe slightly different framing,

RICHARD: Pretty much the same thing, like the warning that binary classification of good or not good is probably the root of what I was talking about, yeah.

SPENCER: And then you mentioned how, if you had the power to impact the FDA and cause it to behave differently, you'd want them to use more cost-benefit analyses. Now, listeners might wonder, "Well, isn't that what they're doing? How could you make decisions if you're a giant, regulatory agency?" So they're not doing that?

RICHARD: Like all laws, their mandate is working on words rather than numbers. Humans, by default, unless they're specially trained, will always try to run everything on words and categories. It's like, "You must approve something if it is safe and effective." Well, what does 'safe' mean? What does 'effective' mean? There's some semi numerical estimates of that. But it really just comes down to a judgment call from experts in the field who have thought about things. It's this very opaque process. It's not mathematical. It's not easy to predict and I would want to change that to be more explicit. Instead of deciding if something is safe or unsafe — nothing is ever perfectly safe — just list up all of the potential side effects and our 90% confidence interval of the probability and severity of those side effects and run a Monte Carlo simulation and just find, what is the expected amount of harm from this? Are you familiar with the concept of a disability-adjusted life year or a DALY?

SPENCER: Yeah, we've talked about that on prior episodes. You want to just give the two-sentence explanation of that?

RICHARD: Sure, yeah. So DALYs or QALYs, quality-adjusted life years, it's basically a way of turning health states into a number from zero to one, where zero is, you wish you were dead, and one is perfect health. A typical member of the population is at like 0.87, and different illnesses will have a QALY or DALY cost for some amount of time. And they come up with these numbers by asking people to compare things like, how much worse is a broken leg compared to depression. And there's big tables of numbers that get produced, of what the numbers are attached to different conditions. And those numbers are not at all used in FDA drug approval processes right now. They don't use it for side effects. They don't use it for benefits, and any cost-benefit analysis. What they should be doing is take all the side effects, find their DALY cost, find the benefit of the drug, its DALY benefit. And you could have different decision theory procedures. I'm agnostic as to which procedure you use, as long as it's known in advance and standardized. It could just be average effects. You might want to say, "Okay, we want to be a bit more conservative. We will look at the 90% confidence interval such that we'll approve something if the 10th percentile of benefit is higher than the 90th percentile of cost," just to make absolutely sure that we're not hurting people with this thing. Different decision-making procedures or medical ethics, there's a lot of different valid things that you could use, as long as we as a society decided on what's the correct decision procedure. But right now, there's nothing so transparent. It's a very opaque process where it's just words and assumptions and people's value judgments, which is hard to replicate and hard to predict.

SPENCER: What would you think of a system where they basically said, as long as the drug is not addictive, then we will allow anyone to buy it, but it will just be labeled like crazy. Like it'll just say 'dangerous' all over it. And there will be a bunch of ways it could hurt you and different rules about how, if you take it with this medicine, then it might kill you, etc. Do you think that would just be chaos or do you think that actually might be better than the current system?

RICHARD: That would be heavily dependent on the culture of people involved. I can definitely imagine a certain society or a certain culture for which that system would be much better than what we have now. And I can also imagine a society and a culture with which that system would fail horribly. Again, I'm very sympathetic towards the rights-based libertarian approach to things. But an important concept that is worth keeping in mind is, sometimes social forces can turn a right into an obligation. A clear example of this is the right to duel another gentleman who has insulted your honor. That turns into something that you're forced to do and kills you because of social pressure saying, "If you don't do this thing, then we'll feel bad." So Hamilton winds up dying because of this right that he has that, in theory, he could have refused, in theory, was completely optional, but he just got forced into doing it. And given the psychological quirks around medicine, oftentimes, if people are desperate and not thinking straight, sometimes there's this desire to... If a relative of yours has some sort of rare thing where other treatments aren't working, then you might say, "Yeah, I'm going to spend a lot of money for this thing that has all the warning labels. But yeah, we'll do it anyway." And you can easily cause more harm than good because you're, in a poker term, 'on tilt.' It depends a lot on the cultural norms, the societies, the values, but I'm pretty sympathetic to the argument that the system that you described could have a lot of bad side effects just because in high stakes medical situations, people are often on tilt and not thinking straight. And the system that takes away this probably harmful option is probably going to be best overall.

SPENCER: What about medicines that are quite safe? Because there's lots of pretty safe medicines that still can be hard to get. One that just personally I've seen cause a lot of harm is people who are really depressed and they're on antidepressants. They're pretty safe. Okay, maybe there's some mild side effects like sexual side effects, but they're unlikely to really hurt you. And they forget to get their prescription and then they don't take their dosage, and then they're depressed so they don't make an appointment with the doctor, and it all completely spirals out of control. I've just seen that kind of thing happen a lot, because they're just so annoying to get.

RICHARD: I don't know enough about that topic to make a prediction about what would happen if those kinds of antidepressants were over the counter and easy to get. Perhaps the solution is just powerful antidepressants being sold over the counter; perhaps the optimal solution is, no, it should be a lot easier to get a doctor to write a prescription for it. The evidence could go either way. And my prior would be, yeah, it's probably that drugs should be easier to get. You look at other countries, for example, lots of kinds of over-the-counter birth control. They can get things over the counter that require a prescription here and nothing bad happens. I am very sympathetic and I would lean towards the idea that, yes, certain kinds of antidepressants probably should be available over the counter, definitely if you've been prescribed them in the past. I might be a little hesitant because the counter-argument there is kids being prescribed powerful amphetamines as basically a performance enhancer in school, which seems wrong to me. And I would worry that, if powerful psychoactive compounds were available without any kind of prescription, then they could easily wind up being overprescribed and cause a lot of screwiness and more harm than good in certain situations, if there was some other cultural pressure or some other perceived benefit of using them.

SPENCER: An interesting example here is supplements, where they have way fewer rules around selling them, it seems. And you see a lot of wacky uses of supplements. You see all kinds of hucksters selling their supplement combo whatever. You have lots and lots of supplements that we just know are placebos; they don't do anything. You have some even that are known to cause significant side effects and even can be dangerous at times, that get sold. So you can see the dangers of the free-for-all in the supplement world already. And if you applied it to medicines that could potentially be more dangerous, then you could see lots of problems there.

RICHARD: Absolutely, yes. I think most people who use supplements are doing themselves more harm than good. Even more people are just flat out wasting their money. A lot of these dietary supplements are being used when there's no reason for them. People who eat a healthy diet are doing more things because the label tells them to. But on the other hand, because of human variation, some people, due to their particular circumstances, some people are very smart about self-experimentation or using nootropics properly. A lot of supplements do cover for problems of bad diet. So I definitely would not want to ban or regulate supplements any more than they already are, except for a few cases that seem to be particularly egregious. Yeah, so everything you said is true. I generally favor chaos and experimentation and let people play around with things unless we know it's irreversible or particularly harmful. But yeah, what you said is entirely correct. Even though most people are doing themselves more harm than good, I wouldn't want to regulate them more. And in fact, I would want to move drugs a bit more because I think in the long run, there's a social learning or development or exploration that goes on. But I guess to steel man the argument of keeping things more under wraps in FDA, you have the supplement industry and it doesn't seem like science ever advances. It seems like everything is driven by cycles of hype. There's a fad that comes in. There's never any cultural knowledge of how to use things more responsibly and it just seems like a churn that accomplishes nothing. And the response to the invisible graveyard people is saying, "You're assuming that the system works perfectly, and that all of this money and investment is put into developing the good drugs." If there was nobody that gave an official stamp of approval of things, if you didn't have the government saying, "Okay, it's approved, now insurance has to cover it," maybe all drugs would look like the supplement world and there would never actually be any incentive to do the proper research and development to make new things. So it's entirely possible that if you screwed around with the system too much, you might actually break the engine of progress that is required to make things happen. It's plausible. I don't think that's the most likely outcome, but it's definitely a possibility that the FDA being very slow and bureaucratic and methodical is part of the secret sauce that allows the scientific process and the investment, the overall engine to actually happen.

SPENCER: That's a really interesting point. One thing I've observed is just how easy it is for many people to believe that something that's a placebo works for them. You just see this constantly where someone tries some new thing, and they start feeling a bit better. They attribute it to the thing, and then they take it for the next ten years [laughs] even though the way we feel on a day-to-day or week-to-week basis can fluctuate so much. I think that, if you just happen to get one of those upswings right around when you started taking a new supplement, it's just very easy to attribute it. And then, going back to labeling things as good or not, you start labeling that supplement as good. And then you're afraid to stop taking it because you attributed positive things to it.

RICHARD: That's exactly right. And that's why we need the clinical trials to actually show, okay, this thing really works.

SPENCER: Funnily enough, in my own life, I am fairly convinced that most things will not work for me. [laughs] I've had people make fun of me for this, my incredible skepticism towards so many things. Someone will be like, "Oh, do you want some cough medicine?" I'm like, "Okay, sure. I doubt it will actually help my cough, but sure." And I feel like I barely experience a placebo effect because of that, because I come in just believing that the chance that thing helps me is quite low.

RICHARD: Yeah, I think there's a lot of natural human variation, some genetic, some cultural, in how much people respond to the placebo effect. Some people really respond to it. Some people don't.

SPENCER: Yeah, maybe another factor for me is I'm constantly running self-experiments. I try to always have a new self-experiment going at all times. So I've just run such a ridiculously huge number of self-experiments, it's just really clear to me that mostly, they fail.

RICHARD: Nice.

[promo]

SPENCER: Another thing I wanted to ask you about is side effect labeling, because it just seems totally bonkers to me. If you look at side effects for medicine, it just lists like 20 things ranging from 'your head will explode and blood will pour out of your ears,' to 'you'll be slightly tired.' And there's no indication usually of how common these are. Is this more common than it is on a placebo? And I don't understand why we don't have a more sane side effect labeling system.

RICHARD: Yeah, you're right. They're not thinking about the value of information. It's not very numerate; although I'm pretty sure that something only gets listed as a side effect if it is statistically more common than the placebo group. There should be effect sizes, there should be severity. And a big problem with all these kinds of warning labels is people don't think in terms of the value of information. They don't think of the costs they're imposing on people by reading it. Yeah, I agree. It should be much more rationed to only be the most important things that matter.

SPENCER: But I've even seen side effect lists that say some of these were no stronger than placebo. I'm like, "Why are they listed here?" It's some ass-saving maneuver, I assume. But I find it almost useless to look at side effect lists because you just don't know if this is like a one-in-a-million thing or a one-in-five thing.

RICHARD: Yeah, I guess I just betrayed my ignorance there. Yeah, I might have been wrong about it only getting shown up on the list. In that case, yeah, it's even more screwed up than I thought.

SPENCER: Well, I don't know. Maybe it varies in different circumstances. Let's go meta for a second. The FDA is a big bureaucracy. I think you have an interesting way of looking at bureaucracies like it. You want to give us a sense of how you think about that?

RICHARD: Sure. Basically, a bureaucracy is a machine designed for producing repetitive action. It's basically a factory and you think about the history of civilization, the industrial revolution, lots of good things in life come from factories, and civilization very much functions on procedure and precedent and predictability and law and order. So a system that's based on, "Well, we'll just hire this fine strapping young man and have him do what he thinks is smart," rarely works. It usually has all kinds of bad side effects. The antidote to that is, you hire people who are very much used to routine, you have a big rulebook, you're acculturated to do things a certain way, and you basically try to produce a system such that people aren't corrupt, people aren't self-interested and, probably, they'll produce the same output given the same set of inputs. You want reliability. That's a really valuable and impressive social technology. You get a building full of hundreds of drug reviewers, and they're pretty much going to ideally produce similar outcomes, which is what you would want. You want procedural fairness in that thing. So it's a very powerful factory for procedural fairness. But of course, it's going to have the predictable side effects of, if you're training a lot of people — specifically hiring and training them to think the same way and in a repetitive way — that system is not going to be good at dealing with novelty. So you have this system that's designed to grind out things in a normal procedure — we're going to do the drug approval, and it's going to be fair and done the right way — you have to almost iron out individual initiative from the system because you want predictability and fairness. That system is not going to function well when things change, or if the rules they were given weren't quite right, or if it's not exactly aligned for what we want. The value and the peril of them should be understood in that way. We have a government that's a factory of predictable actions. There's good and bad of that, but I think it's a useful mental framework. Don't ask, "Why did they screw up?" Just be happy that we have this factory of things that can produce stuff that our society needs on a relatively reliable and predictable basis.

SPENCER: It seems like people are very often infuriated at bureaucracies. If they're inside the bureaucracy, it can feel really stifling. They can't take initiative. They have to follow stupid procedures that make no sense. And then if you're outside the bureaucracy, you have to deal with it. It often feels like it's just trying to prevent you from doing anything. It just gives you all these hoops to jump through that prevent you from taking behaviors you want to take. So you can see why people are annoyed by bureaucracies. But how do you think about those sorts of downsides?

RICHARD: Yeah, again, you have to set a machine where the machine should be used. And I think what people don't appreciate is, your view of the world is going to be unique. Most people assume that, well, the world is the way I see it, and other people need to get with the program. There's a huge variation. Everyone has vastly different ideas about what should happen. It's one of those situations where you think how much harm would be caused if someone with a vastly different world model than mine, who had vastly different ideas about things, started to take initiative and do whatever they wanted. There's certain circumstances like, obviously, in the tech startup world, if you're doing something that doesn't have negative side effects on society as a whole, letting 1000 flowers bloom is the right kind of call. But if something is related to government, or if it has a lot of power, then letting people take initiative, letting things happen at random, is a recipe for corruption, chaos, unpredictability, nobody knows what's going to happen from the legal system, etc. So, in general, as much as people complain about bureaucracies, I think they would complain about legal system and tort reform more. You don't often hear this out loud but, in general, it's much better to run decisions through a bureaucracy than a court system that does allow people more individual initiative, that allows more decision making, because it's just so unpredictable.

SPENCER: By 'court system,' you're referring to a judge that gets to make a decision, that kind of thing?

RICHARD: Yeah, judge and jury.

SPENCER: Right, as opposed to bureaucracy. So what do you do if you're stuck in one of these systems? You worked in a giant bureaucracy.

RICHARD: Yes. So what do you do if you find yourself in a bureaucracy and you want to make the world a better place, as good as possible? As I said, a bureaucracy is a machine that programs people to act in a certain way. And if you find yourself in that situation, it's important that you understand what the rules are. You have to know what is right and expected and know why the rules are happening. But you have to avoid getting programmed by them so that you think of the rules as the only thing that could happen. You have to maintain the capacity for individual initiative. And it's more apparent if you're in a bureaucracy, but this general effect occurs for anyone in any society. Your view of what reality is, is heavily culturally programmed. You grow up a certain way, you learn things in school, you get a certain worldview from your society, you're hired into a bureaucracy, that process happens where you just absorb the bureaucratic worldview. And a way that you can guard yourself against that is techniques of rationality, but also techniques of insight meditation, training yourself, whenever your brain generates a conclusion, to just automatically be mindful of, where did that conclusion come from? Why did I believe this? And you'll realize that you might have been trained a certain way to automatically assume that a model of the world is true, to automatically have a certain stimulus response. So if you train yourself in mindfulness, you generate a pause between stimulus and response. And in that pause, you have the freedom to actually think about what's good and what's right, and what needs to happen.

SPENCER: Did you think when we're in systems where everyone's told to do things a certain way, and you see everyone else doing it that way, that on the inside what it feels like is just that that is the way things work, rather than seeing that that is a completely human-created system?

RICHARD: Yes, exactly. We all have more layers of programming than we have any idea of, and most people who think that they are free thinkers are just running with the free thinker programming on some aspects of their life and not others. It sounds goofy at first. It's like you're training yourself to follow your breath, count to ten, feel the sensations in your body, but you're basically building up the practice of noting what's going on. And then you train that attention inside your own mind, and you'll quickly start to realize that the things that your brain assumes as an automatic truth is actually just artificially generated, and you just keep training yourself that for more and more things, so you see how much delusion that you're wrapped in and you start to be able to use the rules instead of the rules using you. And you can switch to different worldviews. You can try different models of reality on, like putting on a different suit of clothing instead of just automatically identifying with whichever one you were using in the past.

SPENCER: Can you give an example from your own life where you took a moment to pause, notice, "Oh, wait, the reason I think this thing is this other thing," and then it helped you step back and realize that there was some kind of programming going on, rather than coming to a conclusion about the world yourself?

RICHARD: Yeah, the first thing that comes to mind has nothing to do about my job. But it was pretty notable. I had basically taken a bunch of vacation pictures of me going on a date with my girlfriend. And she was like, "Hey, send me the pictures. I want to see them." And I had somehow gotten in my mind that I should only send the best pictures and edit them so they were looking better. So I was going through this pile of 100 pictures of carefully going through things, and I was getting upset at her for asking me to do all of this work to send the pictures. And I thought, "Wait a minute. She never actually asked me to do the work. That was just an assumption that I made of what I should do before I send the pictures out. Why don't I just send her the entire data dump of 100 raw pictures?" And she was perfectly happy with that. And that's kind of how people in a bureaucratic procedure think of, "Oh, well, you have to do this thing in order to make it right," without thinking, "Why am I actually doing this?"

SPENCER: That's a neat example, I think there's an interesting mode you can get in when there's a set of rules you have to follow. Let's say you work in a bureaucratic organization, and there's a whole bunch of rules, where you get in the mindset of like, "Okay, I really want to accomplish this goal, because I think it's important and meaningful. I have this set of rules I have to follow." And now you start trying to say, "Well, how do I get to the goal as efficiently as possible while not violating the rules?" And sometimes I think it can lead to surprising insight that there's a way to do things better, even though you're constrained by the rules, if you view the rules, not as reality, but you view the rules as just optimization constraints.

RICHARD: Yes, that's correct, with the caveat that probably 90% of the people who view rules as optimization constraints are going to be selfish or manipulative and trying to tweak the system for their own end. It's actually quite rare to have someone who is pro-social, mission-oriented, is trying to do the coherent extrapolated volition of the bureaucracy, but also is tweaking and hacking and using the system to work for it. That's a rare combination of things if you can do it. And of course, bureaucracies are specifically designed to not let people think like that because most people who think like that are just selfish, looking for a way to manipulate the system for personal gain.

SPENCER: Yeah, it's a fair point. I think entrepreneurs tend to be better than average at this. For example, one entrepreneur that I know, when he was starting his company, he had been advised that there was a rule about credit cards that had to be followed. And then he actually went to the literal law, studied it carefully, and realized you could actually allow people to open up a credit card in ways that were different than what everyone was doing, because everyone made assumptions about what the law actually said. And that opened the doors for his company to do things differently in a better way. So I think there's just a kind of entrepreneurial skill set that's like, "Okay, the law says this. We're trying to accomplish this goal. We know the goal is good and meaningful. How do we follow the law but not let it hamper what we're trying to get to?"

RICHARD: Yeah, and that kind of attitude is very hard for people to develop without mentoring. There's some things you can do — strategy games, occasionally — but yeah, in general, most people learn things by watching the behavior of others, which is why hanging around other entrepreneurs, people who do patents, learning, seeing examples and case studies is how you learn. And if you haven't been exposed to that culture, it's just very hard to pick up.

SPENCER: Richard, thanks so much for coming on. This was a really fascinating conversation.

RICHARD: Awesome, glad you enjoyed it. I had fun.

[outro]

JOSH: A listener asks: "Our definitions of kindness seem kind of fuzzy. Sometimes we are trying to be kind and we end up hurting somebody. Or sometimes what seems to be the right thing to do isn't to be kind. What is a good definition of kindness?"

SPENCER: Well, as with any word, we can always define it however we want. But not all definitions are equally useful. Like if we define kindness to be a cat in a certain position, that's not gonna be a very good definition. It's gonna confuse other people when you use it. It's not gonna be very useful. Keeping in mind that definitions are inherently fuzzy and they will always be fuzzy, I think that it's useful to distinguish between kindness, which is acting in a generous and compassionate way, from morality, which is acting in a (quote unquote) 'good way,' a moral way. And those are not the same thing. Sometimes you can be kind, but actually, it can be immoral. For example, you could be kind by helping someone who's trying to do bad things in the world. You might be being kind to them, but you're causing harm in the world. There are other times where being unkind is the morally correct thing. For example, maybe you won't do what someone wants, even though they really, really want it, and that will make them feel really bad. But you won't do it because you know the second order consequences of that thing would be harmful to the world. So while being kind is often aligned with being morally good — most of the time being kind is the morally good thing — there are edge cases where it's not and so the two concepts break apart.