Has Anyone Counted London's Respirators?

There are real-world estimates of things that could be useful in preparing for a pandemic scenario. A lot of these estimates don't currently exist, to the best of my knowledge.

Recent Pandemic Scares

There have been several pandemic scares in the past few years. We had the meningitis episode in the UK (Kent, March 2026). Then the cruise ship debacle (April–June 2026). Right now, the plague situation in Irkutsk (October 2026) is unclear. The future trajectory is blurry. Information noise is making it hard to tell what foreign actors are doing, and it is possible that we get more scares like this in the future. There are debates around whether we should aerosolise Ebola for research purposes. There are quite a few routes to a pandemic risk.

How can I help to reduce the risk of engineered pandemics? There are lots of unknowns, and it is unclear who is doing what. At the moment, I'm just a nobody who happens to blog about biosecurity. Is there any way I could make a meaningful contribution? One way that I think I can help is to write more about questions that I think the public is ignoring. And then maybe try to tease out an answer.

One question is what exact information set we need to make decisions during a crisis.

The first instinct of a policy maker might be to try and collect lots and lots of data. After all, more information is usually a good thing. But what data do we collect? Pandemic prevention is an undefined problem with so much uncertainty, and it is unclear what data might help us make better decisions. For example, one might imagine a scenario where we have a state of the art intelligence system that warns us when pandemic risk is more likely. But what exactly would be on those screens? Wastewater samples? PPE stockpiling lists? Sure, we could try to collect as much data as possible, but data collection has a cost.

Because of that, I'm focused on figuring out questions that point us to what useful data is missing. In a perfect world, the amount of data a government could collect is infinite. But we live in a world with human and financial constraints, which means we need to ask what data is actually useful for our aims.

In this essay, I will argue that estimating the immediate supply of elastomeric respirators is an important question, and offer a way to estimate it with some bounds in Greater London (tldr; 2,000 to 20,000 immediately available respirators, but it's very uncertain).

A Brief Taxonomy of Workers During a Pandemic

To motivate this, let's ask how we would reason about society in a pandemic scenario. One approach might be to triage, which is something that hospitals do in an emergency. A naive triaging strategy would be to divide society into two blurry categories, and then make decisions based on these two categories.

The first category would be the set of critical workers. It is unclear what the exact set of critical workers is. But, from just human intuition, it should at least include critical healthcare, electricity networks, agriculture, water sanitation plants, logistics and government.

The second category is non-critical workers; people who don't really fit the bill above. But as I will explain, the boundary is blurry.

First, it is unclear what exactly constitutes the set of critical work. It is almost certain that a doctor is a critical worker. It is almost certain that a water plant manager is a critical worker. It is almost certain that an electrical grid maintenance officer is a critical worker. On the other hand, a professional e-sports gamer is almost certainly not a critical worker. It is morally fine to not be a critical worker (I am not), it's just that in an emergency scenario this division might be useful.

Annoyingly, there is also a huge grey area in and between critical and non-critical workers. Consider truck drivers. They are essential to society functioning. However, in a pandemic scenario, which truck drivers are essential? Is it just the ones carrying food? What about medical equipment? Are the people who run financial infrastructure important? What about government and leadership? Do we need to give the department of culture and tourism critical worker status, or just defence and healthcare?

What a critical worker actually is in a full-blown crisis scenario is unclear. During the covid pandemic we had some guidance from the UK government here, but again this list doesn't really make the distinctions required in a full-blown scenario. GPT Astra estimates a range of 7 to 11 million, but again this number could be orders of magnitude wrong because of how logistical systems are so entangled.

Secondly, it is unclear which critical workers require person-to-person contact. If you considered the government as a part of the critical workforce, it's also possible that they could function entirely remotely. We could have the president or prime minister acting in a clean zone and governing through a video conference. It might be possible that groups could have some VR way of interacting with each other better than through a laptop screen. On the other hand, it would be impossible to be a healthcare worker without in-person contact.

If a pandemic was only airborne transmissible, we might have a way to make a system such that truck drivers may never need to be in person-to-person contact. For example, we could change shifting schedules so that loading times in big stores never overlap.

For the people that are not critical workers, it's also likely that they will eventually need to have contact with other humans during a pandemic. To survive, they need to go to stores to buy food. They need to go to doctors for healthcare. They should probably go outside to exercise. On the whole though, if a pandemic was catastrophic, it's possible that we could use technology to find a way to get the public to avoid going outside. For example, we might have robots do deliveries like they do in Milton Keynes (?). We could transport a treadmill or walking pad to each household so that people could get exercise.

And so this makes me think that we should prioritise one subset of people when it comes to pandemic preparation: critical workers who require in-person contact. The number of such people in the UK is unclear, and probably depends on logistical assumptions. I would like to try and estimate this in a different post.

Would elastomeric respirators satisfy critical worker need?

What is inevitable though is that critical workers would require masks of some sort. In my previous post, I bought and talked about elastomeric respirators. Elastomeric respirators are the dust masks that tradespeople wear. They have strong filters, and a rubber seal around the face to ensure a tight fit. They are also reusable. This potentially makes them better protection in the long run. The CDC says that they are likely to be at least as effective as N95s.

The dominant uncertainty though is that their effectiveness is heavily dependent on their fit quality. It could be possible that, on average, based on fit, N95s are better. But including average number of use and degradation, it's unclear.

But suppose that on average, the reusability does make elastomeric respirators more effective. So this leads me to the question...

How many elastomeric respirators would there be available in a pandemic?

Elastomeric respirators in the UK would probably be housed in warehouses, government stockpiles, or retail stores. DIY stores keep them in stock, since tradespeople use them as dust masks. As for warehouses or government stockpiles, it's hard to guess where those are, since this is likely to be confidential information. Warehouses would be owned by private companies, and so public-facing data is not available on how much stock mask companies would have.

It could be the case that government or warehouse stockpiles never get used. Suppose we had an extremely dangerous pandemic which was highly transmissible. We would then probably have mass paranoia, like from a psychological aftershock from Covid. In this case, a pandemic crisis might cause our logistical systems to fail. If so, private companies might be worried about their future businesses, and might elect not to distribute PPE if it's too expensive. On the government side, I have no visibility on how the government would plan to distribute their own stockpiles, and so I make no judgements there.

In that case, we have a worst-case assumption in case of logistical failure; the elastomeric respirators that are immediately available would only be those available in retail stores at that time. This assumption might be too strict for many reasons. For example, it could already be the case that hospitals and other places of work have their own PPE stockpiles. It's not clear to me how much NHS trusts keep reusable elastomeric respirators, so for now let's assume there are none.

Counting what's in London's stores

If stores are the only places which have elastomeric respirators, then how many elastomeric respirators are in stock, in retail stores, right now in London? To begin to solve this problem, I leaned heavily on AI to build some datasets that would help inform a subjective estimate of how many there are. Claude code seems particularly suited to this flavour of work.

First, I got Claude Fable to use OpenStreetMap data for the major hardware stores in greater London. The stores it got seem sensible; the stores it identified are Screwfix, B&Q, Arco, Wickes, and a handful of small independent stores. The image below is a generated map of those DIY stores in the greater London area: the colours of the dots are meant to show the rough probabilities of that DIY store actually carrying an elastomeric respirator.

And then I just got it to estimate the number of elastomeric respirators by having it look at the websites, and then estimating the low to high bounds for each one, based on inventory space and a best guess. Looking at the estimates, these seem reasonable. I've been to my local Screwfix and it seems reasonable that they stock probably a hundred units elastomeric respirators at a given time.

Multiplying the lower, central and upper bounds by the number of stores on OpenStreetMap gives a lower and upper bound estimate on the number of elastomeric respirators currently available in retail stores in Greater London. It gives us a range of 2,000 to 20,000.

Caveats

Ok, so let's think about this number of 2,000 to 20,000 respirators carefully. We might be being too generous. Even if these estimates were valid, it is possible that zero stock goes to critical workers. In a panic scenario, it is plausible that everyone lines up for elastomeric respirators and gets them first before critical workers. That would render this estimate pretty useless.

Then we have the dominant uncertainty, which comes from the point estimates of how many masks are kept at each store. Claude helped me guess some bounds but we won't really know until we ask. Maybe there's an efficient sampling method, where we could go to a subset of stores and then Bayesian update our guess. Or maybe we could get a local LLM to cheaply and continuously poll DIY store stocking lists to keep a monitor of where the elastomeric respirators are.

The second uncertainty I am worried about is the quality of OpenStreetMap data. Maybe Claude did not actually get the complete list of DIY stores. I can't think of any other significant DIY stores in the UK. But maybe art stores sell elastomeric respirators? What about those homeware stores?

But for now, let's draw a mild conclusion. Even taking the upper bound of uncertainty, 20,000 does seem low and it is mildly worrying. It seems likely that Greater London would need more elastomeric respirators, conditional on them being better than N95s. If you did index on elastomeric respirators being important, and these estimates were correct, then it's also surprising how relatively cheap they are to double in capacity.

Is there a way to circumvent this problem entirely? Well one thing we could do is just encourage people to buy elastomeric respirators over time. People buy condoms 'just in case', and so why couldn't we do the same for elastomeric respirators? And given how dependent the efficacy of respirators is on training, shouldn't there be public service announcements and schools training people how to put these on?