Pooled Testing is Super-Beneficial

Tyler and I have been pushing pooled testing for months. The primary benefit of pooled testing is obvious. If 1% are infected and we test 100 people individually we need 100 tests. If we split the group into five pools of twenty then if we’re lucky, we only need five tests. Of course, chances are that there will be some positives in at least one group and taking this into account we will require 23.2 tests on average (5 + (1 – (1 – .01)^20)*20*5). Thus, pooled testing reduces the number of needed tests by a factor of 4. Or to put it the other way, under these assumptions, pooled testing increases our effective test capacity by a factor of 4. That’s a big gain and well understood.

An important new paper from Augenblick, Kolstad, Obermeyer and Wang shows that the benefits of pooled testing go well beyond this primary benefit. Pooled testing works best when the prevalence rate is low. If 10% are infected, for example, then it’s quite likely that all five pools will have at least one positive test and thus you will still need nearly 100 tests (92.8 expected). But the reverse is also true. The lower the prevalence rate the fewer tests are needed. But this means that pooled testing is highly complementary to frequent testing. If you test frequently then the prevalence rate must be low because the people who tested negative yesterday are very likely to test negative today. Thus from the logic given above, the expected number of tests falls as you tests more frequently (per test-cohort).

Suppose instead that people are tested ten times as frequently. Testing individually at this frequency requires ten times the number of tests, for 1000 total tests. It is therefore natural think that group testing also requires ten times the number of tests, for more than 200 total tests. However, this estimation ignores the fact that testing ten times as frequently reduces the probability of infection at the point of each test (conditional on not being positive at previous test) from 1% to only around .1%. This drop in prevalence reduces the number of expected tests – given groups of 20 – to 6.9 at each of the ten testing points, such that the total number is only 69. That is, testing people 10 times as frequently only requires slightly more than three times the number of tests. Or, put in a different way, there is a “quantity discount” of around 65% by increasing frequency.

Peter Frazier, Yujia Zhang and Massey Cashore also point out that you could also do an array-protocol in which each person is tested twice but in two different groups–this doubles the number of initial tests but limits the number of false-positives (both tests must be positive) and the number of needed retests. (See figure.).

Moreover, we haven’t yet taken into account the point of testing which is to reduce the prevalence rate. If we test frequently we can reduce the prevalence rate by quickly isolating the infected population and by reducing the prevalence rate we reduce the number of needed tests. Indeed, under some parameters it’s possible to increase the frequency of testing and at the same time reduce the total number of tests!

We can do better yet if we group individuals whose risks are likely to be correlated. Consider an office building with five floors and 100 employees, 20 per floor. If the prevalence rate is 1% and we test people at random then we will need 23.2 tests on average, as before. But suppose that the virus is more likely to transmit to people who work on the same floor and now suppose that we pool each floor. Holding the total prevalence rate constant, we are now likely to have a zero prevalence rate on four floors and a 5% prevalence rate on one floor. We don’t know which floor but it doesn’t matter–the expected number of tests required now falls to 17.8.

The authors suggest using machine learning techniques to uncover correlations which is a good idea but much can be done simply by pooling families, co-workers, and so forth.

The government has failed miserably at controlling the pandemic. Tens of thousands of people have died who would have lived under a more competent government. The FDA only recently said they might allow pooled testing, if people ask nicely. Unbelievably, after telling us we don’t need masks (supposedly a noble lie to help limit shortages), the CDC is still disparaging testing of asymptomatic people (another noble lie?) which is absolutely disastrous. Paul Romer is correct, testing capacity won’t increase until we put soft drink money behind advance market commitments and start using techniques such as pooled testing. Fortunately or sadly, depending on how you look at it, it’s not too late to do better. Some universities are now proposing rapid, frequent testing using pooling. Harvard will test every three days. Cornell will test frequently. Delaware State will test weekly. Lets hope the idea spreads from the ivory tower.


The PCR test costs around $3.50 US to run in Australia. If we can drive costs down further by using pooled testing without losing effectiveness that would be great.

From June 30 - "The United States is once again at risk of outstripping its COVID-19 testing capacity, an ominous development that would deny the country a crucial tool to understand its pandemic in real time.

The American testing supply chain is stretched to the limit, and the ongoing outbreak in the South and West could overwhelm it, according to epidemiologists and testing-company executives. While the country’s laboratories have added tremendous capacity in the past few months—the U.S. now tests about 550,000 people each day, a fivefold increase from early April—demand for viral tests is again outpacing supply.

If demand continues to accelerate and shortages are not resolved, then turnaround times for test results will rise, tests will effectively be rationed, and the number of infections that are never counted in official statistics will grow. Any plan to contain the virus will depend on fast and accurate testing, which can identify newly infectious people before they set off new outbreaks. Without it, the U.S. is in the dark.

The delays have already started. Yesterday, Quest Diagnostics, one of the country’s largest medical-testing companies, said that its systems were overwhelmed and that it would now be able to deliver COVID-19 test results in one day only for hospitalized patients, patients facing emergency surgery, and symptomatic health-care workers. Everyone else now must wait three to five days for a test result.

“Despite the rapid expansion of our testing capacity, demand for testing has been growing faster,” Quest said in a statement last week warning of such a possibility. The company then said that orders for COVID-19 testing had grown by 50 percent in three weeks." www.theatlantic.com/science/archive/2020/06/us-coronavirus-testing-could-fail-again/613675/

Seems to be a market failure, not a government one, but Americans have grown adept at blaming whatever it is they want to blame.

About a month ago we were running at 1/3 capacity. That is, we could be testing a 3X more people than were currently being tested.

But EVERYONE is getting tests now. There's no cost--you just drive up and if you indicate you had contact with someone that was infected (no checking is done), then you pass the bar and you get a test.

I know people that have had 3 tests in 3 weeks. First was because they had been to protests. Second was because they were flying, and third was because they were going to be around old people and had just flown. And they believe this usage of the test capacity is 100% legit.

AT shares that belief, writing back in mid-May about reopening America by July 4th - "Getting to green nationwide is possible by the end of the summer, but it requires ramping up testing radically. The United States now administers more than 300,000 tests a day, but according to our guidelines, 5 million a day are needed (for two to three months). It’s an achievable goal. Researchers estimate that the current system has a latent capacity to produce 2 million tests a day, and a surge in federal funding would spur companies to increase capacity. The key is to do it now, before manageable yellow zones deteriorate to economically ruinous red zones."

There are also a number of approving MR posts talking about such a test schedule being desirable. This link is from the end of March - marginalrevolution.com/marginalrevolution/2020/03/a-solution-if-we-act.html

But in conservative capitalism, paying workers kills jobs.

Only profits, money not paying workers, creates jobs.

Profits create wealth.

Wealth creates credit driven consumption.

Consumption creates jobs.

Scarcity of capital limits production.

Production below demand creates high profits.

Paying workers to build capital creates too much capital and too much supply eliminating scarcity eliminating pricing power eliminating profit.

Eliminating profit destroys wealth.

Wealth destruction destroys credit, lack of credit prevents consumption which destroys demand killing jobs.

For example, Obama's socialism created too much oil and gas which destroyed wealth and that's killed jobs.

So, Trump refuses to allow socialist capitalism to increase testing capacity production capital because it would kill jobs by eliminating the wealth creating profits of scarcity of supply below debt driven demand.

Asia and Europe are socialist and create too much supply they export to the US in the socialist conspiracy to destroy Trump's America.

Socialists use Keynes ideology to use debt to pay idle workers to build too much capital creating too much supply that they export to the US stealing US jobs and stealing US consumer money from debt created by the Fed.

Destroying wealth by eliminating profits from scarcity.

Thus Trump levies tariffs and ramps up his trade wars.

> Only profits, money not paying workers, creates jobs.

Workers can participate in the profits of just about any company they wish. Just buy stock. If you aren't sure what company to invest in, invest in the market as a whole. Since 1990, stocks have gone up 10X, or nearly 8% per year.

There's no excuse for anyone not participating in corporate profits.

Your average family spends $3500/year eating out. If you saved $2K/year by being more frugal with food, and invested that int he stock market, $233K after 30 years.

No, Phinton, everyone is not getting tested. You still need a script from a doctor in most states. Its not like its a drive up window, test here while waiting for Wendy"s to fill you order? Pooled testing makes a lot of sense to me since there is only indications not precision in any of this and R factor is recognized in these tests.

> . You still need a script from a doctor in most states.

Which state? You only need self-certify that you were exposed to someone with covid. They do not check. That is true in Washington state, that is true in Oklahoma, that is true in CA and Alabama. Check on google, and it will pop up all the available test sites. CVS is testing nationwide.

If the best the troll can come up with is an article from ten days saying that there's a risk that something bad might happen, with no follow-up since then, I think we can conclude that the bad thing did not happen.

Prior's probably going to warn us next about running out of antibiotics.

You don't have to guess as to what it's going to be. Just watch the blue check marks on Twitter, wait about a half-day to a day, then come here. That's how the phone tree for talking points works.

Are their red check marks too, or is twitter just another tentacle of the Deep State?

To be clear, the hot take hive mind exists among conservatives as well, but to answer your question, the unapproved ones are being deplatformed as we speak.

It should be pretty obvious by now that Silicon Valley is in the tank for the Deep State. If the widespread intelligence/defense contracts didn't give it away, maybe you should consider that these services don't just arise naturally out of the ether. It's a bit incongruous to claim that the American Dream is rigged, but a bunch of college dropouts became billionaires just by coding in their basement.

Don't know about the billionaires, but the TCP/IP protocols transporting HTML to my Linux running system don't owe much to any of those services.

They owe a lot more to Vint Cerf, Tim Berners-Lee, and Linus Torvalds, with Richard Stallman providing the framwork which tied them together into the Internet I use.

Exactly. The problem with open protocols and standards are that the flow of information using them can't be controlled.

Those guys aren't filthy rich (relatively speaking) because, while they created immensely useful things, they didn't create things that could be used. (Satoshi Nakamoto is a billionaire, but we don't even know who he is, or if he's still alive. And his Bitcoin fortune hasn't been touched, we can all see that.)

By contrast, FAANG has billions in market cap with (like many tech companies) questionable business models. We know how Apple makes bank. Amazon barely does it on volume. Netflix has subscriptions that may or may not cover their production budget. Are ad sales driving Facebook and Google? What about Twitter? Seems unlikely. Why do angel investors prop up companies like these? How can they carry on for years with massive losses? How is it that they can still donate hundreds of millions to social justice causes?

But its the public sector that is ordering individual testing instead of pooled testing. How is that a "market" failure?

"The PCR test costs around $3.50 US to run in Australia. If we can drive costs down further by using pooled testing without losing effectiveness that would be great."

Is that the cost of the test? Or the cost that's getting passed along to consumers with the government subsidizing the cost?

Obviously the real cost is what matters, not the subsidized cost.

Apparently that's the cost to run the test but not collect the samples. I don't know the details, but around $3.50 US sounds about right as modern testing is highly automated.

"Unbelievably, after telling us we don’t need masks (supposedly a noble lie to help limit shortages)"

Unbelievably, we are heading into Mask Shortage II.

Some interesting dueling headlines here:

"Fauci says states need to address problems with COVID-19 response: 'If you don't admit it, you can't correct it'" - USA TODAY

"Trump criticizes Fauci and says US in a 'good place' as Covid-19 cases top 3m" - The Guardian

Very sadly, I think Trump is winning on this messaging war. Even in "blue" states, there is a terrible sort of rebellion.

"Many in Orange County resist masks even as coronavirus cases soar" - Los Angeles Times

Unexpectedly, when the "experts" repeatedly lie to the public about the effectiveness of masks, the public stops listening to the experts. What horror! Why can't everyone just love ¡SCIENCE! and unquestioningly do what they're told?

This has been a frustration of mine from the beginning. Wearing a mask while in public has a necessary benefit. However, when you lie, engage in blatant manipulation, and make obviously exaggerated claimed on the benefits of wearing a mask then people will assume you're selling snake oil. The so-called experts who are literally claiming "Myth 1: Mankind can survive without masks" (actually read this) are doing more damage, IMO, than the people claiming that it's their right to not wear one.

It's not okay, morally or pragmatically, to lie in order to manipulate people into doing the "right thing".

A lament to the limitations of "libertarian paternalism" - lie, manipulate, and stretch the truth to the public, because they don't know what's best for themselves. Perhaps they're not as smart as they think they are, in fact. I'm surprised it hasn't backfired more often.

Yeah. Because that's a strictly Libertarian thing. No, for example, Liberal has ever even stretched the truth to sell, for example, a complete upending of health care laws.

This isn't about labels and partisanship.

As an aside, I think pooled testing is fine, but it might underweight collection as bottleneck.

+1, this is a good point. Pooled testing has drastically higher collection costs.

Those costs aren't "free". I believe in the case of a school or even a business you can aggregate and collect the pooled samples fairly cheaply. However, for the general population, that's not going to work as well.

So maybe pooled testing should be initially targeted at schools re-opening, businesses, etc. The high value, lower costs instances.

Costco limit was 2 boxes per customer for surgical masks. Respirators (N95) we stocked up recently without issue. Wonder if it’s regional?

I doubt California and Washington states are exploding in cases because of Tweets from a POTUS with basically zero credibility on Covid among the public there. What percentage of Latinos in LA and Imperial counties are following tweets?

A much more likely explanation is that public choice constraints made the lockdowns unsustainable, and those same public choice constraints prevent any state from achieving a R0 of less than 1 in the medium term.

Low trust gonna low trust.

Yet the Asian and European public choice constraints were sustainable enough to control the virus. Time to move to Asia or the EU considering that public choice constraints prevent any state from achieving a R0 of less than 1 in the medium term.


State, as in one of the United States. Obviously different countries have different publics. And thus different public choice constraints. Low trust societies are gonna get low trust outcomes.

Which I’ve been saying for months:

Taiwan CDC, $200 million budget - 7 deaths
US CDC, $12 billion budget - 130,000 deaths

To pass your Turing test, please look at the screen - Greece is a higher trust society than the U.S.
(Result evaluation - did subject start laughing Y/N? Further criteria - did subject mention Ray Lopez Y/N?)

To continue your Turing test, please look at the screen again - Sweden is a lower trust society than the U.S.
(Result evaluation - did subject start laughing Y/N? Further criteria - did subject mention NYC and/or Cuomo Y/N?)

That's a good point! Greece did much better than Germany. Germany's performance was pretty bad comparatively. However, Australia, New Zealand and SE Asia were even better. Obviously Australia or Asia would be the better destination.

Not to single out this one comment but at some point the word "result" needs to replace "performance."

A solid point, but 'ongoing result' is unwieldy as the pandemic continues. Don't have a suggestion for a better term however.

I actually think performance is more apt because it's ongoing. At least until there's a widespread vaccine or were certain there won't be any follow on waves.

You might want to read this summary of statistician Karl Friston's study that shows government action has had no impact on how many have died. We'll see how this holds up over time.


Having remembered that this was done in SA during the initial AIDS outbreaks, I asked about this in March.

For college: What is the pool you would test: The students on the same dorm floor, or the students in the same class. Go for the group the student most interacts with and would likely be the place they would contact the disease.

Do not need to test the persons in the department faculty office space because faculty are never in their office and say they are working from home.

Workplace testing is better than giving employer immunity.

with 2 outcomes and pooling, you need 1 + ceil(log_2(n)) tests. For 1000 people, you need 11 tests. Apply FEC principles for redundancy to account for false positives / negatives (how many errors do you expect and want to detect / correct)

(intuitively, in the 1000 person set up, there are 2000 binary words that you wish to express, with some error - test error - in “transmission”)

Of course, more testing reveals more cases, and more cases reveal just how inept Trump and his buddy governors have been at controlling the spread of the virus. Which explains why Trump et al. oppose, or more accurately don't support, more testing. Pool testing would provide more bang for the buck, and it would reveal, quickly, the hotspots. Testing the hotspots would, in turn, reveal even more cases. When one speaks of Republican efforts to suppress the count, one typically is referring to the vote count, but it applies equally to the coronavirus infection count. Tabarrok is doing great work, and I applaud his efforts. I don't understand why Cowen has become distracted, distracted by focusing on genetics and strains and other scientific explanations for how the virus affects particular people. But at this time doing so is like focusing on the best way to avoid a fire while the house is on fire. Let's put out the fire then focus on the issues that will help explain how this virus works so we can take actions that would reduce the risk of another catastrophe like the one we are experiencing now.

> how inept Trump and his buddy governors have been at controlling the spread of the virus.

It's not the red state govs that have had the problems....New Jersey death rate is 10X that of FL. New York death rate is 16X that of Texas.

If the entire country had performed as Texas and Florida (140 deaths per million) we'd be at 46K deaths total. But instead, we're at 135K.

NY, NJ, Conn, and Mass are the states with deaths over 1000/ mil. I'd bet they'd be surprised to hear that they are buddies of Trump.

Looks like this week's buzz-word is going to be "pooled testing."

Time for everybody to claim it is THE ANSWER and they have been PUSHING IT FOR MONTHS but the GOVERNMENT WON'T LISTEN.

Then you can forget all about it when the next trendy term shows up. (Any remember Juneteenth? Yeah, well....)

Testing, Testing, Testing ... to what end, exactly ?

If we were able to test every person in the U.S. today -- what specific actions would you take with that resulting data ??

There has been a fixation on testing in the U.S. which no other country seems to imitate. But then, they also have not imitated the months of ongoing (and now accelerating) new cases that the U.S. has.

Hmmm, I wonder if the quantity of testing has anything to do with the number of new cases.


When a state like Arizona has a quarter of its tests coming back positive, the problem is not the quantity of testing, but the quantity of the virus spreading through the community. And 12,273 tests yesterday is not an impressive quantity in a state with a population over 7 million. coronavirus.jhu.edu/testing/individual-states/arizona

Testing became a fixation because the CDC's screw-up made testing an easy club to hit Trump with. If Clinton had been the president stuck with the CDC's testing problems, the media would have treated testing like masks. First we would have heard that only paranoid rubes in the red states want tests, then that would be called a noble lie, and finally there would be a campaign that only paranoid rubes in the red states don't want tests.

This might have been useful in early March, when the CDC and FDA were impediments, but with 500-700 K tests daily, not so much now. There continues to be a lack of interest in antibody tests, because the greater the number shown to have been infected, the lower the infection fatality rate (IFR). Lower IFR undercuts the argument for lockdowns and the case against reopening.

+1 for "lack of interest in antibody tests"

an anecdote to support:

Two days ago I googled "covid antibody test" and paid $10 to sign up for one at LabCorp in a matter of minutes. I showed up after work yesterday, and was in and out in approximately five minutes. There was nobody else in there, and they had about 10 blood drawing booths sets up. I found out my results this morning (negative, which surprised me).

The ease and cheapness of the process was extraordinary. It was much more comfortable than a brain swab test. There was nobody else doing it. Nobody I know is even aware of the possibility of an antibody test.

Assuming no symptoms, why would anyone be surprised at a negative antibody test?

I have fit the profile of a super-spreader for months

If you have had no symptoms, there is basically zero chance of fitting any super spreader profile

Current consensus, subject to revision, is that asymptomatic people are weak spreaders, at best.

Interesting, I was unaware of that

It is important to separate pre-symptomatic from asymptomatic, as the terms may overlap (a lot). If you have no symptoms but develop symptoms, you were asymptomatic in one sense, but only in the narrow sense of not having any symptoms before developing them. Which is why the term pre-symptomatic is used, because a pre-symptomatic person can most certainly shed the virus at an extremely high rate, and be super spreaders, but they also definitely develop symptoms (possibly very mild ones).

But covid had a one-two punch. It took a few months to definitely establish pre-symptomatic spread, but there is no question of it occurring any longer. Asymptomatic people became apparent with antibody tests, where people with zero symptoms had antibodies to covid. The working assumption is that such people are not passing the disease along in any major fashion, based on contact tracing. Nobody is ruling it out, but it does not seem to play a significant role-

Shame you weren't pushing something that would actually reduce the spread of the virus.

How is that reopening plan looking these days? The one you were pushing back in May, called Pandemic Resilience: Getting it Done. And this, from WaPo in mid-May, looks totally out of touch with reality in the U.S. today - "A disease prevalence greater than 1 percent defines red zones. Today, 30 million Americans live in such hot spots — which include Detroit, New Jersey, New Orleans and New York City. In addition to the yellow-zone interventions, these places require stay-at-home orders. But by strictly following guidelines for testing and tracing, red zones could turn yellow within four weeks, moving steadfastly from lockdown to liberty." In Arizona, Texas, or Florida, the liberty of the past couple of months is looking a bit shopworn these days.

This only causes a sad chuckle at this point, "supposedly a noble lie to help limit shortages." There are still shortages, particularly in those places were it would make a real difference - retirement communities in states like Arizona and Florida or primary care offices. Though the market is helping out, by using its vaunted pricing mechanism - "At a legislative hearing, a hospital association executive detailed how one Maryland hospital that spent $600,000 on PPE last year expects to spend $10 million this year."

Pooled testing of police, health care workers, nba players, etc. makes sense. You would expect most results to come back negative. The positive rate for individual tested (excluding repeats) in MA is 10%, pooled testing won't reduce the number analysis run. I also wonder how much of a bottle neck the actual analysis is, the process would mostly be automated in the high throughput labs


This is the HIV testing dilemma again: Consider an HIV test with 99.5% sensitivity (0.5% false negative) and 99.5% specificity (0.5% false positive). If you test random people in Australia, which has a 0.1% prevalence of HIV, then each positive result you get only has a 16.6% chance of indicating an actual infection.

Only when prevalance rates of a disease get to 3-5% of the general population does well-testing help when you have test accuracies as outlined above. In that case, if 5% of the population is infected, then when the test say "you got the disease" it's right 91.2% of the time (see link below with more math)

If anyone thinks today's covid test accuracy are anywhere near that of HIV testing, it's probably wishful thinking. Covid tests will get there, no question. But today, not a chance especially when methodology is considered (drive through versus lab)

In short: If you are aren't practicing risky sex, then an HIV test is mostly worthless to you. And if you aren't sick, a covid test is mostly worthless to you too.


"Tens of thousands of people have died who would have lived under a more competent government. "

I am not convinced that this assertion is true. What we are seeing is that the virus is spreading in places where the prevalence has remained low (California, Texas) and not spreading in places that have reached a kind of saturation level (New York). This is consistent with the same total number of people dying regardless of government response. The response just affects the rate.

> "Tens of thousands of people have died who would have lived under a more competent government. "

I think some of the deaths (eg nursing home deaths due to sick people being sent to nursing home and infecting everyone else) are indeed due to gov performance.

And I think we had a lot of states that actively decided to NOT treat the virus as something to be concerned about the moment Trump blocked the border with China. For example, Trump closed the border on Jan 30, and for the 6 weeks after that, dem leaders went out of their way to suggest to the everyone this was a joke. Comments like "C'mon down to China town and let's show this virus we're not scared!" and "Keep riding the subway, there' nothing to be worried about" were insane in retrospect. And decisions to keep things open well into March.

What we've learned from this is that you get roughly 4 weeks of "good will" from the public. They will do what is needed if told by the gov. But beyond that, forget it. So, for the next pandemic, I hope gov officials are thinking how to best use the 4-6 weeks of good will.

Borders closed tight, followed by 2-3 weeks of full isolation (eg jail if you are found outside the house) VERSUS 16 weeks of half-assed lockdowns needs to be debated. Because the 16 weeks of half-assed locked downs hasn't resulted in much in terms of a win.

We can't assume this episode is typical because the four weeks of good will were poisoned by the lack of good faith on the part of our media, political, and expert class.

While people were getting restless before, there's no denying that the protests, and subsequent approval by those who only the week before had still been Chicken Littling their way through this fake pandemic, blew the lid off any remaining scrap of legitimacy they had.

The bottom line is that our overlords ginned up a pandemic, used it to wreck the economy and get their authoritarian rocks off, then let the DNC fundraising episode known as BLM through the filter while changing their story on the purpose and need for lockdowns all along the way to impose a semi-permanent malaise not to be lifted until they get someone they can control back in the White House. It's sabotage and the people (not to be confused with The Very Smart and Serious People) are waking up to that fact.

I agree that the bias of the media has been a major problem in responding to the pandemic.

Over the past 4 years the blatant partisanship of the media has made half the country skeptical of anything they have to say. There's also a clear kind of cheerleading for the virus because they know it's bad for Trump. There's even the return of the "grim milestone" headline trope from the Bush administration.

I suppose they'd say that it's the Right causing the problem by accusing them of "fake news" all the time. But, you have to admit that a lot of it is "fake news". Not actual lies, but opinions disguised as facts, expert 'advice' that changes depending on who's in the wrong (masks are bad/good, protests are bad/good), and ginning up panic for political purposes (lots of news stories about rising cases in Arizona and Florida, few about California).

Today there are news stories trying to make it appear that Trump is going to rush a vaccine in order to have it ready before election day. That would be fantastic, a major accomplishment! So what if it helps him? He actually would deserve praise for that. Such stories are incredibly irresponsible. Ginning up partisan skepticism about a vaccine will make half the country not take the vaccine, inhibit herd immunity, and prolong this crisis.

Hydroxychloroquine is another example. Trump mentioned the use of this drug early on for treatment. A reasonable media response would be to say that perhaps the medical professionals ought to determine what treatments to use for their patients. Instead they dug up any and all anecdotal evidence of it not working, and blamed Trump for causing unnecessary death (using the case of a wife poisoning her husband with fish tank cleaner as their go-to example).

Two months later, we finally get a study through the media and pharmaceutical stonewalling. And guess what? It looks like it's effective!

How many people died because they weren't given or refused HCQ due to the negative coverage? Who's going to take responsibility for that? Not the Very Smart and Serious People here, who joined in with their handlers but are oddly quiet about it today.

Although I agree that people initially dismissed hydroxychloroquine for partisan reasons, the evidence seems to be that it is not effective. The blinded trials showed no effect. The retrospective study that found an effect looks like demographic biases could have favored the hydroxychloroquine outcome.

People really wanted it to work and there's a lot of "but you need Azithromycin" or "but you need Zinc" and other such things. From what I've read none of those additions make a difference.

On the other side, the assertion that it's particularly dangerous seems silly. Obviously lots of people take it regularly and are not dying of heart attacks.

> the evidence seems to be that it is not effective.

Your (mis)understanding is a byproduct of media bias. Europe pioneered the studies, and has seen good outcomes. A large-scale study in the US was just published a week ago that found that 13% of those treated with the drug died while 26% of those NOT given the drug died. So, it cut the deaths in half.

What an odd world we live in where something that is old, cheap, tested, safe, plentiful, readily available and works very, very well is dismissed simply because the media wants to "dunk" on the president.

How broken, how dysfunctional, how absurd is that?


Sure let's get more tests. Though it's not obvious how to use the test results. But how about N95 masks? If we had effective masks that people could wear in their self-interest instead of to help others, it's a much easier case to make to get them to wear them. I'd guess we'd see greater compliance. For more mask manufacturing, we need more melt blown filter material machines. How are we doing with our crash programs to build those?

This is currently the only thing that someone can do to protect themself. Enabling the ability to do that would have add-on benefits to the national psyche. So would metrics showing how we are progressing towards the goal of ubiquitous masks. The Times could produce colorful graphs illustrating our progress.

The reason people don't wear masks has nothing to do with self interest and everything to do with:
1) rebellion against democrats, "experts" and the media and a signal to others your attitude towards the pandemic
2) denial of the pandemic
3) they are uncomfortable, itchy
3.b) it is hot outside in most parts of the country and the masks make it worse
4) they smear makeup and cause acne
5) they look like you have a diaper or bathing suit on your face
6) inductive reasoning: the chance of getting COVID in a lot of situations such as grocery shopping where you can generally stay away from others and where you only interact with others for short periods of time probably isn't that different wearing a mask vs. not
7) it was inconvenient early on to find a mask
8) young, not worried about getting a cold and lack of understanding of what the mask does, which is to your point, prevent you from spreading to others as opposed to preventing you from getting it
9) not wearing a mask may be an attempted signal of bravery or strength

No doubt those are all reasons people don't wear masks, but you have to believe that compliance would be greater if the sales pitch was that the mask is mainly to protect you rather than others. More importantly, no one would need to care what reason people have for not wearing a mask if it's only to protect yourself. Eliminating the mask sanctimony is reason enough to have effective masks.

You are wrong, Alex. The many public universities and the many bureaucracies at all levels of government have failed America. They failed to produce and intermediate relevant and reliable knowledge for elected officials to use.

And it's not just about Covid-19. It's about almost everything related to the production and intermediation of knowledge; just read the latest joke


Now if you want to blame elected officials at all levels of government, we can discuss that but please be sure you understand the limits to each official's authority.

How does pooled testing affects the specificity? Is it the same as testing one person? Or it increases significantly the likelihood of false positives?

I think the key with pooled testing is to focus more on eliminating huge blocks of your samples with a single test. So, if you have 1000 people to test, and you run 10 tests (100 ppl each), you are going to able to ensure that 900 don't have it very quickly and cheaply (eg 9 tests won't have any positives). You then have a block of 100 people that had a positive. And that is where the real work starts. You will probably need to administer 50 to 150 tests to clear that block. You could again pool those tests. But at some point people won't come back for another test.

don Fernando Irarrazaval, right now neither pooled testing nor any other "gran solución" is reliable. Alex wants to celebrate that his idea is used in a new line of research, but you can bet that it will take a long time (at least a year) before we have reliable knowledge from this research. In the meantime, like all residents of Santiago, Chile, I'm forced to endure the consequences of politicians, bureaucrats, and fake scientists that are conditioning the lifting of restrictions on meeting
impossible requirements.

From the paper

Fortunately, several studies have looked specifically at the feasibility of group testing for detection of SARS-CoV-2 via PCR. A particular focus of these studies is false negatives, driven by sample dilution.Yelin et al.(2020)found that group sizes up to 32 were feasible at usual amplification levels, andShental et al.(2020) use pools of48, albeit with a more complex combinatorial pooling strategy.Hogan et al.(2020) evaluate the false positiverate and find it to be very low. Overall, these studies paint a reassuring picture of the test characteristics ofgrouping for SARS-CoV-2 via PCR specifically; these could be further improved with the method ofLitvak etal.(1994).

Note also that you can do multiple tests from a single sample or collection so it's not technically necessary to recall people for more tests.

One thing that is not mentioned or accounted for, is sample preparation would be done individual samples - the most human time/reagent expensive steps. An aliquot of each of the prepared samples would be pooled and analyzed. Any positive pools would have another aliquot taken from the constituents that would be individually analyzed. Most work is to go from swab to RNA extract for analysis.

At what point do swabs and appointments with nurses and doctors become the limiter instead of pcr?

At least in a number of places in Florida, Texas, and Arizona.

Scott Alexander did good work to counter the noble lie narrative re masks that now seems to be gospel amongst a certain set of bloggers who are all buddies but also readers of SSC. Basically the CDC has made the same statements about masks in the past when there were no scarcities. Other actors may have other motives but it is saddening to see his wisdom fade from the discourse so quickly.

The CDC estimated number of deaths was 1.1 million worldwide and 116,000 in the United States, for the 1957-58 Hong Kong Flu pandemic.


US 1957 population was half of current 2020 population.
World 1957 population was about one third of current 2020 population.

Testing, masks, and lockdowns were not a big deal back then -- civilization somehow survived and prospered.

The Deep State didn't desperately need to get one of their guys back in the White House back then.

With I like Ike and Tricky Dick, the Deep State owned the White House back then.

Maybe I should have been more clear in saying that a high-trust, homogeneous society didn't have an overlord class that felt the need to release a virus on the population and use it as a political weapon to prevent their combination money laundering / pedophilia racket from becoming public knowledge back then.

But we all know what happened to Nixon, don't we? They got their pound of flesh sooner or later. It was a good blueprint for Russiagate nearly 40 years later, but many of these tricks only work once.

Your hot steaming Comet Ping Pong order is ready for take out. Thanks for your business, and remember, we have a strict no gun policy in our pedo cave.

Odd timing, I would have waited until after Ghislaine Maxwell commits suicide next week to start the hand-waving dismissals.

Your properly cropped photo of Ghislaine Maxwell in a properly cropped top is now ready for broadcast on Fox News.

We apologize in advance, but those old Soviet photo artists are hard to reach these days. We are certain nobody will notice Trump's big hand is still in the picture anyways. We value your business, and are pleased to offer 10% off on your next custom cropping job, along with a pre-written apology if you get caught.


Hey, I'm all for her spilling the beans on Trump as well as anyone else who used her and Epstein's "services" (or perhaps, simply used her and Epstein). A pedophile is a pedophile, after all, regardless of their political affiliation.

How about you? Because I see a lot of hand-waving that is starting to look a little desperate.

Prince Andrew is the pedophile liar who is probably most scared of Maxwell, but since he doesn't fit into your political affiliation framework, I guess that would just be more handwaving.

Just like mentioning Alexander Mashkevitch and his yacht docking in Beverly right around the time of Maxwell's arrest would be handwaving. Or who knows, maybe you are already aware of the pedophile connections too when it comes to his yachts - patch.com/massachusetts/beverly/beverly-harbor-yacht-sighting-fuels-speculation-patch-pm

It is a big world out there - pedophiles care nothing about political affiliation, they just enjoy the company of men who can provide beautiful women, as long as they are on the younger side.


-1. It's a different disease, and if masks and other precautions had caught on early, lockdowns may not have been seen as needed.

I hope it happens. But the contact tracing/isolation protocols are still lacking. Air Force basically has mechanics showing up for shifts unless they are a first degree contact. Maternity ward my sister used to work in isn’t even isolating first degree contacts. And that truck driver that was in the LA Times died because someone that knew they had a Covid-19 decided to turn up to the party anyways.

You can’t trust people to do the right thing. I think we should realize that after seeing how governments had to enforce the lockdown in France and Italy. Universities need to be willing to enforce their policies, whatever they are, with suspensions and expulsions — or something similar. I don’t want us ending up with numbers like the UK. I worry that’s where we’re headed. At least deaths are still lagging. Hoping they continue lagging.

+1. The governance of a free, self-governing people works only when citizens are already self-governing. When children refuse to govern their own conduct, then at some point, someone claiming to be the adult will. And amid the chaos, the real children will clamor for fake order offered by the fake adult.



If you really feel that prioritizing testing is more efficacious than air sanitization and other low cost preventative strategies, the FDA could be prodded to empower individuals to test themselves as suggested by Laurence J. Kotlikoff and Michael Mina:

“One variety, paper-strip tests, are inexpensive and easy enough to make that Americans could test themselves every day. You would simply spit into a tube of saline solution and insert a small piece of paper embedded with a strip of protein. If you are infected with enough of the virus, the strip will change color within 15 minutes.”

See: https://www.nytimes.com/2020/07/03/opinion/coronavirus-tests.html?action=click&module=Opinion&pgtype=Homepage

But still air sanitization provides long term reductions in infection and protects against other diseases which can be deadly as well. You can give a man a test, or you can teach him to maintain a sanitary building.

It always perplexes me when you bring this up. Is this actually one of the hold ups for expanded testing?

It just seems like an obvious and arcane detail for testing professionals.

It's like saying "we strongly support the transition from timing belts to modern timing chains in car engines."

Harvard has more money than God. I assume they are just paying for all of this testing themselves. If every college did even monthly testing of everyone who uses the campus, what would the payment model look like? Can we give presidents an estimate of how much this would cost? Would insurance cover some of it?

U of Illinois developed a saliva test that it wants to use to screen students in the Fall, but they are waiting for federal approval. I don't believe it uses pooling, but it also is supposed to be faster and use less materials than conventional approach.


Also may be worth noting that Major League Baseball is currently using pooled testing this season. Thousands of players, coaches and staff will be tested every other day, with results reported within 24 to 48 hours, and an isolation protocal for those testing positive.

The rollout had problems, but none seemed to be fundamental. Seems like about 3 percent of players were infected at intake. There were problems with delivery over the Fourth of July weekend that seemed to have more to do with scheduling practices to start on the Monday after.

Pooled testing is a band-aid for the lack of competent non-corrupt government leadership. How about pushing for competent government leadership?

The soda analogy is reasonable enough. However, the author cannot quite bring himself to blame our corrupt legislative system that insists we must have a deeply broken and corrupt for-profit health care system, and instead the author has that reflex of trying to blame a government agency that has its bosses appointed and budget set by the politicians.

In any case, to make the analogy complete: whenever the soda makers get a new opportunity, they collude to restrict the new products entering the market through patent shenanigans and buyouts, and the high-quality products have to compete with counterfeits, pop-ups, and contaminated products.

When manufacturing things, you can only increase your output at finite rates. Throughout history, whether you are the military making bullets or bombers for WWII, or Apple scaling up production on a new iphone...the fastest you can grow output is is double every quarter. Some quarters maybe a bit faster, and others a bit slower.

Before the pandemic, the US was doing about 3M PCR tests per year (mostly HIV). Simple math says 30M PCR tests per year is in range. But it also says 60M than 600M PCR test per year wont' happen until the next year (after a vaccine).

A rational testing strategy says "test everyone with symptoms, and test everyone that is visiting their doctor for other reason". That would need about 1B tests per year. And that's considered light.

In short, there's not a gov in the world that will be rolling out testing at 3X per capita per year.

Please explain. What is pooled testing? How does it allow us to test fewer people and know the status of every one?
From your first paragraph, it sounds like you want to test one person in a group of 20 and conclude that the whole group is clean.

You test a number of samples combined into one batch. The hope is, if you get a negative on the batch, you're done. After only doing one test.

Positives, on the other hand, lead to more work.

I work in the entertainment industry, and it appears that what they are going to be doing is pool testing. Crew and actors on a certain show will test 2-3 times a week during the roughly three months of, say, a TV show production. It would be an interesting population to study.

Alex makes good points, although I wonder if there might be logistical and data issues with pooled testing: each sample now has to be identified with 10 or 20 or whatever number of patients. If the data-storage procedures can handle that, then fine: any modern database can handle one-to-many relationships, but are the actual data tools used by hospitals and testings labs set up for that?

There is ongoing research, although apparently with mixed results, into something even more large scale and less intrusive than pooled testing: measuring coronavirus levels in sewage. This is not so much for tracking down who's sick and who isn't, but to track overall levels of infection.

Poop testing instead of pooled testing.


+1. Bureaucratic capacity has been overrated with this from the beginning. It's like startups: Ideas are a dime a dozen. Execution is everything.

You can get information on MLB testing here:
Its not clear if pooled testing is being used. MLB will sample Tier 1 people every other day and Tier 2 people multiple times a week. The samples are necessary but the lab could pool the samples and speed up testing considerably. I'm pretty sure the samples aren't pooled because there would be no need for Tiers and every other day of sampling. Of course the lab isn't going to suggest pooling because I'm pretty sure it built in some markup in the tests.
MLB could insist on it however, and either get more frequent tests or get more people (press and tv people) tested as well.

Comments for this post are closed