The economics of putting germicidal light in every room, with Misha Gurevich and Vivian Belenky of Aerolamp
In this episode, Patrick McKenzie (patio11) is joined by Misha Gurevich, CEO of Aerolamp, and Vivian Belenky, its chief scientist and a researcher at Columbia, to discuss why UVC, a wavelength of light that inactivates airborne pathogens but is absorbed harmlessly by the dead outer layer of human skin, remains almost entirely undeployed. They explain the physics and safety case for 222 nanometers, the economics of installation, and preliminary South African trial results showing 90% suppression of tuberculosis transmission in hospital wards.
Presenting Sponsors: Mercury & MongoDB
Complex Systems is presented by Mercury—radically better banking for founders. Mercury's new feature Command brings an LLM directly into your banking interface, so checking balances, finding invoices, or sending a wire is as easy as asking. Apply online in minutes at https://mercury.com/.
What's the point of building faster with AI if your database can't keep up? MongoDB's native data model mirrors the language LLMs already speak. Ship at the speed of AI while staying ACID compliant at Fortune 500 scale. Start building at https://mongodb.com/ai.
Timestamps:
(00:00) Intro
(00:58) What far UVC does to pathogens
(03:25) Why it's safe for skin and eyes
(09:24) What a typical deployment looks like
(11:21) Why isn't this everywhere already?
(13:55) Where the rollout should be prioritized
(17:31) The economics per room and per building
(20:56) Lamp lifetime and maintenance
(21:48) When does the evidence become undeniable?
(23:24) Tuberculosis and long-term care as early proof points
(27:16) Pandemic math and kneecapping the next outbreak
(30:16) Sponsors: Mercury | MongoDB
(32:49) R0, K-factor, and why the built environment beats opt-in
(34:23) Building codes, ASHRAE 241, and competing technologies
(39:11) The deployment curve and the awareness problem
(42:33) Scaling manufacturing: krypton chloride versus LEDs
(45:14) Price versus cost, and where the inflection point is
(47:35) Should you put one in your house?
(51:51) What would make this fail?
(56:09) The hygiene hypothesis objection
(01:01:13) The precautionary principle and reasoning under uncertainty
(01:06:04) Where the first million dollars should go
(01:10:49) The business case and the heckler's veto
(01:14:51) Where to find Aerolamp
(01:15:52) Wrap
Transcript
Patrick: Welcome to Complex Systems, where we discuss the technical, organizational, and human factors underpinning why the world works the way it does.
Patrick: Hideho everyone, my name is Patrick McKenzie, better known as patio11 on the internet. As longtime listeners of Complex Systems will know, I think that far UVC is one of the sleeper picks for among the most important technologies getting developed today. [Patrick notes: See previous discussions with people from 1Day Sooner and Blueprint Biosecurity.]
Far UVC is a wavelength of light that can deactivate viruses and other pathogens, and it is possible that we will be able to introduce this into our built environment, into our homes and offices, via specially made lamps for infection control.
I'm honored to be joined today by Misha, who is the CEO of AeroLamp, and Vivian Belenky, who is the chief scientist of AeroLamp and also a researcher at Columbia University. Thanks very much for coming on the program, guys.
Misha: Yeah. Good to meet you.
What far UVC does to pathogens
Patrick: Good to meet you as well. So just for folks who haven't heard the far UVC gospel yet, can we talk briefly about what this wavelength of light actually does, both on the sort of chemistry/biology level and hopefully on the social/technology level?
Vivian: So on a chemical, biological level, the special thing about far UVC is it is absorbed by the DNA and RNA of pathogens, as well as by essentially all proteins. And this is very important, because the fact that it's so heavily absorbed by proteins is a thing that makes it much safer than other germicidal UV wavelengths.
So essentially, a pathogen — or any microorganism, anything without significant protections like humans and animals — is going to be inactivated. So it's still in the air, but it can no longer replicate, and this works extremely fast. So it's essentially like having an extremely strong air purifier running in a space. Except instead of maybe giving you an extra air change or two in your space, you can get the equivalent of 30 to 50.
[Patrick notes: In infectious disease control, the unit of effectiveness is turnovers of a room’s air supply per hour, originally mostly provided by better ventilation. Some handy calibration numbers: a typically insulated house gets about 0.5 an hour, mostly through passively leaking. An office HVAC system might deliver 1.5. Aircraft and hospitals are somewhere in the 10-20 region, after heroic work on ventilation. The CDC’s post-covid target for the entire built environment is 5, and it will be a long slog to get there via HVAC improvements alone. In particular, note that if you want to move 20X the volume of a room worth of air in from the outside in an hour, you need to heat or cool that amount of air, which makes external ventilation non-trivial in much of the U.S. much of the year. So you get most of it via recirculation of air, which requires some filtration or treatment step. And thus, back to UVC.]
Patrick: And this is just a wavelength of light which happens to be invisible, so it's something that we get in sunlight already, presumably, right?
Vivian: No, actually. So sunlight is primarily UVA and UVB. UVB in particular is what we're most worried about when we're worrying about risk of skin cancer and cataracts. But UVC is actually completely blocked by the ozone layer, and this makes sense, because UVC is quite efficient at inactivating microbial life. Probably the surface of our planet would look very different if it was present here on Earth.
So there is no UVC in sunlight. Sunlight is germicidal just like UVC, but to a much lesser extent. So it's true that while sunlight does kill germs, it does it because there's just so, so, so much more sunlight than there is ever from any UVC lamp. That even though sunlight is only mildly germicidal, you know, per photon, it can still do the job.
Why it's safe for skin and eyes
Patrick: The legal beagle in me has to say that Oliver Wendell Holmes was empirically disproven by the science. [Patrick notes: D’oh, I’ve been misquoting that for a while. Louis Brandeis, not Oliver Wendell Holmes, and he wrote it before he was on the Court: "Publicity is justly commended as a remedy for social and industrial diseases. Sunlight is said to be the best of disinfectants; electric light the most efficient policeman." That's from "What Publicity Can Do," Harper's Weekly, December 20th, 1913, later collected as Chapter V of Other People's Money and How the Bankers Use It. Hat tip to Claude for the correction.]
Sunlight is in fact not the best disinfectant. But okay. So this is an interesting wavelength of light. I personally have done a little bit of the reading and am relatively well-informed by the safety story, but I think that the typical member of the audience probably isn't.
So what is the sort of chemical/biological reason that this is safe for us? You've mentioned that it gets blocked by basically any proteins, so proteins in our skin and epidermis layer presumably, but how is it safe to, for example, look at?
Vivian: Right. So this is actually much more of a mechanical story than a chemical or biological story.
Patrick: Okay.
Vivian: Essentially, it's not good for living cells to be exposed to UVC of any wavelength. But the difference is that humans have a 20-micron-thick layer of dead skin cells that are chock-full of proteins that essentially absorb all far UVC. And I should say this is unique to far UVC, the wavelengths of 200 to 235 nanometers, principally 222 nanometers, which is what is most commercially viable right now.
[Patrick notes: I think of this as being spiritually similar to various forms of radiation. Alpha particles are not a major radiation concern unless you swallow the emitter, because they’re stopped by a sheet of paper or your skin. Gamma rays are much more concerning and you need much thicker shielding to protect from them. I presume that people talking to the general public about UVC safety will probably not use this analogy frequently because radiation freaks people out, but just between us geeks, the entire earth’s surface gets exposed to radiation every 24 hours and it is a good thing, too.]
Longer UVC wavelengths — 254 nanometers, 265 nanometers, which are used in water disinfection — these do not have so significant protein absorption. So they are — I'm not gonna say they'll give you cancer. Relative to UVB, they're thought to be less carcinogenic, but they are not pleasant to be exposed to.
So it's really something that is unique to the shorter wavelengths in the far UVC. So this high protein absorbance, it is due to the stratum corneum, the outer layer of the skin. It absorbs almost everything, and what is not absorbed is absorbed in only the very upper layers of the skin. And those skin layers will generally slough off and become part of the stratum corneum within, you know, typically a couple of days.
So I've never seen a study that showed any significant biological activity down at the basal skin cell layer, where you would also worry about cancer if there was any damage to the DNA there.
Now, the story with the eyes is a little bit more complicated, and this is because there isn't a convenient dead skin cell layer — or dead eye cell layer, whatever that might mean — to protect us. But eyes are protected in the same way that they are protected from sunlight. We have eyelids, eyelashes, eyebrows, brow ridge.
[Patrick notes: If you want to put a number on it, Zamudio Díaz et al did. They stuck UV-sensitive film on manikin heads under a ceiling-mounted fixture and found the eyes received about 5.8% of the peak dose landing anywhere on the head. The ranking, from most exposed to least, was top of the head, bridge of the nose, upper lip, and then the eyes, in every position they tested. As Vivian says, you should sort of expect this: think of where you feel discomfort after a day in the sun, and how infrequently it is your eyes (unless, I suppose, you were skiing without protective eyewear, because being in a maximally reflective environment might get you past the SPF of your eyebrows).]
And all of these things reduce the effective dose to the eye. So this does mean that, despite these mechanical protections, the eye is more vulnerable. There is the tear layer, which has some lipids and proteins, and that absorbs some of the far UVC, but actually only about 15%.
The rest of the incident dose would be absorbed in the epithelium, and this is a — so a relatively lower dose might produce eye pain or discomfort. The positive story there, though, is that this is still — even though these are living cells, they're still full of proteins, and the absorption is thought to essentially totally stop after the first few cell layers.
So any chance of long-term damage is, in my opinion, fairly low, although we haven't — this technology hasn't been around for long enough to do extremely long-running studies. There have been a few long-run eye safety studies. Actually, there's a one-year one and a three-year one going on in, I think, Japan.
[Patrick notes: Shoutout to Shimane University Faculty of Medicine, Department of Ophthalmology and Sugihara et al. Three year follow-up on the one-year cohort coming out soon.]
But it does mean that the safe effective eye dose is lower than the safe effective skin dose. We're still figuring out exactly what that is and exactly how to translate that into industry practice. We are generally pretty conservative, but the good news is that it's sort of like looking into a bright light, and not like looking into an infrared laser where, if you feel any pain, that means that you flinch away and then you're not really worrying about long-term damage. And this is very different from, say, infrared lasers where you could get a blinding dose and feel nothing. So it is thankfully nothing at all like that.
Patrick: Yeah. If anyone has ever been in a lab that has lasers, the safety briefings are, one, terrifying, and two, if you feel anything, it's far too late.
Vivian: Yes, yes. With far UV, if you feel something, stop looking at it. And just to clarify, it's not where you might be kind of just ambiently sitting in the room and suddenly your eye starts hurting. It's more like, did you accidentally climb up and stare right into it for a couple of minutes without turning it off? So we're talking more about accidents of that nature.
Misha: One of the big benefits of it being more effective than sunlight as a disinfectant is that you can actually use it at pretty low power levels. So the emitters we're using, they're putting out like 100 milliwatts, right? Which is really not a lot. So over the course — like, over a distance of a room, it basically decreases to around zero, especially if you're further and further away, right?
Vivian: Yes. So you need very little dose of this stuff to get a pretty rapid germicidal effect.
What a typical deployment looks like
Patrick: So we're starting to see these emitters in the corners of rooms in, let's say, tech-forward places in the San Francisco Bay Area. I see your branded emitters more than most. But can you describe what the typical deployment of this would look like?
Vivian: So for now, we are doing these corner-mounted units, mostly just because it's a little bit more cost effective to mount them in the corner. This is because these lamps have a fairly narrow beam angle, and that means that to maximize the average dose in the room, you want to maximize the path length of that beam. Which means usually you mount it in a corner, and you point it to the opposite corner, and that gets you a larger average dose over the space.
In the future, I think ultimately we're looking at just kind of normal overhead ceiling lights. You know, just a boring ceiling fixture, like a smoke alarm or any number of random pieces of building infrastructure that are in drop ceilings in offices, hospitals, schools.
And yeah, so we're doing it right now this way just 'cause it's a little bit easier to install. We wanted to emphasize that, hey, this is something that you could just buy for your space and get on with. And Misha can talk a little bit more about that.
Misha: Yeah. I basically sort of see the current lamps we're sending out as a very important existence proof that the technology is fundamentally ready to go. There's not any insurmountable technical difficulties. There's not any insurmountable logistical difficulties to deploying it. There are still many difficulties, as there are in running any business, particularly any business that involves high tech components.
But it's not something that's like, you know, you can only get this for $10,000, or you can only get this if you're a secret government lab. This is something that's pretty much ready to be deployed.
Why isn't this everywhere already?
Patrick: Mm-hmm. One of the things that I like about this is that, as you mentioned, the tech is basically proven. We've done extensive lab studies about this, et cetera, et cetera, and it is something that, bluntly, fits on a shipping container from China. And one of my theories about the world is that everything that fits on a shipping container from — without loss of generality — China craters in price over time, particularly as you scale up production a bit.
And so while many of our other medically oriented interventions are consistently high priced due to Baumol's cost disease and other reasons, this is something that we should eventually be able to buy for not much more than the price of lighting. And there's very few people or institutions that go, "You know, darn, we're building out a hospital, but we just can't afford the lights in it."
And so it's largely a matter of will, I think, for deploying it, although you've probably had this conversation many more times than I have. Why does it not exist already in all of the hospitals?
Vivian: I think you completely hit the nail on the head that it's just a matter of will, and it's a matter of people even knowing that this is a thing that they might do.
So right now, I think far UVC is sort of in this category of things that only weird people might get or think about. And it's like wearing a respirator everywhere, and what it needs to be is more like having hand sanitizer stations around, or having just normal ventilation in your building.
So I think there is, to some extent, we already know that people care about not getting sick, but not getting sick in this particular way with this particular product category. I think it's some combination of social normalization of, you know, we should be treating our air at all, and knowing that there is a way to do it cheaply and effectively. And yeah, as you said, will.
I got started in this field looking at what are the bottlenecks to deployment. Like, is it a cost barrier? Is it a research barrier? And I think it's not that there's no room to do more research. I think there is definitely more room to do research and refine the recommended practice, best practices. But for the most part, it's just, yeah, there's not really any good reason that it can't be everywhere.
Where the rollout should be prioritized
Patrick: I think when we look at the history of improvements in sanitization practices, sanitizing water is helpful in one way in that it isn't in fact the case that there is a single choke point in the water delivery system, but it is largely centralized in the typical deployments, and so you only need to convince one organization.
When we were trying to convince doctors to wash their hands before surgery, that was an intensely difficult conversation, which is still ongoing in some places, because you have to convince all the doctors to do it consistently every single time. This is sort of a one-time intervention, but it's a room-by-room intervention, and we don't necessarily need it in all the rooms, but we do need it in many of the rooms in areas that people congregate.
Can you talk about what the research suggests with respect to, as we're prioritizing the rollout, where should we prioritize?
Vivian: Right. So this is a tough one. If I had to say where do I think there would be the most benefit, I would say elementary schools, maybe middle schools, because they are often extremely poorly ventilated, and children are immunologically naive.
So I mean, one issue with doing airborne disease transmission studies is that you're using healthy adult volunteers. Turns out that healthy adults just generally don't get the flu. You kind of need a lot of shots on goal to successfully catch the flu. Children are kind of a different story.
It is just a bit trickier, though, in that I am convinced on the strength of the safety evidence. I don't know that we have a large enough pile of safety evidence that a large enough fraction of parents of children would be thrilled to try this relatively new technology in schools. I do expect schools to be relatively conservative, perhaps with the earlier rollouts happening in private or specialty schools.
So I think ultimately occupational settings of some sort are gonna see it earlier, even though I don't necessarily think that's where the very largest benefit is. I think long-term care centers and hospital waiting rooms are big ones where the benefit is high. So any concern about photobiological risk is going to be more acceptable.
Patrick: Presumptively places that have a relatively transient population. So in a hospital bedroom, you might have a patient there for a relatively long time. But presumptively, people do not spend plural days in the hospital waiting room. And so if you have some worry that there's some dose level that would be potentially inimical to a person, they are less likely to get that dose level in the waiting room than they are in other spaces.
Conversely, when you're not counting an individual human's dwell time in a space, the aggregate amount of dwell time in the waiting room is quite substantial. It's just spread over hundreds of patients.
So I do think that it is likely to happen in private schools maybe faster than public schools. And one reason is that when I do the back of the envelope numbers for it, it's something that a single parent could probably fund just by deciding to do it if they get the school on board. And indeed, a few of my tech friends have discussed potentially doing that for the schools their children attend.
[Patrick notes: Full disclosure: I haven’t spent my relationship points on this, but have considered it. My back-of-envelope-math was on order of $10k for a fairly typical Chicago neighborhood Catholic school with a few hundred students. Many such schools stay open in material part because the neighborhood has many families who write similarly sized checks every year.]
The economics per room and per building
Patrick: Misha, can you talk a little bit about the economics of this, like on a per room and per institution basis?
Misha: Yeah. So we usually say something like 250 square feet per lamp is like how much coverage you would get. So for a standard classroom — so this is sort of, I guess a standard classroom is a tough question, because there's like the standard classrooms as they actually exist. And if you look into, say, California guidelines for a standard classroom, they're actually much bigger than standard classrooms tend to be.
But that would be like two to three lamps per classroom, maybe four for a bigger one. And so for a lot of schools, you're looking into a lot of lamps, right? You know, dozens. For really big institutions, like if you're talking a university, you're probably looking at hundreds of lamps.
I also think — so universities are kind of a front runner, 'cause again, because it's not children, it's older students, it's a little easier to get installs set up. One of our competitors recently set up a big installation in a Florida university. I haven't heard any details yet, but they're apparently collecting a lot of good data that way.
But so this is sort of the kind of thing where on most small buildings, you're looking to spend maybe four to five figures if you wanna outfit the whole building. And then once you're getting bigger, you're looking to spend maybe six figures purely just on lamps, and then installation also. So usually we say installation will cost you probably about what the lamps will cost you if you're hiring professionals to do it. If you're just doing it yourself, installation is basically free, right? It takes like 10 minutes to stick one on the wall and plug it into an outlet. But if you're doing permanent installations, you're gonna wanna run wiring through the ceiling, that kind of thing, right?
Vivian: Yeah. So budget as much for the installation as for the lamp itself. That's kind of just like a rule of thumb, on average across all possible electrical systems and ceiling types. Even if this were happening on a mass level and we were hiring professionals to do all of it, I think for many typical scenarios, it could be quite a bit cheaper. But just on average, when we're talking about mass social modeling, I think roughly double the cost for installation is a reasonable conservative estimate.
Patrick: And so at the moment, it's something like $500 a lamp. So if you need two of them for a room, that's $1,000, and then $1,000 for installation, and then multiply by number of rooms that are at top of your priority list.
And useful to point out that this is something electricians are very qualified to do already. You just say, "Hey, there is a weird light fixture that goes in the corner." And they say, "Okay, I have done light fixtures before." That isn't an unsolved problem in material science.
Misha: One of the reasons that UV has not really been adopted more generally, even though it's sort of been a known technology since the forties and fifties, is that older wavelengths are more dangerous, so they have to be installed a lot more carefully, right? Like you need expert installation. You can install UV in the upper part of a room where it's safe for people, but if they mess up an upper-room install, then people are getting eye damage very quickly. Whereas you can't really mess up a 222 installation that badly, because it's just innately a lot safer. And so the level of expertise is, like you said, any electrician can do it, instead of needing like an expert UV installer.
Lamp lifetime and maintenance
Patrick: I'm familiar with standard lamps, although they changed a little bit in the LED era, where you have to go and replace the bulb every once in a while. What's the average lifetime of these installations?
Vivian: Yeah, we think that the bulbs that we're using right now — the manufacturer will admit it'll last at least 10,000 hours maintaining 70% output. I've seen some data that suggests it's actually a bit longer than that, maybe more like 13 to 14,000 hours. If you are only using it in an occupational setting for eight hours a day during the work week, that's about five, six years of usage before you need to do any replacements. And if you're running them 24/7, which I don't know what scenario that would be most desirable in, but that would be just about a year and a half.
When does the evidence become undeniable?
Patrick: I could imagine maybe a transportation sort of scenario where you had them in, for example, a train station or an airport where they would want to run them 24/7 or something pretty close to it.
And then I guess the big $64,000 question — or more, now that a dollar's worth less than it used to be — is, when will we start seeing the actual results in the physical universe? When does the data start getting sort of undeniable that's like, if you install this, then you win in terms of the amount of sick days you have and the clinical consequences?
Vivian: Yeah. So I think this is a really, really tough question. There's a number of reasons that a randomized controlled trial for an environmental disease transmission intervention is quite hard to do. We're not entirely sure that we know how to design such a trial, but in principle, you should expect that this will roughly follow a sigmoidal shape.
So the first few adopters are going to see sublinear benefits because — maybe you install it in your office and nobody gives each other the flu in your office, but then your kid's school doesn't install anything and the flu goes around there, and then you just catch the flu from your kid. So there's going to be some critical coverage point within a community where the suppression really takes off. And we have modeling on this, but yeah, I think it's really, really uncertain.
Tuberculosis and long-term care as early proof points
Misha: I think I'm actually a lot more optimistic than that. I think a lot of specialized use cases are gonna see fairly strong evidence a lot faster and a lot more easily. Basically, situations where people are not as social, not as mixed, or for pathogens that are particularly susceptible.
So there's been really good results on tuberculosis, for example, and this is not really something we think about day-to-day in America, but there's a lot of institutions and places that are like tuberculosis hotspots that are also not doing a ton of social mixing. This is a big problem in a lot of countries as well that are not America. But also even in America, there's places that have a lot of tuberculosis, and I think those places will see, for example, pretty noticeable drops in transmission fairly fast and pretty reliably. And I think that kind of thing will be forthcoming relatively quickly compared to, if you put this in a school in a community, we don't really know that fast.
I also think long-term care centers and basically senior centers are another place where we're probably gonna see results relatively faster, because there's just not as much social mixing there, right, with the rest of society.
Vivian: I would agree with that. And also I should point out that we have some really encouraging results on tuberculosis specifically. There's a trial going on in South Africa that already has preliminary results, and they're seeing 90% transmission suppression in these TB wards. This is an animal study, so the way it's set up is that there's guinea pigs which are exposed to the humans only through the air, and they're monitoring what percent of the guinea pigs get tuberculosis.
But tuberculosis is actually not very sensitive to far UVC at all. It's relatively resistant. And I would say it's about maybe ten times more resistant than your typical respiratory virus like flu or coronavirus. So, is — are flu and coronavirus transmitted through the air in the exact same way as tuberculosis? No, probably not. But there is in fact a lot of reason to be optimistic.
I think Misha's totally right about long-term care centers probably seeing much more immediate benefits than the school or office case. Or perhaps boarding schools. I think boarding schools could be a relatively more immediate example.
Patrick: I think this is going to be interesting because there are some institutions that require, we're gonna need a stack of academic papers on this that have confidence intervals, et cetera, et cetera. But when you're talking about 90% decreases in infections, for example, the anecdotal evidence will pile up in certain communities extremely quickly, and it is just nakedly and obviously incentive compatible for a case like a long-term care center.
Regardless of who owns it, PE firm or otherwise, they would strongly prefer that the residents not die. And that is both the humanitarian mission for them, plus also they get paid based on how many residents are still living at the moment. And so you could imagine it having a happy viral coefficient in that community of practice as soon as a chain installs it in one location and then is able to check at the end of the month. Well, we had 34 infections in the median one of our locations, and the one where we installed this, we had three, and then make the obvious decision very quickly after that.
So knock on wood, I hope that we will see evidence of the obvious decision getting made all over the place on a timeframe of, like, months to short number of years from now, followed by the broader societal rollout that this will likely take.
Pandemic math and kneecapping the next outbreak
Patrick: So we've talked about decreasing infections locally, but I think the huge societal upside here might be decreasing the total load of infections. In particular, kneecapping future pandemics before they start. Can you talk about some of the pandemic math that we were unfortunately all forced to become experts on back in 2020, and what this does to the potential factors of virality and similar?
[Patrick notes: If you’ve blocked this out, and I can’t blame you, the epidemiologists focus on changes in r over time, where r is the number of expected infections from a marginal patient, and a virus’ innate biological power level is roughly expressed by r0 (r naught), which is r at time = 0.
r is not a cosmological constant. It is sensitive to the biology of the pathogen, the biology of humans, and—more useful for our purposes—the built environment the pathogen finds itself trying to spread in. We routinely engineer environments to be extremely inhospitable to pathogens. These are described with complex epidemiological vocabulary such as “swimming pool.”]
Vivian: Yeah. So I think actually the pandemic prevention case is by far the most exciting element here. I think when you're trying to market something to ordinary people in non-pandemic times — and when I talk to people about this, I really avoid mentioning the pandemic, 'cause, yeah, people do not wanna hear it. I think we're all sort of collectively traumatized, and we just don't wanna talk about it.
So we're talking about like, oh, we're preventing colds and flus. My honest assessment is that I'm much more uncertain about our ability to prevent the average cold, which is just not very contagious. And probably if you're getting someone's cold, it's because you spent an extended period of time interacting with them quite close. Can we prevent some colds in immunocompromised people and in some situations? Yeah, I think so.
But I'm actually much more confident that we could decapitate a future respiratory pandemic like COVID-19 or worse, just because, paradoxically, the more contagious something is, the more surface area there is to keep it from getting as bad as it could get.
In fact, one of the earlier use cases for traditional upper-room UVC, which uses a different, more dangerous wavelength — the 254-nanometer wavelength — that was used to control measles. And measles is just absurdly contagious. It is probably the extrema of how contagious a pathogen can even be. I think it has a reproduction number of 20, so each person who has it will on average infect 20 others. And I believe that COVID-19 at its worst had a reproduction number of like one point something.
So UVC was able to successfully control measles outbreaks, and I think that makes me really optimistic that a relatively low level of coverage in key areas — transport hubs, for example, and other gathering places where people who aren't actually in communities together are mixing — is where we can do pandemic suppression for relatively low cost.
R0, K-factor, and why the built environment beats opt-in
Patrick: Yeah. The math that is common to epidemiologists and people who market video games for a living is the difference between the reproduction number — it's called r0 in epidemiology and the K-factor in marketing.
[Patrick notes: Engineering viral growth was an extremely common metaphor in a certain time and place, and I’d bet most growth teams avoid that one these days.]
But 1.08 means that something goes exponential, because for every person that is exposed to the thing, you get more than one people down the line. And 0.99 fails to go exponential. You have to keep dumping in marketing budget, and the viruses just don't have a marketing budget.
So the hope is that, for what in absolute numbers is a relatively small impairment in a COVID-19 or something, to just get it from slightly above one to slightly below one is the difference between having a pandemic and not. And knock on wood, hopefully we'll be able to avoid some pandemics at the margin.
Misha: No, I definitely think that's a huge benefit. I also think because it's part of the built environment, you need a lot fewer people to be involved in the prevention process, right? This is like one of the big problems with COVID: vaccination requires a lot of people to opt into vaccination. Masking requires a lot of people to opt into masking. If you have — you know, whoever owns a building can decide unilaterally. Sometimes it's just one person, sometimes it's a board of directors or something. But that takes way fewer people to reduce the infections in an area than previous prevention methods, right?
Building codes, ASHRAE 241, and competing technologies
Vivian: Yeah. And if a certain amount of infection prevention gets written into building codes, and then these are broadly adopted in whatever the authority having jurisdiction is — that might be on the state level or on the county level — this means that these interventions just get built into buildings on a ten-year renovation cycle.
And so I think there is some potential here in going through building codes, because right now we only have one infection prevention standard, that's ASHRAE 241, and it is an amazing standard that was put together very quickly. But it's still under construction, and it's not yet broadly adopted by any authorities having jurisdiction. But if the standard says you need this much infection prevention in your building to be compliant, and UVC is the cheapest, easiest way to do that, I mean, people are gonna do that.
And I do think that one really, really major advantage — maybe the advantage of far UVC, and UVC generally — is just it is such a cost-effective way to get the required amount of infection prevention into an air treatment space.
Patrick: Mm-hmm. The sort of competing technologies — they're not quite competing, presumably this is something that you could deploy in parallel — but your other options might include a huge upgrade to the HVAC system to cause more changes in the actual physical air in the room per hour, which requires upgrading both the central HVAC and also presumably all the vents, et cetera, et cetera, and might be either impossible or extremely cost-ineffective for buildings that already physically exist. Where this is, again, just plop the light in the corner and then you're done.
Vivian: I mean, another comparison technology might be portable air filters. Actually, not necessarily HEPA, but a lower rating can actually be just as effective, and much quieter. So just a MERV 13 portable air filter. And I do think these things are additive.
I mean, yes, we should be absolutely operating ventilation. We should be getting more outdoor air, more cycled filtered air. We should be using — you know, we call them portable filters, but you can also install them kind of in the room. We can broadly call these in-room air cleaners. I think these are some of the most cost-effective options. And for filter-based in-room air cleaners, you're gonna want them alongside UVC anyway, because pathogens are not the only airborne pollutant. I mean, there's also chemical pollutants, there's particulate matter, there's dust, there's allergens. Far UVC actually does have a modest effect on allergens by the same protein absorbance pathway. But it's much smaller than just conventional filtration. So we should be doing all of these things.
But for pathogens specifically, just moving the air tends to be really, really insufficient, especially for very large spaces that are relatively densely occupied — so auditoriums, lecture halls, gyms. It would be really, really difficult to meet the clean air standard for these spaces without UVC. Essentially cost-prohibitive for most buildings.
[Patrick notes: Also empirically in e.g. classroom environments teachers will prioritize silence over infection control, not entirely unreasonably. Every programmer who has ever Googled “reduce fan noise” understands why in their bones.]
Patrick: And so this is one of a panoply of options we have with regards to infection control in our built spaces, but it's a kind of — I guess I was about to say additive, but it's probably multiplicative, for better, if one actually does the math with regards to other non-pharmaceutical interventions. Or pharmaceutical interventions for that matter.
So if we have vaccines, they have some penetration rate in the community, et cetera, et cetera, the vaccine almost certainly doesn't become less effective just because there is less of the virus circulating around.
Lights are easier to deploy. They require less coordination among people. Also, there is a political economy question here, which is dancing in the background, where hopefully there will be less opposition to simply having light in the corner than, you know, needles deployed all over the place.
And therefore far UV makes all of our existing and future technologies better for having this deployed alongside them.
Vivian: Yeah, I think that's definitely true.
The deployment curve and the awareness problem
Patrick: I once made a bet with someone that we would have a broad deployment of this in the United States by 2030, and I think I'm going to lose that bet, unfortunately, because we're not moving at quite that speed. But if you were to put finger to the wind, what does the curve look like for deployment of this over the next couple of years in maybe an optimistic scenario and then sort of a baseline scenario?
Vivian: Yeah. So I think you might say that we're not exactly even on the curve. I think worldwide, maybe a couple thousand of these lamps are sold ev— no, I don't think it's even that high. It might be as low as just a couple of hundred a year worldwide. And I don't even have a great sense for whether this is going up.
But I think essentially it's a matter of, does this idea go viral? Har har. You know, if it takes off, if it kind of "Hey, this is a thing you can and should do" — and then I think we can maybe look at the speed of deployment of LED lighting as a case. I mean, we're still looking at at least a decade after it really takes off to get to truly wide deployment. But, like I said, that ten-year renovation cycle for commercial buildings is going to play a role here.
So it's really just a matter of, when do we get to that tipping point, and how do we get there? And this is something that really keeps me up at night, because I talk to people and people think that like, "Oh, there must be some good reason that this isn't getting going. There must be a regulatory barrier, there must be a cost barrier, or there's a critical piece of safety research missing."
And like I said, not that there isn't lots of safety research and other kinds of research still to do, but there's nothing super critical where it's like, "Oh, we just need to know this, and then we can get going." And there's not really any regulatory barrier either. It's maybe just like a social diffusion question. It's like, do we need just like a global awareness campaign? How do we get this out there? And I have been in research for my entire career, so I am not an expert in how to run a global awareness campaign or how to make one actually effective. And I started to really wish I was.
Patrick: Calling back to something that you said earlier, we're all somewhat traumatized by the pandemic, but implicitly we are racing the next pandemic, and hopefully we have it up and running in as many spaces as possible prior to the somewhat inevitable crash efforts to, again, improve our physical spaces and other resilience, that would be sort of activated in the ordinary course during pandemic times. Ordinary course, to the extent that anything is ordinary during pandemic times.
But it is combination of frustrating and intellectually interesting that there is no barrier at the moment. It is a product that is commercially available.
Scaling manufacturing: krypton chloride versus LEDs
Patrick: I guess we'll say a few words on the scaling. So this is presumptively, again, manufactured in — without loss of generality — China, and one thing that we've learned from LEDs among many other technologies is that China and the industrial ecosystem there is very good at scaling up production of things for which there is a demand. But is there any particular reason why this would be harder to scale than, for example, LED was?
Misha: I mean, there's like a few reasons, but they're not that major, right? The current manufacturing of the best emitters, which we think are the most cost-effective or the best lifespan, are only manufactured by one company that's from Japan. And they're generally in the business of making high-end, high-margin products. So even if we cut down all of their margin, we're still looking at a per unit cost of like 15, 20 bucks per emitter, which puts them at like — they're just sort of a different category than LEDs.
They're not sort of as easy to scale. They require hydrogen fluoride gas as part of the manufacturing process. This is something that's technically just a lot less simple than LEDs. It's still something that's very feasible to scale up. It's not something that's, on first principle, not scalable. It's just, this is something that's not as easy as LEDs.
Vivian: Yeah. I mean, I would push back that it's more complex than LEDs. I mean, the thing about LEDs is that it is extremely complex and capital intensive to produce LED chips. But once you have made that capital investment, you can scale it very, very effectively, in that you put in tens of billions of dollars into the capital, and if there is enough of a market, that makes sense to do.
So right now these krypton chloride excimer lamps are on a different, less aggressive cost curve. I don't think they could get as cheap as the white LED. But there are hopes for some solid-state chip-based solutions for far UV emission. I just think that we're looking sufficiently far out that we are just nowhere near the cost floor for even what we have here.
And I think once we are in this beautiful unicorn world where we're selling tens of billions of lamps and really rolling this out — how do we push the cost floor of a krypton chloride lamp even lower, and how do we get solid-state chip-based scalable technologies online? I would love to have that problem. I would be so joyful if we had that problem.
Price versus cost, and where the inflection point is
Patrick: And given that the cost doesn't seem to be the major barrier to deployment right now, and that, at least in the United States, you modeled 50% of the cost as being labor, where that is very difficult to compress — I would assume that the cost drops as we scale production of things. That's generally how it goes in manufacturing. But there isn't a radical step change in the likelihood of deployment or the ease of deployment as a function of reducing cost, it seems to me.
Misha: The existing industry is just so small, and like I said, it's kind of high margin as a business model. So I think there is a — I don't know if I would call it a radical step change, but I think there's probably a pretty big inflection point at some point, 'cause a lot of the lamps in this market are being sold for like $2,000. I recently heard someone sell their lamps for $3,500 each, right? So I think once we're talking for installations like, well, you need 100 for your building, that adds up really fast if you're spending $3,000 a lamp, right?
So I do think that cost is probably a barrier in a lot of deployments. But that cost is not fundamental to the industry. That cost is just sort of an artifact of the fact that these are tiny companies that need to have high margins to survive. And I think this is a very tractable thing if you're looking at it from a, can the world throw money at this to get deployments a lot faster and cheaper? And I think this is very feasible.
Patrick: As we're talking about a total addressable market at the moment of hundreds of lamps per year transitioning to tens of millions or hundreds of millions, yeah, our prior should heavily be on that the cost per unit goes down pretty aggressively.
[Patrick notes: I sort of want to cite a confident claim like this, but even a software guy understands manufacturing well enough to understand that. You can invent devices which have increasing-marginal-cost-per-unit over several orders of magnitude but there has to be an unfortunate fact like “can only be powered by kryptonite and we have a sharply limited supply.”
What we’re far more likely to see is learning curves like for LEDs or solar cells or similar: repeated decimation of per-unit price. Haitz’s Law, for LEDs, formulated in 2000 was approximately “cost per lumen falls by a factor of ten per decade; light output per package rises by a factor of twenty.” That curve was sustained through 2020.]
Misha: Well, I mean, I'm saying the cost per unit is not the cost, right? That's like the price per unit.
Vivian: I guess this was sort of our theory for, you know, we're offering a $500 lamp kind of on the theory that there is this inflection point, and our goal is to drop that even further. You know, I don't think there's any reason that in the near future, next couple years or so, or even sooner, that the price couldn't be on the order of 100 bucks per lamp. And that's with zero technological innovation of any sort.
Should you put one in your house?
Patrick: And for the benefit of people who haven't seen your website, not to put too fine a point on it, this is a thing that you can literally go over to AeroLamp's website and buy right now, and it comes in a box to your home or office, and then it is as easy to install as any other lighting fixture.
Vivian: Actually, even easier. It's sort of just, you stick it on a tripod, put it on top of your bookshelf. Or just use a drywall anchor and stick it on your wall. For typical overhead lights, it can be quite a bit more annoying to install. I've done it in my home a couple of times.
Patrick: And, full disclosure, I don't have it in my house yet, but I have considered it, and it exists in several commercial spaces I've been in in the San Francisco Bay Area, among others.
Misha: Well, so this is actually one of the things that I get asked this question all the time: "Oh, should I get an AeroLamp?" But I think most private homes don't — like, the cost-benefit analysis is not really in the favor of getting one, just because there's not that much disease transmission. Most people don't live in like a big house with like 20 people or anything, right? If this is you and your family, I don't think the benefits are gonna be that high.
Although, obviously, this sort of changes based on your own personal cost-benefit analysis, right? Like, if you're particularly rich, particularly value not getting sick, if you're immunocompromised, this sort of changes the balance. But I think for most people, having it in their home for like 500 bucks doesn't actually price out very effectively. Maybe once they're down to 100 bucks, then yes.
Vivian: I don't know. I would actually maybe push back on that. I think there's quite a lot of families. I think the societal benefits are not particularly concentrated for individual homes. I think for individuals, in quite a lot of situations — I mean, for example, I had a baby two months ago, and babies do not have immune systems, essentially. And yeah, I quite valued having people over and around to help out postpartum, and I was using my lamps. I have two in my living room, and turning them on when I had company. I turn them on when I host events and gatherings.
I think for individual homes, it's less something that you might maybe have on all the time, but maybe something — it's sort of like the ventilator in your kitchen. It is something that you quite like to have for home health reasons.
Maybe not at the $500 price point, though that's maybe — I think if you actually monetized the potential health benefits, it would actually look quite a bit better. Getting sick is very, very costly, so you don't need to prevent that many episodes of illness. Especially in tiny babies, where if a tiny baby gets a fever, you're looking at an emergency hospital trip, hospitalization, and potentially a spinal tap, and it's all deeply upsetting for everyone involved. So I think you should rationally be extremely willing to pay to lower the probability of small babies, other immunocompromised people getting infections.
But the societal benefits is where homes are just kind of like, meh. It's not where I am most going to be pushing for this to go out. Where I think schools, transport hubs, that sort of thing is where you would see much more broad social suppression.
Patrick: When I was doing the back of the envelope math — and I have not devoted the last couple of years to researching this — my thought was that the case for a home shaped like mine, where there are four occupants, two of whom are school-aged children, is not that great yet, except for the aesthetic benefit.
You're putting your chip on the table: this is something that I hope gets deployed widely in the future.
But for special circumstances, sure. And also given that most people who listen to Complex Systems are employed tech professionals or similar, the absolute number is probably not that big of a barrier.
What would make this fail?
Patrick: But be that as it may, there is a question that we sometimes ask in investing, which I think is a clarifying one. If it fails to work, if this doesn't achieve the societal level benefits that we expect it to — why? Like, what is the thing that we don't have confidence on that we could potentially learn more in the next couple of years in a way that would disappoint us?
Vivian: For my part, it would be that it turns out that just a huge chunk of airborne disease transmission is strongly short range, and that there is actually very little long range transmission of common pathogens like your typical cold, your typical flu. I think when it comes to pandemics, for something to be that contagious, it does sort of need to be more long range.
So I think that it is imaginable to me that we end up in this uncomfortable world where we can't do very much for, you know, quote, normal disease transmission, but it still is totally crucial anti-pandemic technology. But now there's not really a clear business case for deploying it, because people hate paying for prevention.
It's a lot easier to sell someone something if you can tell them like, you're going to benefit from this in the next year, and see your money back through prevented illnesses and prevented absences and so on, versus, hey, this is something that you install just in the event that there's going to be another mass pandemic in the next decade or two. That's just a much tougher sell. [Patrick notes: This is one reason so many people are so interested in building codes. Smoke alarms and sprinklers sit around doing nothing in 99% of rooms every year and you have them anyway. Though, in fairness, smoke alarms are almost certainly in your city’s building code and sprinklers probably are not for single-family residences.]
But I think it's the transmission dynamics and, you know, how actually do any of these diseases transmit in real life, under what circumstances, for what sort of people. I think that's our biggest question mark, and it's very challenging to study.
Misha: Yeah, I mean, needless to say, we're pretty optimistic. But if it fails, it'll be for reasons like that, right?
Vivian: Yeah, like you could imagine that — so we think that for a typical far UVC installation, you're looking at the equivalent of one air change happening every two minutes. So that translates to 90% of coronavirus or influenza virus being reduced in about eight minutes, and double that to get to 99. So about 15 minutes for 99% reduction.
If the transmission dynamic is, you're sharing air with somebody who is ill, and after 30 minutes, after an hour, you have inhaled enough of infected air to, on average, produce an infection — far UV will totally cut down on that. If instead the transmission dynamic is, you talk to an infected person from two feet away and you get a massive dose, and you just are definitely for sure getting whatever it is that they have, then that's not really enough room for far UV to do much, or any kind of environmental intervention to do much.
You can improve that a little bit by mixing the air and interrupting the airstream between people at the social distance. But it's definitely going to be more challenging. You might still maybe reduce the severity of an infection just by reducing the viral dose. We saw this with COVID, is that even if you don't fully prevent an infection, if you get less virus in you, you're going to have a less severe infection. And probably quite a lot of diseases work this way as well.
But I think it would be quite unlikely that we would see no benefit at all. Whether the benefits are relatively marginal — like, probably still worth it, especially at a lower price point — or whether the benefits are truly socially transformational, you know, that we can just see vastly less airborne disease. That's sort of the uncertainty. But I think no benefit at all is quite unlikely. I would be very surprised by that.
The hygiene hypothesis objection
Patrick: One hopes that, given numbers like 90% deactivation of pathogens, that in an idealistic case, the amount of evidence, both formal evidence and anecdotal evidence, will pile up very, very quickly and lead to a high rate of diffusion in the built environment.
I have heard one objection from people, and I think this is interlocutors sort of overemphasizing the precautionary principle, but I’m here to ask the experts. We talked about children, infants being immune-naive, and the way you get to be not immune-naive is to have time in the world and get exposed to pathogens and have your immune system built up.
Is it possibly the case that decreasing people's contact with pathogens is on net a bad thing?
Vivian: So actually I think I would push back on that quite strongly. The modern formulation of the hygiene hypothesis does not actually posit that you need to be exposed to pathogens to train your immune system. We now think that the primary immune system training that you get as you're growing up is with environmental and commensal bacteria, the microorganisms.
There doesn't actually seem to be any benefit to having a clinical episode of illness. I was actually just looking at a couple of studies on this the other day. For example, there was one, the effect of a child having RSV — it was a neutral to negative effect on future illness. And in the extreme case, catching measles wreaks total havoc on your immunological memory. It's purely bad for you.
And I think the emerging understanding is that all viruses are like this. Like, it is purely just bad to catch a virus and get sick. You might not be able to prevent it, but you would still rather it not happen to you at all, and if it has to happen, you would rather it happen to you older. So I would much rather my kid catch whatever it is that's going around at ten than at five, and I'd rather it happen at five than at one. But I'd really rather it just not happen.
There's just no benefit. The immune training comes from contact with the environment, and — just, you know, the world is just teeming with microorganisms. There is absolutely no lack of immune training. So if we just totally ended viral infections, I think that that is just overwhelmingly good. No downside.
Patrick: That is great to hear. And a thing that I've told people is if it turns out that there is some optimal level of viral infections that we have to have to have the perfectly tweaked immune system, we can always intentionally infect ourselves in the future. If it's just like, "Oh man, my child just hasn't gotten sick enough in the last nine years."
Vivian: You know, I think this is actually called vaccines. I think intentionally exposing people to viruses — we have a technology for that.
Patrick: Exactly true. We deal with, I think, on net less disease load than we had in historical environments and haven't seen major impacts as a result of that. But we have engineering and pharmaceutical options to raise up and down the level if it turns out that we ever go below the optimal level of illness. Where, I think, finger to the wind on how things likely are, it's very likely that we are above the optimal level, if there is indeed an optimal level.
Misha: Yeah, I think people go too far in the hygiene hypothesis just because we can really look at the data, or just our lived experience of what it's like to be a human living in our society, and we can say, "Oh, asthma is up, allergies are up." The thing is, those are relatively trivial compared to childhood mortality, which is way down.
Vivian: Yeah. And I think there is just gathering evidence that increased allergies and asthma certainly has nothing to do with viral infections.
There might be some questions of, is widely deployed far UVC going to mess with the indoor microbiome in a way that affects us? And I think, yeah, my answer is mostly no, just because microorganisms, bacteria, especially those living on surfaces and not in the air, they're extremely hardy. And UVC is generally much less effective on surfaces than in the air. So this has not been explicitly studied, what are the effects on the indoor microbiome. But my bet is that if there's a cost, it's relatively minor. And in the event that there is a substantive effect, you go outside, there's gonna be germs there. There's dirt, there's bugs, there's no lack of potential exposures.
The precautionary principle and reasoning under uncertainty
Patrick: Yeah. And not to say something spicy for the podcast for the sake of spiciness, but I think that the precautionary principle ends up just being a really rough way to live life. Where you can always argue that, oh, in the multiple million dimension space that we live, there is at least one dimension where a given technology could disimprove you, and therefore a precautionary principle says never do anything.
Where, like, we know what the numbers are from infectious diseases in terms of deaths caused per year. They're horrific. If there is hypothetically a future in where there is a line in the history books, pre ending of flu and post ending of flu, that will be a world historical achievement.
We're not necessarily a hundred percent expecting that to happen as a result of far UVC. [Patrick notes: You’d be surprised, or not, how many times I’ve been told “You can say that ending the flu is on the table, but I can’t say that. You get drummed out of the field for being above a socially accepted level of optimism. But I didn’t pick this field because I am a pessimist.” Please note that I am not reporting comments from podcast guests.]
But that is something that is possible in the solution set, or possible in the outcome set, given deployment of this. And then when you compare some percentage of that to some percentage of, well, it might subtly tweak the indoor microbiomes — those are two very different numbers in terms of their impact on the human experience.
Vivian: Absolutely. I mean, I'm not saying we shouldn't study it. We should absolutely study it. I think the space of things that we don't yet know and would really like to know is huge. But you always have to reason under uncertainty and reason about different risks. And this is what I talk to people about when I talk about, centrally, photobiological safety.
In that, yes, we don't have long-term data. Yes, it is reasonable to be cautious about these kinds of exposures. We haven't studied this in this specific population. Yes, there are unknown unknowns, and I cannot fully characterize this risk. But that is just massively underrating the risks from infectious disease.
I think there is also quite a lot of unknown unknowns with how bad exactly are episodes of viral illness. I have a strong suspicion that long COVID, for example, is not special, in that quite a lot of viruses have long-term sequelae. I mentioned measles has long-term impacts on immunological memory, and there's just lots of stuff like this, like HTLV. Where we are not necessarily fully capturing the downside of an episode of clinical illness, even a relatively mild clinical illness.
Patrick: It is possible that basically the entire population, for all of human history — or at least all of human history post, say, the move to cities — is suffering under the effects of long flu. [Patrick notes: See the provocatively titled “Is the 1918 Influenza Pandemic Over?” and many other places in literature.]
We don't have a word for “long flu” because we assume that getting the flu was just something you were priced into. But in a future where hypothetically we are less priced into that, there might just be far less long-term health impacts than there were previously.
Vivian: Yeah. Some viruses can cause cancers. You know, me personally, anytime I get a bad cold in the winter, half the time I'll be coughing like crazy for two months after that. Even just, not beyond flu, we don't even fully know everything that we're constantly passing to each other. So I think, you know, unknown unknown downsides — I think there are some unknown unknown upsides as well.
Patrick: Mm-hmm. I will say relative to other sort of infection prevention measures, this one is relatively easy to desist if we decide that the math doesn't work out in favor of it, and, you know, you turn the lamp off and people stop getting doses. Where it is sort of difficult to desist from, say, vaccines. Well, obviously you can stop giving them in the future, but tough to remove a chemical change from people after they've been exposed to the chemical change. Whereas structurally, this is only affecting the top layers of your skin cells, most of which are dead, et cetera, et cetera. And after you stop getting doses, you stop getting doses.
Vivian: Yeah. I mean, you could imagine that maybe there's some crazy edge case where getting a sufficient dose to the upper layers of skin causes some protein structural change, and it diffuses down through the lower skin layer. Like, these are — you sort of have to struggle to think of them. And yeah, this is, at a certain point, reasoning under uncertainty, reasoning about risk, it just gets kind of nuts.
Where the first million dollars should go
Patrick: Yeah. Well, this has been a very informative conversation for me. Are there any thoughts that you would like to leave the audience with with respect to far UVC or AeroLamp in particular?
Misha: There's a lot of interest in clean air recently. There's big foundations that are putting a lot of money into it. Speaking as a capitalist and as a guy who runs a company, it'd be great if people bought a lot of AeroLamps. But I think in the industry and for the world in general, the really important thing is awareness.
So I think some sort of global awareness program, some sort of marketing informational campaign, some sort of thing like that could easily benefit the world more than almost anything else you can do, right? There's just so little knowledge about this as a technology. Most people you meet — to begin with, most people have never even heard of UV for disinfection, but even among people who have heard of UV for disinfection, almost no one has heard of 222.
This is just such a — the pitch is really easy. It's a cool science fiction technology where you're using special wavelengths to decrease the risk of getting sick. Everyone is eager once they know about it, but almost nobody knows about it. And I think that's sort of like, if you're thinking, "How can I make this more widespread? How can I make it more common? How can I enable the world to get better faster?" I think just information is the main bottleneck.
Vivian: Yeah, and to mention clean air, I think far UVC is only one part of the solution. Air filtration, ventilation is also part of that. And if you are a parent of a kid, you can lean on your kid's school, on their daycare, on your workplace, on any community centers or social events that you participate in, of like, "Hey, how clean is the air in this building? Can we make it cleaner? Can I help?" Whether that is far UVC with AeroLamp or with a different company.
We are very cheap, so we think that's attractive. But we're also an open source company, so in principle, anyone can just take the CAD files published on our GitHub and make their own, in principle. But clean air needs to be just much more of a thing. I mean, there are these groups, like a lot of these Still COVID-ing groups, and they're fighting the good fight, and it's just not enough. There needs to be a social movement for this, and there needs to be a broad coalition for upgrading our buildings so that they're not making us sick all the time.
Patrick: So for your finger to the wind here, if the end goal is a social movement and a widespread deployment of this into many of the built environments, if there were hypothetically a well-resourced capitalist listening to this and thinking, "Okay, but where do I spend the first million dollars?" — would you want it at this point on a trial deployment and some papers about that, or would you want it on, I don't know, a social media marketing campaign to get Taylor Swift or someone to adopt this and achieve adoption that way?
I'm saying something that's kind of absurd, but are we more limited on the formal evidence at this point, or are we more limited on just getting more people to know about it?
Misha: I think the latter, but you could easily argue the other way around.
Vivian: Yeah, I mean, I would say the former, just because I think that might just be a more robust way to get word out relative to, you know, if Taylor Swift does it. I do think that we are maybe early enough that too much attention could have something of a backfire effect. So we wanna scale responsibly.
I'm a fan of trial deployments. And I think we've already kind of started seeing effects with AeroLamps just kind of around in SF and Berkeley venues. Also here in DC, an increasing number of some of these group houses and event spaces are having these. And just the more that they're kind of around and people are like, "Oh yeah, that's just like a normal piece of infrastructure," I think that can function on its own. Yeah, I think Taylor Swift talking about far UVC right now, it could be very good, it could be very bad, or it could have no effect at all.
Misha: Yeah, honestly, if I was gonna pick a celebrity, it wouldn't be Taylor Swift. I think it would be Paris Hilton. Because if high-end hotel brands get associated with clean air, I think that would be pretty valuable.
The business case and the heckler's veto
Patrick: Yeah. Oh, boy, we will be paying for the sort of political economy consequences of the pandemic for a long time, unfortunately. But one of them is that there is in some quarters some skepticism with regards to the actions of anything broadly associated with public health. There is much one could say about that.
Misha: Well, so I think the political angle here is — it is tough, but I think there's a completely non-political capitalist angle that's pretty feasible. Whereas if you run a business that employs people that are highly paid, you're kind of internalizing the costs of sickness there yourself, right?
Vivian: Or whether it's in your office or if someone's kid gets sick, and then they're out taking care of their kid. I was doing some economic analysis on this a couple of years back, and a significant part of the economic costs of just colds are actually caregiver absenteeism rather than direct absenteeism, just 'cause kids get sick a lot more. They're more vulnerable.
Misha: Yeah, and this is something we saw a lot during COVID, is a lot of private companies had interventions that were not state-level mandated, right? So this is something we've been trying to do, which is get in contact with people at Google, because Google is big enough and has enough things going on that they have someone who's in charge of the average health of Google employees, right? [Patrick notes: I bet against above 20% for the strictest version of this statement, but generally speaking, yes, the tech majors very definitely have people in charge of workforce safety. Those first-aid kits didn’t arrive on the wall by magic, and neither did the consulting reports from epidemiologists in 2020 on easy-to-adopt NPIs in office environments.]
And if we can get someone like that on board, they don't need to have a campaign that's convincing the average person about these interventions. They just need to do the math and say, "Oh, we think this'll benefit Google on the bottom line," right? And that's a very different sort of calculation.
Vivian: A lot of finance firms during COVID quite early on saw the writing on the wall and implemented clean air interventions in their offices quite quickly. I mean, this was — high quality far UVC was not broadly available then, but high levels of ventilation and filtration, companies absolutely saw the business case there. Obviously now we are in a different situation, post-pandemic things are tougher, but I think not impossible. I think at a certain price point, the business case is quite good.
Patrick: Generally speaking, an optimist, but the thing that I worry about a little bit is that it could turn into a situation where there's something of a heckler's veto, and the one person in an organization that is most opposed to public health measures might decide to fixate on this as a thing that they definitely don't want. But hopefully that doesn't happen.
Vivian: What we've seen is, this is a huge problem in trying to formally run a study. If you're trying to — you know, I wanna run a study on clean air effectiveness or far UVC effectiveness, we would like to install these in this space for the study. That is extremely vulnerable to the heckler's veto. You just need one person who is just not totally comfortable with it, and the IRB will never let you do it.
It's actually quite a bit easier if a building owner decides, "You know what? We want this. We're installing it." Generally, in offices, we found that the employees are usually much more on board, and it's the building manager that's like, "Eh, do we really wanna spend this money?" We find that it is often substantively driven by employee demands, for the office case.
But I think the vulnerability for just, like, normal technology diffusion or deployment, it's substantially less vulnerable. And the nice thing about that is if somebody has decided to install this technology because they think it's good, they feel they want to benefit from it, there's nothing stopping you from later on studying the effects of it. So we can still get good evidence without rendering ourselves super vulnerable to the heckler's veto.
Where to find AeroLamp
Patrick: And, you know, just playing out the micro politics of this sort of thing. The vaccines have a great individual cost. You have to take time out of your day, go out, get jabbed, and it's a very unpleasant experience, and which is more unpleasant than it needs to be.
(It's amazing to me that we haven't made ShotBlocker or similar standard of care everywhere.)
[Patrick notes: A ShotBlocker is a patented disposable medical device. It makes injections hurt less by non-painfully poking your skin in a few adjacent places.
In fairness the evidence is thinner than I'd like: the meta-analysis finds a real but small effect, with the trials disagreeing wildly, and cold-and-vibration devices beat it head to head. But that's rather the point. There exists a family of nearly-free things that make getting a shot less unpleasant. We implement none of them at scale. We then discover that 7-8% of people skip immunizations specifically because of needle fear. Do we then e.g. let clinicians get another billing code for doing more-than-zero pain management? No, we jawbone tech companies, saying that people will like pain actually if they only see the right series of posts.]
But be that as it may, there's a visceral unpleasantness to vaccines in a way that there isn't a visceral unpleasantness to a box in the corner. It doesn't emit anything visible, and it will simply stand there for hopefully the rest of time.
But hoping that we successfully get this deployed and that it performs to our expectations, or outperforms our expectations, and the amount of actual impact it has on lived experience. So where can people find AeroLamp?
Misha: aerolamp.net
Patrick: And thanks very much, Misha and Vivian, for being on the program today. And we will follow along with interest as this diffuses into society.
Misha: Yeah, it's great to be on here.
Vivian: Yeah, great to chat.
Patrick: Thanks for tuning in to this week's episode of Complex Systems. If you have comments, drop me an email or hit me up @patio11 on Twitter. Ratings and reviews are the lifeblood of new podcasts for SEO reasons, and also because they let me know what you like.
