Comments Off on Chen Sheng and the Twitter mob

Chen Sheng and the Twitter mob

Let’s say that one day you’re minding your own business, watching a film on Netflix or something similarly mundane, and because you haven’t learned the first lesson of 21st century living — never tell anyone what you think — you make the fatal mistake of writing down your thoughts and posting them online. Before you know it, you’re trending globally (#HasHeLoggedOnYet), the BBC want to interview you, the Daily Mail has people rooting through your bins, and some website called “4Chan” is torn between asking whether you’re “OurGuy” and trying to hack your social media accounts to post excerpts from Mein Kampf.

What, precisely, logging back on to Twitter, should you do when you see the dreaded 99+ sign in your notifications tab?

The one thing you absolutely should not do is back down and apologise.

Genuinely. By the time an outraged mob forms it’s far too late for that. They’re already too amped-up  to be placated, and now they want the ceremonial slaughter of your social life. Instead, given that the world as you know it is burning down around your ears, you should double down and at least earn that cancellation. This approach has the added benefit of providing onlookers with genuinely spicy takes as opposed to some milquetoast sentiment about how it would probably be bad to abolish the police, and also allows you to get a few things off your chest. Effectively, think Chen Sheng (1).

Chen Sheng was an officer under the Qin dynasty. One day, Chen and a man named Wu Guang received orders to bring soldiers to the north to assist with the defence of the realm, which they duly set about doing. After all, the punishment for disobedience was death. And just so that no canny soldiers attempt to weasel out of having to fight a pitched battle by dawdling along the way, the punishment for being late is also death.

The problem, as ever, was the weather. Halfway to the rendezvous, Chen and Wu found themselves halted by flooding. They couldn’t go forward. They certainly couldn’t get to the battlefield on time. This posed a number of interesting philosophical questions, the most prominent of which was this:

“If we follow our orders, we will be relieved of command and executed. If, on the other hand, we raise an army of ten thousand men and launch a rebellion to seize the throne for ourselves… then if we fail we will be relieved of command and executed. So why, precisely…”

And so it was that the problem of perverse incentives was discovered and made painfully clear to the men at the top, who proceeded to learn absolutely nothing and were eventually toppled by the founder of the Han dynasty in a startlingly similar revolt caused by precisely this problem.

Right. That’s the history covered. Now back to 21st century online norms. God willing, you’ll never be in the position of a Justine Sacco or any of the other people who the internet decided to destroy on the basis of a badly worded tweet. But if you’re online long enough and you do anything even vaguely political, there’s a pretty good chance that you’ll get the minor version of it at some point, when some 100k+ account decides to quote tweet you to the hard left or hard right, and your mentions flood with weird cartoon avatars. And when that happens, just remember: an apology won’t make it stop.

Like Chen Sheng, you may as well be hanged for a sheep as a lamb. The punishment for transgression is social and career death, and they’re determined that you’ve transgressed. So why not see where holding your ground gets you?

Image courtesy of Andreas Eldh, used under a Creative Commons licence.

Comments Off on The Tower and the Office Block

The Tower and the Office Block

1. A couple of years ago, during my postgraduate studies, I was doing some research into the theoretical properties of public housing allocation systems in the UK. British councils tend to use a type of programme known as a ‘choice based letting’ system, and I was interested in how close these would be to the algorithms we used in matching theory problem sets.

The answer turned out to be ‘not at all’, and while there were probably a number of reasons for that one of the starkest was the difference between the theoretical agents in an economic model — who knows what they want, and will choose their ‘strategy’ accordingly — and the people who actually went out to find public housing to live in.

One of the most interesting discoveries from that brief sojourn out of the ivory tower was that people often held totally erroneous beliefs about the way the matching process worked. These beliefs were not confined to more general myths (‘they all go to refugees’), or a reasonable lack of understanding of how an algorithm might work, but took in some truly bizarre misunderstandings of the relatively simple part the user played in operating the system.

While the allocation algorithm is simple — applicants nominate their most preferred houses in order, then the applicant highest on the list picks a house, and so on down — the interactions between bidders can make the end result hard to predict, and people often fail to follow instructions. In turn, this leads to a sort of mythology about the best way to get a council house. People would stay up to bid in the middle of the night, convinced that being ‘first in line’ would finally get them into a permanent house, or would bid on houses they didn’t want out of concern that failing to use all their bids would render them ineligible. Neither of these things were true.

Fast-forward to May 30th, 2020. Deputy Chief Medical Officer Jonathan Van Tam is being asked whether the UK’s new Coronavirus contract tracing system could lead to fraudsters calling people up to get access to their personal data. Apparently not: “I don’t think people are going to fall for this”.

Well, maybe we’ll be lucky. But anyone with a landline knows that the moment your number gets out there, you will be inundated with calls from ‘Microsoft technical support’, ‘HMRC’, ‘Apple’ and someone who knows you were in an accident that wasn’t your fault. Generally speaking, the people making these calls aren’t doing for the sheer joy of having people hang up on them. They’re making them because they work. And they work because the people who pick up the phone are generally less switched on than the people Van Tam usually associates with.

What we have, again, is a system that makes perfect sense from the point of view of the intelligent people that designed it, but that might not be quite so intuitive for those who will end up interacting with it.

The common factor between the cases as I see them is the failure of the designers to respect a hidden constraint on their plans: the ability of the end user to understand the system and their place in it.

I would guess broadly that this type of design failure becomes more frequent the greater the distance between the designer and the end user; an awful lot of information is filtered out as we move up the levels of an organisational hierarchy, and while the broad strokes of a problem may be presented the details are very unlikely to be.

Compounding this, I suspect that that the people towards the top of an organisation tend to be rather brighter than those at the bottom. Certainly, they are likely to be better educated and have more experience with the sort of task involved in building out a product, and for many purposes these things are pretty much equivalent. The tendency of talent to rise to the top — where it can be better put to use — leads to two further problems. The first is that the people who engage with customers day-to-day as people rather than points of data are unlikely to have the skills to build systems for themselves, or to identify how to meet user needs. The second is that the people at the top who build the system are surrounded by people of similar ability who are not at all representative of the general population.

At this point, I want to take a brief detour. It is all but certain that you, the person reading this, are also not representative of the population at large. It is also very likely that you don’t know how unrepresentative you are. Robin Hanson made this point well in a 2009 post; in a random sample of US adults, only 52% of respondents could look at a table in an almanac and correctly answer a question about whether US oil exports went up or down between 1976 and 1978.

Take a different example. Around the world, 15 year old students are asked questions to compare performance across countries. One item begins by telling students that when it is 1:00 AM in Berlin it is 10:00 AM in Sydney, then asks what the time in Berlin will be when it is 7:00 PM in Sydney. What proportion of American teenagers do you think got that question right? 90%? 80%?

Just under 46% of American students answered it correctly.

The thing is, these people aren’t stupid. That’s the point. They’re average, and you don’t really know what that looks like. And that’s a problem, because it’s going to be someone like you who designs the systems they rely on and work within in everyday life.

From this point of view, the biggest challenge in designing a system isn’t building in the capability to hand every possible case fed into it; it’s designing one that can be used and run by the less capable without totally falling apart.

What is particularly difficult about this is that the construction of the problem (the less capable operators serving the less capable users) rules out the obvious solution of staffing your structure with benevolent social planners who can carefully work out the optimal solution for their client (1). Instead, whatever you come up with has to be resilient to the normal on both ends.

I suspect this understanding is what results in some of the more frustrating rigidities in bureaucratic systems. A lot of time and work goes into defining the plausible use cases, and minimising the difficulty involved in handling them; call centre workers are given scripts, websites provide neat prompts, and the entire edifice is set up to funnel people into the best path to a solution with minimal fuss. The problem with this structure appears when something slightly out of the ordinary arises, because it really doesn’t offer much in the way of flexibility or independent decision making to the people operating it. It doesn’t do this because it can’t; the rules need to be simple and predictable, because discretion is likely to result in chaos.

From this angle, the primary problem of civilisation isn’t how to stop ourselves destroying one another. Humans might be selfish, grasping, greedy, venal, petty, tainted by original sin, and so on, but we are also members of a social species that generally manages to avoid the state of all against all. Instead, it’s how to harness people towards a common scheme; sure, preventing violence and theft is a part of this, but so are safety regulations and making sure we don’t build cities on top of active faults (looking at you, America). Less Hobbes’ Leviathan, and more Duns’s Leviathan (2).

So, back to Van Tam. In a plan’s transmission from the ivory tower to the concrete office block, quite a lot can go awry. Van Tam’s failure to understand the people who will be using his service — the elderly, the vulnerable, the scared, and the sick — is going to leave people open to scammers, who I expect will take full advantage of this golden opportunity.

And this really isn’t good enough. Highly capable people like Van Tam owe the less capable people who rely on them better than this. We can argue about how intelligence works until the cows come home, but I think we can probably broadly agree that (a) it exists and (b) people possess it in differing degrees. Why it does so is a different matter entirely, and perhaps with good schooling and tuition we could all operate on the level of Euler or Goethe. But for the most part people don’t have access to the time or resources to undertake that sort of schooling, and we have to take them as we find them.

This is, in some senses, the strongest case for redistribution and paternalism. We owe these people. We were lucky enough to be born with good genetics, or into a family that valued education, or in the catchment area for a good school, or with whatever your preferred explanation for differing ability is. As Scott Alexander has remarked, this debt is especially marked in a society which has carefully removed every outside source of support that they could once have relied on; they can’t hunt deer or gather berries or grow wheat because they have no land and have no right to it. And short of the long-overdue unwinding of the Norman conquest and the enclosure movement, they are’t going to get it. We are responsible for making the world we build one that they can navigate.

And when the government sets up a system that gives scammers a golden opportunity to fleece these people, then flings up its hands and says ‘well, I don’t think they’ll fall for it’, that is a total abnegation of that responsibility.

Comments Off on New paper: Fatalism and COVID-19

New paper: Fatalism and COVID-19

I appreciate this isn’t what the three regular readers of this site have come to expect, but I am going to briefly halt the flow of half-thought-out musings to direct your attention to some of my academic work, and one new paper in particular. The study is on the effect people’s beliefs about COVID-19 have on their behaviour, and it comes up with a few cool outcomes.

The starting point is that people think COVID-19 is way more infectious than it actually is. The mean participant in our project thought that a person with the virus will go on to infect 28 others on average. The actual figure, even at the ends of confidence intervals, seems to be between one and six.

Showing how beliefs affect behaviour can be really tricky, but because of the design of the project we were able to artificially create variation in people’s beliefs about infectiousness of the disease. We designed information treatments exploiting variation in expert estimates about the infectiousness of COVID-19, seen below, showing people upper-bound or lower-bound estimates of infectiousness.

We were able to show that this information did change beliefs, and that in turn beliefs generate changes in intended behaviour.

What we found shows that exaggerated beliefs really matter: when people thought COVID-19 was massively infectious, they were less likely to comply with social distancing measures. We call this the ‘fatalism’ effect, and it actually makes quite a bit of sense. If the disease is infectious enough, then social distancing won’t keep it at bay. And if you’re going to get it anyway, then there’s not much point in going through all the inconvenience and pain of locking yourself away for weeks on end. There’s a good bit of circumstantial evidence supporting this interpretation, perhaps the strongest of which is the fact that higher perceived infectiousness led people to be less optimistic about their future prospects.

The good news is that people responded to our treatments by updating their beliefs to be closer to the expert estimates. Giving people good information that causes them to revise exaggerated beliefs downwards is an effective way to encourage social distancing and other desirable behaviours. A very conservative (and very back-of-the-envelope) calculation shows that the benefit to revising down beliefs could be worth roughly $2.7 billion in reduced mortality loss in the US alone, simply through encouraging people to wash their hands more frequently.

The full paper goes into significantly more detail on what we found and how we found it, and can be read here. Perhaps most importantly, I’d like to thank my co-authors for all their work on this. Jesper in particular put in a magnificent late night shift to get it over the line for submission, and any praise for the article should be expressed in willingness to shout him a round if you run into him in London. Any blame for the paper should be directed at me, on the basis of the bountiful evidence of this website.

Comments Off on Just because there’s a law.

Just because there’s a law.

4,400 years ago Urukagina, ruler of the Mesopotamian city-state of Lagash, set down the first known code of law. The full text is lost to history, but cuneiform script on a clay cone tells us that, among other provisions, it made clear that the powerful could not force others to sell them their possessions. From that point onward, the very notion of such a compulsory purchase of land was unthinkable, particularly in the region of piece and prosperity that stretched from the Mediterranean to the Persian Gulf.

Of course, the development of legal systems did not halt there; a thousand years later, God himself took the pen and handed down to the Israelites ten commandments. Without wishing to give any one divinely ordained law greater attention than any other, it is useful for our purposes to note the strict rule against blasphemy. After all, it is thanks to this law that people swinging an unguarded foot out of bed and onto the waiting caltrop of a fallen plug curse the concept of random chance in a cruel and uncaring universe, rather than directing anger at an anthropomorphic personification of the forces of creation and destruction driving the whole show along.

While the laws set out by the almighty were clearly beyond criticism, they were also naturally somewhat limited in their scope. They have therefore left plenty of room for innovative legal thinkers to build upon them and so work ever harder for the attainment of utopia. Several promising initiatives have emerged in relatively recent years (1), including the simplification of mathematics by the Indiana General Assembly (which was set to rule that Pi equals 3.2 prior to the intervention of an unhelpful professor) and the prohibition of death in certain localities in Brazil — noble initiatives which will surely be adopted worldwide when the resistance of the medical profession is at last overcome.

Given this brief history of law and its effect, it is curious therefore that certain people — motivated one assumes by either simple ignorance or complex bad faith — insist on constructing arguments which fail to respect the straightforward fact that once a law declares something to be so, it is so. And yet to anyone accustomed to spending any time in political discussion, the following exchange will be familiar:

“But of course X can’t happen. The law says Y.”

“Ok, but what if X happens anyway?”

Readers who slogged through that opening, take heart: you are nearly at the end of this essay (2). While the arguments I made above are clearly ridiculous this does not seem to stop people from advancing variants upon them, claiming that because a law exists it will be obeyed. This does not quite qualify as an argument to authority, so for the sake of conciseness I’d like to refer to it as an ‘appeal to the rulebook’ (3); the tendency to argue by the rules as written, rather than the situation as it is.

This was on full display in the interminable Brexit wars (4) where proponents of EEA membership made much of the legal fiction of sovereignty built into the arrangement, it appeared throughout the austerity debates where activists and government went back and forth on what benefits assessments meant for the vulnerable, and it is back with a bang in the discussion around exit strategies for lockdown, where people have a tendency to treat behaviour as axiomatic.

While economists have many faults this is one of the few sins that they do not regularly commit. The misapprehension that making something illegal means it won’t happen (or conversely that creating a legal right means that it will be used) is not one that can survive in a discipline that cares only about the incentives that people face; you can write down what you like on paper, but what matters is the consequence assigned for breaking it, and the probability that that will be enforced.

This was neatly summarised by Sir Ivor Jennings, who pointed out that Parliament, being supreme, can outlaw smoking on the streets of Paris; the law simply would not have any effect; barring the odd case involving English tourists dobbed in by Public Health England’s legions of collaborators, no-one would have any incentive to follow it (5). Ultimately, the choice of whether or not to follow a law — and it is always a choice — is made by people weighing up the relative costs and benefits.

What brings this obvious point to mind is the discussion around the potential use of immunity passports. The UK is currently examining the available technological options for verifying the status and identity of those claiming to have had the coronavirus, with the eventual intent being an easing of lockdown restrictions for these people. The submission by Onfido to Parliamentary committee makes much of ensuring that immunity passports can’t be faked or traded, “allowing at-risk individuals to continue the proliferation of the virus”. For the lucky few, life could return almost to normal; they would be afforded significantly greater freedom to work, socialise, and travel. For everyone else, tough luck and lockdown continues.

How long do you think it would take the second group to work out that catching the virus would be the socially and economically sensible thing to do? How long would it take them to work out that the chances of being caught breaking the rules are significantly lower when they can blend in with the share of the population exempted from them? And would they, while doing this, continue to assume that others would follow the laws as written — removing any concern about spreading the disease to those not yet exposed?

Image courtesy of Steven Zucker, used under a Creative Commons licence

Comments Off on The strange rites of the deep country

The strange rites of the deep country

The following text is drawn from the papers of my late uncle Theophilus. Upon his passing in the soothing and pleasant grounds of his final residence it fell to my part to order his final words and documents, seeking among them debts to be paid or wrongs to be righted. In his prime, Theophilus had been an antiquarian of some status within the small academic community associated with Miskatonic University, and it was hoped that we could pull from his final ramblings some semblance of the clear-eyed thinker who had once contributed so much to the study of folklore. I found this task disquieting; in his last days, Theophilus had become increasingly deranged, warning of strange and blasphemous cults in the fenlands beyond the river, or of glyphs beyond the sight and comprehension of ordinary men left upon the interior of his locked room by the furtive agents of some hideous, nameless cult. It was among these brittle pages marked by his distinctive, wavering hand that I came upon the following chapter.

The inhabitants of this place had lived here long before the earliest records began, and from those antediluvian depths of time survived unchanged by the forces of modernity or evolution; the singular form of the local names borne both by the townsfolk and the long, unbroken chain of headstones in the local cemetery standing as witness to their physical and genetic isolation.

There were certain casts to their features that I did not like; something reptilian in their unblinking eyes. The town stood in a torpor of decay; unpainted houses and rotten, ruined piers, civic buildings left to the squalor and neglect of the populace. Above the bulk of the settlement on a small rise the remnants of some great ancestral home looked out over the bay, the shattered windows and crumbling roof a testament to the decadence that seen this great family fall into disrepute and nameless obscurity.

I had come to this remote place to study the local folk religion, expecting some twist upon the gaudy decoration and litanies that mark the papist church in this region. What I found was troubling. The pantheon worshipped at the altars and shrines that marked the edges of the street at six foot intervals bore little resemblance to the saints and theologians so honoured by our apostate friends, nor to the degenerate gods and daemons paid tribute by our distant ancestors, and examining the carvings it seemed to me that some madness lurked there, within the proportions and detailing of these nameless icons.

Very little of their beliefs was set down in writing, and it is likely that this isolated and ill-educated populace would be quite incapable of having explained in any significant detail the beliefs and practices that drove the hours of their days. Some small amount, however, was passed to me in the form of a brief and slanting set of hieroglyphs, curiously asymmetric and distinctive in their structure. There was nothing in this text to link the local faith to practitioners of the black arts in our wider continent, no fragment of the forbidden Eltdown Shards or the curious Pnakotic Manuscripts. The faith practiced in that unhallowed place was, for all of its abnormality, one that had taken root and festered there in the long dark beyond civilization without the influence of the world beyond the bay.

The days of the locals revolved around a series of peculiar rituals, the meaning and intent of which had long been lost to the ages, but whose observance was watched as closely as these strange beings could manage. Of these strange and abhorrent rites and litanies I will set nothing in ink and paper; it is better that the noisome chants and ululating shrieks that accompanied that bizarre and crude practice be committed to the abyss of memory with my passing. However, in the interest of furthering anthropological study I will make some brief note of the pinnacle of the ritual cycle.

At eight o’clock each Thursday the denizens of that benighted town emerged from the dark dwellings in which they spent their days to make obeisance to the greatest of their many gods; sweat sheened mounds of pallid flesh raising a clamour in praise, narrow eyes carefully noting those who failed to take part in the ritual tribute. When such a one was observed there emerged from their throats a most detestable and inhuman cry, gargling and rising from the back of the throat – the signal for this strange class to fall upon the unbeliever. The consequence of this cry ranged in my observations from simple shaming to fists and vicious beatings, all accompanied by ritual words demanding that their unnamed deity withdraw its protection from the outcast.

I did not last beyond a few days in that place. As I walked the tenebrous streets the blank faces of the worshippers raised the hairs upon my neck; there a hostility in those steady gazes that I have not felt before or since that day. Upon my my third morning, I awoke to find the door of my hostelry smeared with blood, with an unstated threat that should I not take my absence swiftly I would find myself dwelling for the rest of time at the bottom of the deep, black waters of the reef.

I found that village in the fourth month of 1900; I was glad to leave it. But it is troubling to me that the spread of the motor car and the curious decision to construct a railway line to that forsaken and insignificant place has lead to the spread of this people and their blasphemous cult through our fair land. I know now that they are watching me, and that my days run short. Soon their ages long machinations will reach their final, terrible fruition, and even now I lie awake at night dreading the first time that the alien and unknowable cry rises above the rooftops of my own town: Ia! Ia! EnnAitch’Ess fhtagn!

Photo courtesy of Tim Ellis, used under a creative commons licence

Comments Off on How risky is commuting to work?

How risky is commuting to work?

Officially, there are 1,950 coronavirus cases in the UK. Given the limits on testing, and the rapid growth in infections, the government’s own advisers believe the true number could be closer to 55,000.

These cases are not evenly distributed, and we know that London is a few weeks ahead of the rest of the country. Let’s say say London accounts for roughly half of these cases (25,000). The population of the city is roughly 8.9 million. Official advice is to avoid non-essential use of public transport and where possible avoid rush hour or work from home.

How risky is ignoring this advice? If you’re in a crowd of 100 people, what’s the probability that someone in that crowd is infected? How many people fit into a tube carriage with you? How many would you actually be exposed to?

Assume for the sake of argument your commute involves one train journey, that your fellow commuters are drawn at random from the population of the city, and that you all get on and off at the same station. Standing capacity on the most recent vintage of tube trains is about six people per square meter. A cough travels about six metres, and a sneeze about eight metres 1.

The diameter of the circle we’re thinking about is roughly sixteen metres. A tube train is about 2.68 metres wide and 16.5 metres long, so if you stand in the middle of the carriage you’re roughly within range of every other commuter.

At this scale, a rectangle will serve for an approximate area. This gives 44.22 square metres – or at peak travel times 265 people.

The probability that any one of them is infected is (naively) \frac{25,000}{8,900,000}, or 0.28%.

The probability that none of them are infected is 47%.

Surprising, right?

Now think about how many people you’d actually come into contact with on a commute – walking through stations, switching trains, coming into the office. How many people might you ‘meet’ in a day? Even if you travel when our example train is at half capacity – 133 people – there’s a 31% chance one of them will be sick.

You can play with the maths yourself if you want: the formula is (1-(25,000/8,900,000))^x, where x is the number of people you meet 2. At about 246 people, the probability hits 50%.

Obviously you may only be exposed to some of these people for a short time, and just because someone is infected doesn’t mean they will pass the disease on to you. This is just the probability of passing someone with coronavirus.

Still, probably best to follow the official advice. And if you can, work from home.

Header image courtesy of Clement127 on Flickr, used under a creative commons license. 

Comments Off on Coronavirus and behaviour, part 2

Coronavirus and behaviour, part 2

This follow-up to my previous blog post goes into more detail on the UK approach to tackling the coronavirus outbreak, the assumptions driving it, and the alternative policy measures available. The usual disclaimer applies; I am trying to pull together information so I can get the shape of this in my head. Any errors are my own responsibility.

(1) The UK government’s plan

(a) The plan does appear to be developing herd immunity. 

Three days ago we had Peston telling us the strategy is “to allow the virus to pass through the entire population so we acquire herd immunity, but at a much delayed speed… such that the health service is not overwhelmed and crushed”.

This was in line with junior minister Lord Bethell’s statement in the House of Lords, where he noted that “creating some kind of herd immunity… is clearly the objective — well, not the objective; rather, it is one of the results of the virus passing through… so herd immunity will actually provide resistance to future visits by the virus.”

It was also in line with the chief scientific officer, who told the world that “our aim is to try to reduce the peak, broaden the peak, not suppress it completely; also, because the vast majority of people get a mild illness, to build up some kind of herd immunity”.

Following the somewhat heated reaction to the revelation that the government intends to allow at least 60% of the population to be infected with a potentially deadly disease with unknown long-term health outcomes, there has been a degree of walking back from such blunt statements, with Matt Hancock wheeled out to tell the public that herd immunity is not the plan.

Adam Kucharski presents a reasonable case for the defence on Twitter, explaining that herd immunity “has never been the outright aim” so much as “a tragic consequence” of a virus that will not be fully controllable; if you can’t control it, you’ll get it, so it’s about managing the timing with which it arrives.

This is probably fair; whether or not herd immunity is ‘the aim’ is a matter of splitting hairs. It is more the case that the government does not believe the virus can be stopped, so that herd immunity is how this will end. The major differences between the government and its critics are on how, given the assumptions underlying this view, we might best manage the flow of cases, and on whether the assumptions hold.

(b) The initial policy measures announced were very slim, and did not appear to reflect mechanisms for spread.

The early British response effectively consisted of asking people to wash their hands to avoid falling ill. If that fails, people developing a persistent cough or temperature were encouraged to remain at home for a period of seven days.

This was — as with much of the British strategy — out of step with the rest of the world, where an isolation period of two weeks is recommended. While patients may only be contagious for ten days or so, a two week period is simple and sure.

The virus appears to have an incubation period of approximately 5 days before symptoms appear. Patients are still contagious during this period, with a study finding that roughly 50% of patients in one studied group and 60% of another transmitted the virus to another person before showing symptoms. A pre-review paper suggests that up to a quarter of infections generated by a person may result from this phase. Moreover, a substantial share of infected individuals may remain asymptomatic throughout their experience.

This suggests that those living with self-isolating individuals are likely to catch the disease and transmit it in the period before they themselves develop symptoms sufficiently worrying to cause them to enter isolation.

Focusing purely on patients who are currently symptomatic will therefore miss a large number of carriers, and will be of limited efficacy in slowing the growth of the virus. It is not clear whether the modelling assumptions that generated the government’s understanding of the efficacy of different measures accounted for this.

It is worth noting that there is research suggesting that self-isolation may fail to have the anticipated effect when asymptomatic transmission is possible.

(c) Delaying delay.

Beyond this initial response, the government anticipated installing sterner measures over the following month. Concerns about ‘social fatigue’ led the government to choose this timetable; people would comply with social measures for a time, before changing their behaviour — potentially raising infection rates again at the peak of the disease.

From the government’s point of view, this reduces the policy response to a matter of timing; you get to introduce restrictions once, and then their effect decays. The game is to introduce them at the point where they will have the largest possible impact on the peak NHS caseload. Hence Robert Peston’s “senior government source” stating that the government is “waiting for the optimal time to force restrictions on our way of life that will be very painful”.

This is a very different perspective from that given if you believe that restrictions can be made to work long term, as Hong Kong, Singapore, and China appear to. In that framework, once the disease arrives you commit to rooting it out as early as possible; the nature of exponential growth means that a reduction in the rate of growth has a correspondingly larger effect the earlier it is introduced.

This raises an interesting possibility; if containment elsewhere does prove successful, and the UK approach results in the disease becoming endemic — or at the least, uncontained — then we may find ourselves facing travel restrictions when looking to leave these islands.

(d) Not met with universal acclaim.

As I’ve said before the UK response is very much out of line with that of other countries, and there is hardly a solid consensus behind it here. As Birmingham’s Professor Alan McNally says, “social distancing has worked in China, Singapore, and other countries”, while the Lancet’s editor-in-chief describes the British approach as “playing roulette with the public”. As covered later, the assumptions underlying it have also been questioned.

Others not specifically responding to the UK policy approach have made comments that clarify just how unusual it is, with the WHO director general telling the world that the requisite policy mix was “Not testing alone. Not contact tracing alone. Not quarantine alone. Not social distancing alone. Do it all.”

In the American context, Susan Rice — in addition to noting the time a containment strategy focused on keeping out certain groups of travellers might have bought — called “aggressive social distancing” the “last key tool for slowing the spread of the coronavirus in the United States”. Without it, there was very little hope of flattening the curve sufficiently.

(e) High Risk

This last comment points to one of the biggest flaws in the government approach. If it works, fantastic. But if it fails, it will all but ensure that no plan b is viable; infection will be widespread and growing fast, and hospitals will be ensured of a period of intolerable strain.

In order to pull it off, the government will need every assumption to hold. And it will need to time its interventions all but perfectly (or at least err on the side of moving too soon rather than too late) — something that could prove challenging without sufficient testing capacity to monitor the spread of the virus through the country.

From this perspective, it is worth noting that even a perfect application of the strategy could see the hospital caseload peak well above capacity.

(f) This plan has changed, partly because the government has been caught off-guard by assumptions failing to hold.

Since the initial steps outlined above, the government has begun to backtrack. There is now a wealth of contradictory information available.

Robert Peston, the closest to thing to a government spokesman currently available, has set out a list of potential measures in The Spectator. These include asking the over 70s to remain in strict isolation for a period of four months (likely to be enforced within the next 20 days), banning mass gatherings (over the next weekend), and temporarily closing pubs, bars, and restaurants some time after that. The option of temporarily closing schools to the children of those who are not “key workers in the NHS and police” is also being considered.

Matt Hancock, meanwhile, is telling the world that “We have a plan” and that “Herd immunity is not a part of it”. Except, as covered above, it very clearly is. His description of the government’s approach as possessing “maximum transparency” could itself be somewhat more transparent.

There are three possible explanations for this shift in tone and rhetoric. The first is that the government more clearly wants to communicate the policy set out in (1a). The second is that there has been a genuine shift generated by a desire to be seen to be acting in-line with the public mood; retaining confidence is essential in generating behavioural change. Even if policies are not being introduced earlier than planned, greater transparency could set minds at ease and win over approval, as well as emphasising how seriously the government is taking matters.

The third and least optimistic is that the assumptions underlying the policy are already being shown not to hold, and on this note it is probably relevant that the Times has been reporting that the virus has been spreading faster than government models predicted.


i) The plan, such as it is, is to try and control the spread of the virus through the population rather than attempt to contain it
ii) The measures adopted to do so (so far) may not achieve this, as they could potentially allow for significant spread
iii) More effective measures may be put in place down the line. It is unclear whether they will be timed correctly.
iv) Other countries are still attempting containment, or are attempting a similar controlled spread using dramatically more stringent tools.
v) It is not clear that this strategy will allow for an effective plan b. It is very reliant on assumptions holding.
vi) It is possible that the plan is already changing, reflecting either a need to retain the confidence of the public, or the early warning signs that needed assumptions will not hold.

(2) What assumptions underlie the government’s approach?

(a) The virus will continue to be a threat for the foreseeable future.

Public Health England believes the coronavirus outbreak could last until spring 2021. There will be no vaccine within a year, with a less than 50% chance of one being developed in this period. Containment within the UK — to the point where the disease dies out — will not be possible, and even if it were it is very likely that the disease would be reintroduced from overseas.

(b) The virus is likely to display multiple peaks when we would prefer one

The prospect of a double peaked outbreak underlies much of the UK’s planning. There are, as far as I can tell, three components to this logic. The first is the lower slack NHS capacity in the winter under usual circumstances; as seasonal influenza hospitalises the elderly, fewer beds will be available to treat coronavirus patients, and comorbidities are likely to occur which will reduce patients chances of surviving.

The second is that a number of viruses find winter conditions considerably more suitable for reproduction. It is possible that the coronavirus will spread more rapidly in the latter part of the year. Moving patients from the ‘unexposed’ to ‘immune’ shares of the population over the summer is vital in preventing a dramatic rise in cases in the winter.

The third is the expectation that behavioural changes might naturally drive a multiple peaked pattern of infections, as with the 1918 flu pandemic.

Of these, the second and third points are most questionable. Behaviour depends on beliefs and incentives, which may be very different in the more connected world we live in, while the reason why some diseases display seasonal patterns is unclear making it hard to infer whether the coronavirus will do so.

There are some suggestions that an environment with greater absolute humidity provides certain viruses with greater viability. It is possible that coronavirus will be one of the viruses which benefits, but there does not appear to be sufficient information to say with any certainty whether it is. It is worth noting that the virus has sustained transmission in supposedly hostile environments. As Harvard epidemiologist William Hanage notes, flu rules might not apply. Exposing people to the disease now to avoid a hypothetical second peak is risky.

(c) Social fatigue/Isolation fatigue/reduced compliance/disobedience

The assumption underlying the delay in harsher messages was simple: people would only change their behaviour for so long, went the logic. When instituting a new plan, you would therefore have a window of obedience before people began to change again — potentially just a few weeks, with any change in behaviour arriving at the peak of the crisis. Ask less now, get more later. Avoid “self-isolation fatigue”. But what precisely was this based on?

Over the last few days the Guardian managed to get a member of the advisory group on record pointing towards two relevant bits of evidence. The first was “a rapid review published in the Lancet last month on the psychological impact of quarantine”, which listed some potential negative psychological effects upon those undergoing it, emphasised the limitations of the papers reviewed, and recommended keeping such measures in place for as short a period as possible. The second was “a paper by the Economic and Social Research Institute in Dublin on how to harness behavioural science to fight the coronavirus”, which found that “extending isolation periods beyond initial suggestions risked demoralising people and increasing noncompliance”. The same paper also noted the danger of highlighting the potential negatives of isolation in advance, as people might understandably try to avoid such measures.

Spot the problem? Neither of them make the case for delaying the introduction of measures; they say to keep them to shorter periods if possible, and to be clear about how long they will last. The evidence base for the fatigue effects is not clearly visible in the literature.

This is all the more confusing when we consider the November 2018 update on pandemic modelling, which noted that “all social distance measures depend on compliance by the population which, in turn, depends on the social acceptability of the measures. Without good behavioural research on these it is difficult to predict the impact of such measures being deployed in a future pandemic”. Quite how we moved from a situation where fatigue was a total unknown to an agenda-setting fact is something of a mystery.

Perhaps more importantly, the research the government does have emphasises the need for transparency and good information in alleviating concern — something we are sorely lacking.

Other research has, for example, found that believing an outbreak has been exaggerated or having lower information understandably reduced engagement with social distancing in previous outbreaks. What message does the government believe ‘business as usual’ will send?

Similarly, while a UCL review of behaviour in previous outbreaks found that at-risk groups were more likely to comply with protective behaviours, a high level of trust in authorities was also important. Given the very conspicuous doubts over the government’s approach, could it be possible that attempting to over-optimise the timing of interventions may be reducing their efficacy through other channels?

A relatively recent paper examining willingness to self-isolate found ‘attitude’ (towards self-isolation) and ‘social norms’ (public perception of actions) played the largest role in regulating behaviour; what effect do we think telling people they are expected to break quarantine, can’t be trusted to behave sensibly, and indeed don’t need to take significant actions might have on agent beliefs and actions?

Research in an Australian context suggests that creating credible concern over personal or familial wellbeing would be associated with improved adherence to public health measures. Telling the public that everyone will get it, and all will be well, seems likely to work against this effect.

It is worth noting that questioning the efficacy of measures is consistently found to be an effective way to undermine cooperation. Legitimate, competent authority, normalising desirable behaviours (and creating strong norms against what we might term ‘selfish’ behaviour), and effective communication are all tools through which we can increase compliance with measures.

On which note, there is research suggesting that one major factor causing people to end self-isolation early is not fatigue or boredom, but simply money. The best way to improve the efficacy of social-distancing measures is to ensure people can afford them.

I have thrown a lot of papers at you here. As you can probably guess, I can’t vouch for the design and statistical power of each one. My point is that the government has placed a great deal of weight on a single assumption that may not hold.

It is unclear, for instance, why people would become bored with quarantine and leave home at the peak of the crisis, or continue to do so if their actions caused the rate of infections and deaths to spike upwards once more. It is also unclear whether the concern of non-compliance with voluntary measures is significantly reduced when such actions are imposed in top-down fashion.

It is also unclear whether the current behaviour of the British public lends support to one side of the argument or the other. On the one hand, you have people like Michael Head, an academic at Southampton, who suggests that “People already can’t be trusted to buy toilet roll properly, so how about long-term compliance when significant levels of freedom are removed and there’s a need to stay indoors for long periods of time? The evidence, as we have it right now, suggests it will decline.”

On the other, editor-in-chief of the Lancet Richard Horton notes “the public acted to socially distance, avoid mass gatherings, and work remotely ahead of govt advice”.

The distinction between the two is partly based in a misunderstanding of incentives; while only one is socially desirable, both reflect a strong perceived incentive to take actions to preserve personal and familial welfare — a perception that could potentially be harnessed for the better.

That the assumption of fatigue is at best questionable has been borne out by the response of the behavioural science community at large; an open letter signed by 481 individuals (at the time of writing) in the field of behavioural science expresses concerns that the assumption of “behavioural fatigue” is insufficiently based in evidence, ignores the possibility of radical behaviour change of the sort seen in South Korea, and that the approach of “carrying on as normal” may undercut the required urgency of action when government advice changes.

(d) Building up long term immunity to the virus is possible

In order for “herd immunity” to build up in a population, it must be possible for exposed individuals to develop immunity in the first place. There have been some suggestions that individuals have been reinfected after recovering from the virus, but it is not clear at all that this is common or even anything other than an artefact of inaccurate testing.

The larger question is how long such immunity would last. As the British Society for Immunology has said, “Because [the virus] is so new, we do not yet know how long any protection generated through infection will last. Some other viruses in the Coronavirus family, such as those that cause common colds, tend to induce immunity that is relatively short lived, at around three months.”

(e) Timing interventions is possible

Note that given our low testing capacity this may be difficult.

(f) We have enough capacity to manage the peak caseload generated by this approach, and can get away with an attempt to “flatten the curve” short of the measures imposed elsewhere. 

Fair warning, this one is dark.

The core assumption used by the government is that without measures the NHS will collapse, as we have seen elsewhere. Every measure taken by the government is taken with the aim of avoiding that.

The problem with this is that the government’s intended strategy still implies a considerable peak caseload relative to the number of beds available. Up to 7.9 million people may require hospitalisation according to Public Health England, with an extant supply of 170,000 hospital beds (~4,500 critical care beds). There are already “growing reports of problems on the ground”, with bits of the NHS “already falling over”, and staff exposed to the disease. As I said in my previous post, the mathematics is not optimistic.

A team of American public health specialists have gone somewhat further. Even among the seemingly healthy, there will be deaths as a consequence of the herd immunity strategy; immunisation by infection is considerably less safe than immunisation by vaccine.

The real risk however is the potential explosion of cases under the light touch measures envisaged by the government. The Americans note that even using more conservative (favourable) estimates than those adopted by the UK government, 20-40 year olds alone will require up to twice as many intensive care beds than are available. With older groups even more exposed, it’s not hard to see why they chose the paper’s optimistic title: “The direction of the UK Government strategy on the COVID-19 pandemic must change immediately to prevent catastrophe”.

One of the authors of the paper, Professor William Hanage, believes that “strong social distancing” will be required to do so, with measures (discussed below) including encouraging working from home and providing financial support for those doing so.

(g) An alternative framework is available.

As Anthony Costello notes, other countries have adopted a very different strategy. The approach in China, in particular, has been to crack down on pockets of infection: people are tested, contacts are traced, family clusters are identified, and individuals are isolated. Other countries, such as South Korea, Singapore, and Taiwan, have adopted less extreme versions of this strategy.

They will all face, as the UK has been keen to point out, the issue of how to unwind infected areas. But relative to a disease spreading freely through the population this seems like a good problem to have. If people are likely to vary their behaviour to become more risky as cases drop, and then less risky as they rise, then the caseload will oscillate in a manageable fashion.

Ultimately, taking strong measures gives a government far greater control over the speed of infection than those indicated by the UK approach. By varying social distancing it would be possible to pursue a policy of herd immunity over a more extended period, or to hold the disease in check for the twelve to eighteen months that a vaccine might take. The approach will place more strain on social organisation, but less on healthcare. It seems worth considering.


i) The coronavirus outbreak is likely to last for an extended period, and total containment in the sense of ‘eradication’ is not possible.
ii) It is possible that the virus could display multiple peaks in cases. This is however uncertain, and basing a strategy around the hypothetical poses its own risks.
iii) The evidence for social fatigue is threadbare at best, and we could likely implement longer term measures as used elsewhere. It certainly does not suggest delaying action.
iv) It is not clear whether building up long term immunity is possible
v) It is not clear that we will be able to time measures correctly.
vi) It is not clear that the measures proposed by the government would ‘flatten the peak’ sufficiently when implemented.
vii) Alternative frameworks for action are available.

(4) Things we can do

(a) Communicate better.

I don’t think it’s too strong an ask for the public to learn about massive changes to their way of living from official, named government sources with context, information, evidence, and offers of assistance, rather than through Robert Peston’s blog.

This is really, really basic stuff. The section above covered the importance of credible and authoritative messengers in generating desired changes in behaviour, and the section above that demonstrated the confusion the government’s current stance has left us in. Secrecy and spin might be the substance of politics, but it’s poorly suited to the task of gearing a nation up for the task of fighting a pandemic.

If I were to lay out the principles for an alternative system, I would choose the following:

(i) Being open and transparent;
(ii) Using clear and simple communications;
(iii) Acknowledging the existence of uncertainty;
(iv) Using absolute, as well as relative, risk;
(v) Framing ambiguous messages negatively [e.g., 1 in 100 will be ill, rather than 99 in 100 will not]*;
(vi) Using visual aids wherever possible.
*This principle recently broken by Chris Whitty

These principles seem sensible to me, but more importantly they’re the ones the government set out in its scientific summary of evidence for pandemic response planning. It should consider following them.

(b) Make it easy to follow the rules, including providing fiscal support to individuals and firms.

The best way to ensure compliance with the rules is to make them easy to follow. Even medical workers will find multiple rules tricky to follow, reducing the probability of compliance as the work required increases. Fewer large rules may be more effective than an array of smaller ones, and stating the rules simply is likely to help.

As noted above, fiscal ability is a major constraint on people’s ability to stick to regimes requiring them to withdraw from economic activity. As Greg Mankiw says, “Fiscal policymakers should focus not on aggregate demand but on social insurance. Financial planners tell people to have six months of living expenses in an emergency fund. Sadly, many people do not. Considering the difficulty of identifying the truly needy and the problems inherent in trying to do so, sending every American a $1000 check asap would be a good start. A payroll tax cut makes little sense in this circumstance, because it does nothing for those who can’t work.”

These are pretty much the textbook circumstances under which running a fiscal deficit is acceptable.

As Steven Hamilton writes over at The Bulwark (and if you have any interest in economics you should read this piece), any package should not only target vulnerable households. If we want to emerge on the other side with the minimum possible damage, businesses which would otherwise have been viable should be given support to bridge the gap in revenues. The alternative would be “a surefire recipe for a long and painful recession”.

Providing support to businesses can come alongside measures such as mandating sick leave and subsidising home working; there is no need for an unconditional transfer of money. By combining the two we will hopefully be able to improve compliance with measures taken to reduce the spread of the virus, while ensuring that we don’t set ourselves up for a painful recovery.

(c) Test widely.

As the 2018 Scientific Pandemic Influenza Group on Modelling said, “It is important to ensure mechanisms are in place to measure rates of infection in the community at different stages of the pandemic”.

This is not influenza, but it is a pandemic. Perhaps due to limited capacity the government appears to be giving up on community testing, reserving such capacity as is available for the most vulnerable.

Given the stated desire to time interventions for maximum impact, the aim of acquiring mass immunity, and the potential demand for localised measures to prevent regional health system collapses, it seems unwise to measure the flow of cases by the number arriving in hospital — particularly given the potential lag between infection and hospitalisation.

(d) Consider carefully the evidence on behaviour.

Rather than using the somewhat nebulous notion of fatigue as a reason to avoid action, the government could consider looking at the research on the interaction of behaviour and policies. Creating social norms that encourage social distancing (such as emphasising taking sick leave or working from home as the right and unselfish thing to do), maintaining public trust in the efficacy of measures and the strategy adopted by government to ensure continuity of supply and public safety (ensuring they understand what is required, and reducing less desired behaviour such as hoarding), and ensuring that people understand the level of threat posed to themselves and their families could all work to improve compliance.

If nothing else, switching messages while doing nothing may not maintain confidence in plans.

(e) Implement marginal social distancing policies. 

There are easy marginal gains on the table. For many people, working from home — at a reduced level of productivity — is an option. As Hanage says, anyone who can do so should. For a disease which is contagious in asymptomatic people, this additional space could help limit spread. While the broad evidence is that the effect is limited, where this is already possible it seems likely that the cost will be outweighed by the gain. Perhaps more importantly, it would create room to consider policies on school closures.

For a disease with no background immunity, the government’s pandemic flu research suggests school closures could reduce the peak by up to 50%, and the total number of cases by 10-20%. The latest modelling figures for the coronavirus suggest a 14 week closure would reduce the peak caseload by 10-15%. Given the role of young children in this epidemic as largely asymptomatic spreaders, this seems like a useful potential outcome. If mitigation is the aim — as is currently the case — then reactive closures upon a case arriving for a period of three weeks would produce roughly the same effect.

Closing schools changes behaviour. It is possible that it will do so in ways that will work against the effect we hope to achieve; as David Halpern says, “The models rest heavily on what people will do. Will people comply with instructions, and to what extent? If kids don’t go to school, what will happen?”

With that in mind, it is still worth noting that many of the alternatives are preferable. If children end up in a daycare — undesirable — it is still likely to bring together fewer than a school of several hundred, with associated limits on the spread of infection. If they are likely to be looked after by grandparents, efforts should be made to dissuade this and explain why it is to be viewed as undesirable.

If key workers — not just in healthcare, but in energy, transport, security, and so on — are unable to find childcare, then the option of limiting access to schools to only the children of those parents who cannot work from home or find alternative arrangements is open to us. It should not be impossible for a state that hopes to manage the peak of an epidemic to sort out a rough and ready solution to the childcare needs of NHS workers.

For limiting large scale events, evidence is limited, although the government estimates a direct reduction in cases of 5%. However, it is worth noting that from a psychological standpoint alone it would emphasise the need to change patterns of behaviour in order to limit exposure to the virus.

(f) Back up policies with enforcement.

This should go without saying, but relying on voluntary cooperation for all measures is a non-starter. The government should be willing to add some stick to the incentives for compliance.

(g) Cut the red-tape.

The American response has been hamstrung in part by CDC restrictions on testing, and a refusal to use the German-made kits supplied by the WHO. In Britain, we are eyeing the possibility of drafting in final year medical school students to help provide necessary manpower.

Where otherwise sensible restrictions can be removed in a way that will not endanger patient safety to a greater degree than inaction, governments (including ours) should be looking to clear the way. Providing care worse than that provided by a fully trained doctor is better than providing no care at all.

(h) Desynchronise if possible

This penultimate thought is my own, so I suspect it will make less sense than the rest. As I understand it, the USA is currently making a hash of its response to the virus, while the UK is attempting to deliberately target a peak of cases in the summer.

This strikes me as suboptimal for the following reason: if we experience our peak outbreak at the same time as somewhere else, we are likely to be competing for resources produced by stretched global supply chains. If we can desynchronise our pattern of infections from that of our American and continental cousins, then we may be in a better position to fight the disease.

Over the longer term, this suggests that countries might wish to coordinate when loosening restrictions on behaviour in order to avoid simultaneous flare-ups.

(i) Remember healthcare workers are human.

We are likely to be asking an awful lot of NHS staff in the coming year. It is possible that alongside sickness absences, some healthcare workers may be unwilling to come in at all. Arranging childcare, paying for hours actually worked, ensuring the supply of personal protective equipment, and taking steps to reduce stress should help to limit lost hours.


i) The government needs a clear and unambiguous message on strategy and measures taken.
ii) The government should make it easy for households to follow its requests both in what it asks of them, and through financial support.
iii) If your policy is dependent on timing, widespread testing seems important.
iv) Rather than pointing to ‘fatigue’ as a catch-all excuse for delaying action, the government should consider the behavioural consequences of action and inaction carefully.
v) Implementing marginal social distancing policies should be on the table. These may be necessary to enact the government’s current strategy.
vi) Relying on voluntary compliance with measures may be unwise.
vii) If policies designed to ensure safe treatment in normal circumstances are preventing treatment at all, the government should consider temporarily loosening them. The range of outcomes is sufficiently changed that they may do more harm than good.
viii) If possible, when timing the peak of infections the government should probably avoid coinciding with other nations, or at least staggering regions across the UK.
ix) Healthcare workers are human and will require additional support. 

(5) Summary

God, that was long.

It is not at all clear that the assumptions behind the government’s plan are holding. Given that, more aggressive social distancing measures will probably be needed, among other policy measures. Accounting for the speed with which this virus spreads, putting these into operation sooner rather than later would be desirable.

Header image courtesy of Sanofi Pasteur on Flickr, taken by CDC/ F. A. Murphy, used under a creative commons license. 

Comments Off on Coronavirus and behaviour

Coronavirus and behaviour

In this post I want to roughly synthesise what I think I know about the UK government’s response to the coronavirus outbreak, the public reasoning behind it, and whether I think it will work. I am attempting to be as clear as I can about my assumptions so that people can spot my mistakes, or at least understand my reasoning. 

(1) The virus itself. 

(a) Without changing behaviour, we will see significant total caseloads

Coronavirus appears to be transmissible via direct contact and airborne spread, as well as from contact with unclean surfaces. People may be mildly contagious when they are asymptomatic. The disease is spreading “easily and sustainably” in the USA, and has already done so in China and Italy. Each person with the disease can be expected to infect roughly 2.28 other people, although this figure is dependent on policy decisions.

The UK currently expects the virus to reach between 50% and 80% of the population. Germany provides a similar upper-end estimate of 70%.

(b) Mortality will be low overall, but higher in sub-groups

The total case-fatality rate in China appears to be around 2.3%, but varies significantly with age. Those aged 70-79 face a case mortality rate of 8%, and those aged 80+ one of 14.8%. Those with pre-existing medical conditions are also at risk.

(c) Approximately 10% of cases will require hospital treatment.

The CDC suggests that 80% of infections are mild or asymptomatic — a category which still allows for “mild pneumonia” — 15% are ‘severe infections requiring oxygen’, and 5% are ‘critical infections requiring ventilation’. Italian doctors have stated approximately 10% of cases are admitted to intensive care.

(d) The pattern of infections rises to a peak then drops. Under business as usual, the peak of infections is 10 to 14 weeks away. 

This has been publicly stated by the UK chief medical officer. We can expect to see 50% of all cases in a three week period, and 90% over a longer nine week window.

(e) Patients can take a period of ~3 weeks to recover.

Drawn from examples here and here. Patient groups may not be representative but a 2-3 week treatment period does not seem unreasonable.


i) Somewhere between 33 million and 53 million people will be infected in the UK
ii) Some 5 to 8 million could require hospitalisation assuming all severe infections are hospitalised.
iii) If we assume a less aggressive treatment regime targeting only the 5%, this leaves us with 1.66 to 2.65 million patients requiring hospital treatment.
iv) Implied peak load is then somewhere between 2.5 million and 4 million people requiring hospital treatment over a 3 week window, or 830,000-133,000 on the less aggressive assumption.
v) Given the duration of hospitalisation observed elsewhere, a substantial number of these cases will require simultaneous treatment.

(2) Health system capacity

(a) The NHS has approximately 140,000 hospital beds in England, 20,000 in Scotland, and 11,000 in Wales.

(b) There are around 3,700 critical care beds in England. For patients incapable of breathing even with mechanical assistance, 15 beds have provision for providing oxygen.

Assuming a similar proportion of around 3% of Scottish and Welsh beds allocated to critical care, let’s say a total in the region of 4,500.

(c) 80% of intensive care beds are occupied.

(d) People will continue to get sick or injured while the outbreak is ongoing

(e) Under a scenario where patient numbers continue to grow, we will likely be forced to triage patients.

Under normal circumstances we would expect significantly better outcomes for coronavirus than we are likely to experience. It seems plausible that a majority of the deaths we will see will follow because the virus has the capacity to collapse the health system.

Italy is already triaging patients, and the UK’s 2011 pandemic preparedness strategy suggested that in a high impact scenario of the sort we might expect to see that this would follow.

(f) An overloaded health system will increase the mortality rate associated with the virus

Patients who might otherwise have recovered will not do so.


i) The most optimistic peak load assumptions require a multiple of the total beds available to the NHS.
ii) If every hospital bed were turned over to treating coronavirus patients, and another 100,000 conjured up from nowhere, we would still be well short of the required total.
iii) Without closing the gap between capacity and demand, patients who would otherwise live will die. We will see a death rate potentially on par with that seen in Italy. 

(3) Closing the gap between capacity and infections

(a) Significantly adding to the total number of beds available is not a short-term project. 

We will not be able to increase capacity to meet demand. We will have to lower demand to meet capacity.

(c) We don’t have the tools to accurately track the spread of the virus and we are under-counting existing cases.

We can infer this from two points. The first is that the UK can currently carry out around 1,500 tests per day. The NHS is working to increase capacity, but this is what we have to work with for now. The response to this was (initially at least) to limit testing to those who had returned from certain infected areas – but not all such locales – and the people who knew they had been in contacted with the sick.

Combine this restrictive regime with fact (1a) and you have a perfect scenario for community circulation; an infected person coughs in an underground carriage, spreading the disease to people who don’t know about their recent movements, and the virus is free to circulate untested. Given that some of these people will arrive in hospitals to be treated, we could expect to see the disease take hold among vulnerable people and their visitors.

The second observation is simply the suspiciously high number of celebrities testing positive. Individuals in this group are more able to afford scarce testing kits; NBA team the Utah Jazz, for instance, got hold of 58 testing kits to check its employees after a player fell ill. Those 58 tests represented almost 0.4% of all the tests conducted in the USA.

While high-net-worth individuals are more likely to travel or engage with large numbers of people — raising their risk of exposure — it’s unlikely that this would sufficiently explain the high numbers testing positive. It is also unlikely that the rate of the illness is massively higher in this group than in the general population; we are missing cases.

The government agrees; the official estimate is that there are somewhere between 5,000 and 10,000 people infected. As we can’t keep up with demand for tests, only those with serious symptoms will be tested.

The likely mechanism for tracking total infections will be back-engineering estimates from positive tests on hospitalised patients. Given the uncertainty over precisely how dangerous the disease is, the error margins on these figures could be quite substantial.

(c) Containment may no longer be possible. 

This follows from the above. Given that the disease is in community circulation and our lack of testing ability, we will not be able to root it out.

With that said, containment does not appear to have been the plan in the first place. Going back to the 2011 strategy (here), the rough timeline of measures anticipated was to try and spot the arrival of the pandemic, then actively find and isolate cases. Once in the community, there’d be a slow ramping up of localised public health measures.

Attempting to prevent as many infectious arrivals as possible – as successfully achieved by Taiwan and Singapore – was not really considered. A 90% reduction in the passenger inflow would delay the peak of the disease by one to two weeks, and a 99.99% reduction by two months or so.

This, I must say, seems questionable in the light of the continued success of countries with high flows from China in reducing exposure to the disease. Reducing the total number of sick arrivals, or effectively screening at airports, seems like a plausible mechanism for reducing the number of unknown cases in the community.

(d) Reducing the total number of cases was highly desirable, but spreading the load of cases is an important secondary priority. 

If you can’t prevent people catching coronavirus, you can at least try to make sure they don’t all catch it at the same time. We have moved too slowly to effectively contained the disease, but that does not mean movement now is worthless.

(e) Lowering patient numbers could be achieved through social distancing as well as through stricter lockdowns. 

The Chinese government lockdown has successfully arrested the rate of increase in coronavirus cases. Singapore, Taiwan, and Hong Kong have managed to avoid these measures by acting rapidly to reduce the inflow of cases through travel restrictions, quarantining existing cases, and encouraging reduced contact.

These measures have involved — deep breath — mandatory and enforced quarantine or isolation for returning travellers, screening at all points of entry, banning flights from heavily affected regions or from any but essential areas, tracking the movements of infected people through financial data and other records, suspending large gatherings, encouraging working from home, conducting daily temperature screenings in workplaces, public education on hygiene, encouraging organizers to defer or cancel events, distributing surgical masks to reduce the airborne spread of the virus, conducting temperature screenings at points of entry, closing schools, and so on.

Each country has taken a different combination of measures, but the total effect in each has been similar.

(f) Reducing the rate of infection is more valuable when done early. 

This observation follows from thinking through the nature of any time-varying Markov process. What we are concerned about is the is the rate of healthy people falling sick in a period, the number requiring hospital care simultaneously, and the number who will die. These are not independent; the more people who fall ill, the higher the peak load. The higher the peak load, the higher mortality.

Given the nature of exponential growth, it’s better to implement measures to reduce the first variable early than it is to do so late.


i) We need to close the gap between capacity and peak demand; as we can’t raise the first significantly, we need to lower the second.
ii) We can’t contain the virus as we don’t have sufficient testing capacity, and it is already too widely spread in the community.
iii) Measures to lower the total spread are still available, as are those to increase the temporal dispersion of the caseload.
iv) These measures have a correspondingly larger effect when implemented early. 

(4) What the UK is doing

(a) For now, almost nothing. 

The current government tactic has been termed the ‘delay’ phase. This effectively consists of asking people politely to wash their hands properly, and to self-isolate for seven days if they show symptoms.

(b) Some light social distancing will be introduced in the next two weeks.

The government also intends to ask the elderly and otherwise vulnerable to consider self-isolation for their own good.

(c) The government’s strategy appears to be based on an assumption generated by a tenuous application of behavioural science.

The logic behind delaying more effective social distancing is that there is a risk of people suffering from “self-isolation fatigue“, leaving their containment at the height of the disease period with the result that it spreads more rapidly.

This has been described as the result of epidemiological modelling, but I think this is probably incorrect; this decision is the output of a model which took in as an input the idea that people would breach isolation after a certain point.

(d) It is not immediately apparent that this assumption is true.

Singapore, Taiwan, and Hong Kong are our test-cases for the extended application of social distancing. So far things appear to be working rather well.

The assumption behind the government model is that you can’t maintain this forever, so it’s best to try and introduce it for (say) four to six weeks when you think you can have the largest impact on the peak of the virus.

It is not immediately apparent that an extended regime of social distancing over a period of months would prove unsustainable. We know that the effect of social nudges can decay over time (example), and that people can tune out messages they’ve heard before.

But all of this is incredibly dependent on context; a new letterhead on a tax letter may be rather less effective than the image of a tumour on a packet of cigarettes. In this case, we are seeing a large change in the incentives to change behaviour (“don’t kill granny”) and a practically unique context. Relying on international evidence from previous outbreaks (such as the swine flu pandemic) displaying effect attenuation, for instance, does not quite account for the comparative predicted severity of current circumstances – and indeed that research also shows that reductions in transmission were achieved, as does research on purely behavioural responses from 1918.

Perhaps more importantly, the idea that interventions are once-and-done is deeply curious. If we were to intervene now, see a rise in cases as fatigue set in, and then repeat the intervention as death tolls begin to rise, it is plausible that there would be a degree of ‘resetting’.

(d) It is not apparent that we will be able to time measures effectively

This is an interesting strategy given that we will be restricting testing to hospital cases, and that demand for tests may well outstrip our capacity to administer them. The idea is that there is an optimal time period where action will have the greatest effect on demand. Act before, and the effect will be lower. Act after, and it will be too late.

But timing for that window is incredibly dangerous strategy when you can’t monitor the spread. The whole thing is very reminiscent of the ‘catching a falling knife’ analogy to timing buys in falling stock markets; if you can pull it off then you will profit substantially, but you risk some truly catastrophic losses.

(e) Measures which do not ‘flatten the peak’ but merely delay it are still incredibly valuable. 

It is likely that we are still some way off a functioning vaccine, manufactured and administered at scale. But with each week that goes by we are gathering data on the best approach to treating the virus, as well as improving the odds that a breakthrough will occur in the interim. Measures which delay the viral peak one to two months should be welcomed, not ignored.

Update 14/03/2020: Fixed language from ‘move viral peak forwards’ to ‘delay’.

(f) The government is not thinking at the margin

Both the 2011 pandemic response plan and the government’s current rhetoric is marked by a dismissive attitude towards measures which would ‘only’ reduce or delay the peak by a small period.

The government believes that encouraging people to wash their hands and self-isolate if symptomatic could cut the peak case-load by 20%.

Restricting mass gatherings, on the other hand, is not worthwhile, as it would only cut peak case-load by 5%. And of course people would socialise in smaller groups, swapping football grounds for pubs. Besides, as the 2011 plan sets out, these are a “useful indicator of normality”. Right. Here’s the thing; this isn’t normal, and with hospitals vastly over capacity cutting the case load by 5% could make a big difference to a lot of people, and removing mass events limits the degree to which large and unexpected clusters of transmissions can occur, making tracking easier and the infection total lower.

School closures are not planned, again following the 2011 document (which suggested localised closures in earlier phases, dropped entirely once the pandemic took hold). The government has shown no interest as “children off school would end up socialising anyway or spending time with grandparents”. Quite why infecting 20-30 is preferable to infecting 4-5 is not immediately apparent, and what worries me about this is the binary thinking it implies: ‘if you infect one person, you may as well infect many’. Anything we can do to reduce transmissions at this point is valuable, and the government’s own evidence is that it would reduce the peak load by 10-15%. The idea that school closures would affect care and health workers is also a non-starter; there is always the option of keeping them open for related students only.

Broadly, this attitude is curious. Firstly, the combination of many small measures — particularly those where the government is in a position to effectively enforce changes in behaviour — can sum to a significant change in overall outcomes. Secondly, delaying the peak of the virus is in itself incredibly valuable.

(g) The rest of the world is entirely out of step.

As covered above, the rest of the world has adopted a very different approach to that taken by the UK. Either our researchers have identified a pattern which China, Singapore, South Korea, Italy, France, Australia, Iceland, and so on have failed to find or consider, or we are basing our policy on a degree of confidence in the available evidence on behaviour which is clearly not shared by the rest of the world.

France is closing all schools and universities, as is Ireland. Japan has closed its schools. Iceland and France have banned gatherings over 100 people, and Australia over 500. Italy has progressed from a national quarantine to something akin to a total lockdown of activity, banning public gatherings, preventing non-urgent travel, limiting opening hours, and encouraging citizens to limit social contact.

And the UK is waiting to see what happens.

If you turn up in an empty lecture theatre at 9am, it’s entirely possible the professor and the rest of your classmates have mistakenly headed to the wrong room. It’s just rather more likely that you’re wrong. Other countries have perfectly competent scientists and policy makers. In many cases I suspect theirs may be rather better. Precisely how confident are we in Public Health England?

(h) The long term nature of the plan is also dubious.

The government’s concern is that when you end social distancing — assuming it would end prior to the production of a viable vaccine — you then experience an increase in cases once more. By having people infected over a short (but not too short) window, you may be able to acquire a degree of herd immunity, whereby enough of the population are resilient to the disease that it will not recur.

We can probably assume that under the government’s plan enough people will be sick for this to work in theory — 60% of the population, for the record — but it is not immediately clear that long term immunity can actually be acquired. There are also early suggestions that the disease can cause long term damage to lung function, the central nervous system, and other systems of the body.


i) The government’s strategy is based on an at-best-dubious application of behavioural science not borne out by observations elsewhere.
ii) The idea that interventions can be timed precisely enough to maximise impact is at odds with our lack of capacity for accurately tracking spread.
iii) There appear to be options open which would substantially reduce the peak caseload if taken in combination, and potentially the total number of infections.
iii) The plan to acquire herd immunity risks long-term health consequences and is not guaranteed to actually work.
iv) Other countries with access to expertise have adopted a very different approach. 


We’re embarking on a novel experiment in the response to pandemic diseases based on unclear behavioural assumptions, an evidence base cobbled together from context-dependent results likely drawn from very different circumstances, and without an understanding of the long-term effects of the disease in play.

While total containment might not be possible it is not clear that the total number of infected could not be dramatically reduced by following a different strategy. Options taken to delay the peak for as long as possible would also be valuable, as we could gather information on viable treatments.

Header image courtesy of Sanofi Pasteur on Flickr, taken by CDC/ F. A. Murphy, used under a creative commons license. 

Comments Off on Antisemitism is breaking activists’ brains

Antisemitism is breaking activists’ brains

“Isn’t it infuriating that yesterday all of Labour was tweeting furiously that the New Statesman refused to endorse Corbyn, and today they’re tweeting about how sad the Jewish Labour Movement dossier is?” ⁠ — Anon. 

Say what you will about the Conservative party but it does at least have the following advantage for those attempting to follow its internal culture wars: the bastards may be bastards, but at least they’re consistently bastards.

If only the same could be said of the Labour party’s more vocal supporters. Over the past week, we’ve had them claiming that accusations of antisemitism are simply a Tory smear, saying ‘the problem is real – but the Conservatives are the real racists’, arguing that the Jewish Labour Movement dossier ‘shows there is a problem that must be dealt with, but not that voters should choose Boris and Austerity’, and then, once enough time was judged to have passed, returning to the first point in the cycle.

For a single movement to contain this many viewpoints on the issue of antisemitism would show an alarming lack of coherence. For the same person to veer between them shows that something in their internal model of reality is rendering them fundamentally incapable of handling the world as it really is.

From my vantage point it looks as though at least part of the problem is that the issue of antisemitism, combined with a snap election, falls neatly onto the major fault-line in online left dialogue.

Those fortunate enough to have experienced the persuasive rhetoric of left wing activists will be aware that there is a certain tendency is to look at issues in terms of oppression, hierarchy, and privilege, and then to formulate arguments around policies and parties in these terms: “X is a transphobe. Y is islamophobic. Z is anti-black. Therefore you must vote for L – unless you, too, are all of these things.”

At the same time, the fundamental rationale for most of the left wing parties is a broadly utilitarian calculus which by its nature places a heavy emphasis on economic issues and redistribution. While the rise in emphasis on social issues has driven away some economically-left-but-socially-conservative voters, there is no inherent contradiction between the values of economic collectivism and social liberalism. And while some voters might be lost, they were racists, or transphobes, or some other flavour of bigot whose vote you didn’t want anyway.

The problem is that their own party is now a glaring exception to this stance, and there is nothing in their debating algorithm that tells them how to resolve it. The consistent line has been that the oppression/hierarchy/privilege axis is the single most important thing in politics, that it holds the power to disqualify utterly and without hope of redemption, and that it can never – never – be ignored. This is built into the core of their worldview. Their colours are pinned firmly to it.

But with an election looming, activists are now having to weigh their values against each other in a way they did not before. Do they stick with their rhetoric about the disqualifying nature of racism, or do they think that redistribution is more important? When forced to choose between an economically collectivist Labour party riddled with antisemitism, and the Conservatives, it seems that the economic motivation wins out.

For most voters, that would be the end of the matter. There would be a recognition that it’s a grubby sort of compromise, but that in the choice between two evils they have chosen the one that they believe to be lesser. For the activists who pride themselves on their social justice credentials, that option is unavailable.

The way they appear to be rationalising their decision while maintaining their self-conception as uncompromising anti-racists is to say that they aren’t making a moral compromise; the stories must be false, because Labour is the good party. After all, they are the good people – and they’re backing Labour! Q.E.D. The activist must simultaneously hold on to the beliefs that we must listen and believe to minorities who report bigotry, that the Labour party is not antisemitic, that bigotry is disqualifying, and that we have a duty to elect a Labour government that will do better by the poor and disadvantaged.

What this means in practice is that we end up with a peculiar whiplash between positions. The activist will declare that the problem of antisemitism is obviously dreadful, but tear into people for saying that they will vote against Labour on account of the same. They will accept the individual cases of antisemitism brought to their attention, but deny that there is any aggregate or institutional problem no matter how high the evidence piles. They will claim that Boris Johnson is a bigot who cannot be Prime Minister, but refuse to apply the same standards to Jeremy Corbyn. The conflict between the activists’ principles and their debating algorithm produces a total failure of rhetoric.

Header image courtesy of mrgarethm, used under a Creative Commons license.

Newer Posts
Older Posts