Streetwise Professor

In the film Pulp Fiction, moronic hoodlums Jules (Samuel L. Jackson) and Vincent (John Travolta) pick up a guy who had stolen a briefcase from the back of their boss Marcellus Wallace’s car. While driving him away, Vincent accidentally shoots him, leaving the back of the car splattered with blood and brains. In a panic, they drive to friend Jimmy Dimmick’s (Quentin Tarantino’s) house. Dimmick tells them his wife will be home in an hour and they can’t stay. In a panic they call Wallace, who calls in Winston Wolf. Wolf says: “It’s an hour away. I’ll be there in 10 minutes.” In 9 minutes and 37 seconds, Wolf’s car squeals to a halt in front of Jimmy’s house. Wolf rings the doorbell, and when Jimmy answers, Wolf says: “I’m Winston Wolf. I solve problems.” Within 40 minutes, Wolf solves Jules’ and Vincent’s problem. The car is cleaned up with the body is in the trunk, ready to be driven to the wrecking yard to be crushed.

The Imperial team that relied on Microsoft/Github to fix its code should have called Winston Wolf instead, because MS/Github left behind some rather messy evidence. “Sue Denim,” who wrote the code analysis I linked to yesterday, has a follow up describing what Not Winston Wolf left behind:

My favorite one is “a fix for a critical error in the random number generator.” In 2020? WTF? I remember reading in 1987 in the book Numerical Recipes by  William H. Press, Saul A. Teukolsky, William T. Vetterling and Brian P. Flannery a statement to the effect that libraries could be filled with papers based on faulty random number generation. (I’d give you the exact quote, but the first edition that I used is in my office which I cannot access right now. Why is that, I wonder?). And they were using a defective RNG 33 years later? Really?

“Algorithmic errors” is another eye popper. The algorithms weren’t doing what they were supposed to?

Read the rest. And maybe you’ll conclude that this was a mess that even Winston Wolf could have cleaned up in 40 days, let alone 40 minutes.

https://www.youtube.com/watch?v=oPoh2OpbyGs

By far the most important model in the world has been the Imperial College epidemiological model. Largely on the basis of the predictions of this model, nations have been locked down. The UK had been planning to follow a strategy very similar to Sweden’s until the Imperial model stampeded the media, and then the government, into a panic. Imperial predictions regarding the US also contributed to the panicdemic in the US.

These predictions have proved to be farcically wrong, with deaths tolls exaggerated by one and perhaps two orders of magnitude.

Models only become science when tested against data/experiment. By that standard, the Imperial College model failed spectacularly.

Whoops! What’s a few trillions of dollars, right?

I was suspicious of this model from the first. Not only because of its doomsday predictions and the failures of previous models produced by Imperial and the leader of its team, Neil Ferguson. But because of my general skepticism about big models (as @soncharm used to say, “all large calculations are wrong”), and most importantly, because Imperial failed to disclose its code. That is a HUGE red flag. Why were they hiding?

And how right that was. A version of the code has been released, and it is a hot mess. It has more bugs than east Africa does right now.

This is one code review. Biggest take away: due to bugs in the code, the model results are not reproducible. The code itself introduces random variation in the model. That means that runs with the same inputs generate different outputs.

Are you fucking kidding me?

Reproducibility is the essence of science. A model whose predictions can not be reproduced, let alone empirical results based on that model, is so much crap. It is the antithesis of science.

After tweeting about the code review article linked above, I received feedback from other individuals with domain expertise who had reviewed the code. They concur, and if anything, the article understates the problems.

Here’s one article by an interlocutor:

The Covid-19 function variations aren’t stochastic. They’re a bug caused by poor management of threads in the code. This causes a random variation, so multiple runs give different results. The response from the team at Imperial is that they run it multiple times and take an average. But this is wrong. Because the results should be identical each time. Including the buggy results as well as the correct ones means that the results are an average of the correct and the buggy ones. And so wouldn’t match the expected results if you did the same calculation by hand.

As an aside, we can’t even do the calculations by hand, because there is no specification for the function, so whether the code is even doing what it is supposed to do is impossible to tell. We should be able to take the specification and write our own tests and check the results. Without that, the code is worthless.

I repeat: “the code is worthless.”

Another correspondent confirmed the evaluations of the bugginess of the code, and added an important detail about the underlying model itself:

I spent 3 days reviewing his code last week. It’s an ugly mess of thousands of lines of C (not C++). There are hundreds of input parameters (not counting the fact it models population density to 1km x 1km cells) and 4 different infection mechanisms. It made me feel quite ill.

Hundreds of input parameters–another huge red flag. I replied:

How do you estimate 100s of parameters? Sounds like a climate model . . . .

The response:

Yes. It shares the exact same philosophy as a GCM – model everything, but badly.

I recalled a saying of von Neumann: “With four parameters I can fit an elephant, with five I can make him wiggle his trunk.” Any highly parameterized model is IMMEDIATELY suspect. With so many parameters–hundreds!–overfitting is a massive problem. Moreover, you are highly unlikely to have the data to estimate these parameters, so some are inevitably set a priori. This high dimensionality means that you have no clue whatsoever what is driving your results.

This relates to another comment:

No discussion of comparative statics.

So again, you have no idea what is driving the results, and how changes in the inputs or parameters will change predictions. So how do you use such a model to devise policies, which is inherently an exercise in comparative statics? So as not to leave you in suspense: YOU CAN’T.

This is particularly damning:

And also the time resolution. The infection model time steps are 6 hours. I think these models are designed more for CYA. It’s bottom-up micro-modelling which is easier to explain and justify to politicos than a more physically realistic macro level model with fewer parameters.

To summarize: these models are absolute crap. Bad code. Bad methodology. Farcical results.

Other than that, how was the play, Mrs. Lincoln?

But it gets better!

The code that was reviewed in the first-linked article . . . had been cleaned up! It’s not the actual code used to make the original predictions. Instead, people from Microsoft spent a month trying to fix it–and it was still as buggy as Kenya. (I note in passing that Bill Gates is a major encourager of panic and lockdown, so the participation of a Microsoft team here is quite telling.)

The code was originally in C, and then upgraded to C++. Well, it could be worse. It could have been Cobol or Fortran–though one of those reviewing the code suggested: “Much of the code consists of formulas for which no purpose is given. John Carmack (a legendary video-game programmer) surmised that some of the code might have been automatically translated from FORTRAN some years ago.”

All in all, this appears to be the epitome of bad modeling and coding practice. Code that grew like weeds over years. Code lacking adequate documentation and version control. Code based on overcomplicated and essentially untestable models.

But it gets even better! The leader of the Imperial team, the aforementioned Ferguson, was caught with his pants down–literally–canoodling with his (married) girlfriend in violation of the lockdown rules for which HE was largely responsible. This story gave versimilitude to my tweet of several days before that story broke:

"People he'd allowed to see his code." "Hey babe, want to swing around my place and see my code?" https://t.co/ZFWuNg7Upt

— streetwiseprof (@streetwiseprof) April 25, 2020

It would be funny, if the cost–in lives and livelihoods irreparably damaged, and in lives lost–weren’t so huge.

And on such completely defective foundations policy castles have been built. Policies that have turned the world upside down.

Of course I blame Ferguson and Imperial. But the UK government also deserves severe criticism. How could they spend vast sums on a model, and base policies on a model, that was fundamentally and irretrievably flawed? How could they permit Imperial to make its Wizard of Oz pronouncements without requiring a release of the code that would allow knowledgeable people to look behind the curtain? They should have had experienced coders and software engineers and modelers go over this with a fine-tooth comb. But they didn’t. They accepted the authority of the Pants-less Wizard.

And how could American policymakers base any decision–even in the slightest–on the basis of a pig in a poke? (And saying that it is as ugly as a pig is a grave insult to pigs.)

If this doesn’t make you angry, you are incapable of anger. Or you are an idiot. There is no third choice.

I think it’s fair to say that I was one of the early lockdown (AKA Karentine) skeptics. In one post (19 March) I asked rhetorically whether we were destroying society to save it. Evidence accumulates daily that the lockdowns were indeed wildly costly compared to the benefits, and in the case of the healthcare industry in the US in particular, it is only slightly rhetorically excessive to say that yes, we destroyed it to save it.

Recall that the primary justification for lockdowns was to prevent the healthcare system from being overwhelmed. With a few very localized (and highly publicized) exceptions, it never was, nor was it in any danger of being so. (And for those who claim this was the result of the lockdowns, there is mounting empirical evidence both within the US and internationally that variations in lockdown policy in terms of stringency and timing have trivial impacts on the trajectories of the epidemic.)

But to “save” the healthcare system, the non-coronavirus-targeted system was almost completely shut down. One major effect of this has been financial: hospital systems and individual physicians and nurses have been financially devastated. I can’t put my hands on the citation right now, but I read that something like 40 percent of the large decline in US GDP in March was due to reduced output in the healthcare sector.*

And there will be severe health consequences as well. “Elective” procedures are not unnecessary ones: avoiding or even delaying treatment of many conditions increases the risk of death, and the number of deaths. Such avoidance or delays can also lead to serious declines in quality of life.

These effects are part of the “unseen” that I mentioned in my earlier post. People will die because of the lockdowns, but there are no Lockdown Death Trackers. And those responsible for the lockdowns, or for cheerleading for them (which includes most major media outlets), have absolutely no incentive to create them. Because you can’t handle the truth, apparently.

It is sickly ironic (deliberate word choice) that many of the most strident defenders of lockdowns were, in normal times, almost equally strident in their insistence that the lack of adequate government provided or funded health care caused people to receive too little care, thereby leading to increased mortality and greater incidence of debilitating ailments. And no, don’t bother trying to reconcile these positions: it’s impossible.

I also note that the persistence of the lockdowns, especially in places like California, Michigan, Illinois, Wisconsin, and Minnesota gives the lie to the “flattening the curve to save the healthcare system” justification for them. In California in particular, but also in the other states (outside of perhaps Detroit, which has its own pathologies decades in the making), there has been a yawning gap between ICU capacity and utilization. Yet despite these what should be welcome data, and increasingly restive citizenries in these states, their governors refuse even to countenance relaxation. So it wasn’t really about the healthcare system, was it? Or if it was, it’s now about something else.

And what is that something else? Power.

Gavin Newsom’s tirade against those who dared to go to beaches is a particularly odious example. He basically treats the 40 million people of his state as would a control-freak father incensed at the audacity of his children in defying his authority. This is not a statewide lockdown. This is a statewide grounding–and if you break the rules again, sonnie, the grounding will be extended! Indefinitely! Not because it has any public benefit. But because you dared to challenge Governor Cartman’s–excuse me, Governor Newsom’s–authoritah.

And, of course, Google and Facebook in particular are willing–nay, eager–accessories to these mass deprivations of rights. They have become censors in chief, extirpating (sorry, “deplatforming” and “removing” are too weak) content that crosses the authoritahs’ official line.

The logic behind the lockdowns was always dubious, and they were NEVER based on reliable data. (And don’t get me started on the models.). But whatever logic there was, it is non-existent now. They have become self-perpetuating. Or rather, they are being perpetuated in order to perpetuate the powers seized by governors and local officials.

And if it destroys the healthcare system (not to mention the livelihoods of tens of millions of Americans)? Well, as a man who is apparently the role model for many governors once supposedly said, you can’t make an omelet without breaking some eggs.

*Here are the data. (See p. 9.) GDP fell $234b in the first quarter. Since GDP was rising rapidly–about 3.5 percent annualized–in Q4, and the lockdowns didn’t kick in until March, virtually all of this likely occurred in the last half of March. A pretty staggering decline in a few weeks.

Expenditures on health care fell $110b, almost exactly the 40 percent I mentioned in the original post. This contrasts to a $27b increase over Q4.

A little more than a month ago, when the China virus was first wreaking havoc in the oil market, Vladimir Putin and Igor Sechin demurred when the Saudis requested an extension of OPEC+ oil cuts. Furious, the Saudis retaliated by promising to ramp up oil production by more than 1mbpd, and prices truly cratered.

No problem!, said Vladimir and Igor. This is part of our master plan to destroy US shale! (Never mind such “logic” is right out of Economics for Mouth Breathers.) Furthermore, Russia has ample financial reserves, and its government can ride out a long stretch of low prices. (The Russian people? Ha! Demonstrating the state-centric nature of Russian policymakers.) The ruble crashed. No worries! In fact this is GREAT for Russian oil producers, because their costs are in rubles and the weakening of the currency cushions the impact of a lower dollar oil price on revenues. (The Russian populace? Again Vova and Igor DGAS.)

But despite the bold talk (which in some ways reminded me of the old story about G. Gordon Liddy holding his palm over an open flame to prove how he could take pain without cracking), the magnitude of the collapse, and the possibility that it would endure, obviously caused considerable consternation in the Kremlin.

The first indication was when the Russians claimed that the Saudis had taken their action to harm US oil producers, a complete flip for Russia’s bragging that that’s why they were taking the hard line.

In the past few weeks, Trump cajoled the parties to the bargaining table, and yesterday–after overcoming a Mexican holdout–the parties agreed to a deal. And what a deal it is, involving Russia agreeing to cut production by 2.5mbpd–almost 23 percent–an amount 4 times what they refused to accept a little more than a month ago. (I also note that it flies in the face of repeated Russian statements over years saying that it was operationally impossible for them to cut more than trivial amounts.)

The Russians had insisted on formal commitments from the US to cut. Trump–who brokered this deal–said that US production would fall due to price declines. (True dat–I’ll try to provide some estimates tomorrow.). The Russians said that wasn’t good enough (apparently demonstrating an ignorance of the limited ability of the US government to force output reductions)–they wanted mandated cuts. But in the end, the deal was done, absent any formal US commitments.

In other words, Putin and Sechin cried uncle. Hell, they tapped out like an MMA fighter in a chokehold. This makes a mockery of their bold front of a month ago. I think it reveals that they were truly shocked by the Saudi reaction, and were desperate to find a way out, but they didn’t want to show it. They needed some sort of fig leaf, and Trump gave them one.

Now this is being spun as some sort of Putin master plan to curry favor with Trump to gain negotiating leverage on other issues.

All together now: Suuuurrrre it is, Vova. Suuurrrre. Trump don’t owe you–you owe him.

I oppose the deal, despite the fillip it will provide to the US oil industry. Artificially boosting oil prices will harm already devastated consumer economies (including the US). Further, this will potentially rejuvenate OPEC post-crisis. Again that would be good for the US oil industry, but not for the US.

I also wonder whether Putin (and the Saudis) will come back later with proposals to get the US government to take measures to reduce output in a coordinated fashion. Just how that would be done is rather mysterious. The only plausible measure that I could think of would be some sort of tax (either on production, or on exports). But this would require Congressional action–it’s not something Trump or any president could negotiate unilaterally–and I don’t see that happening.

But that’s for the future. In the present, Putin and Sechin reaped what they sowed. They thought they could force the Saudis into bearing the brunt of supporting oil prices, while they could continue to produce at capacity. The Saudis responded like it was 1986, and the resultant collapse in prices brought the brilliant schemes of the KGB masterminds crashing down. In the end, they had to capitulate. No matter how hard they try to spin it, it’s obvious that they did.

The Marine Corps credo is that every Marine is a rifleman. This credo reflects, and is reflected in, the Corps’ history of and excellence at close combat. From Belleau Wood to Guadalcanal to Tarawa to New Georgia to Peleliu to the Marianas to Iwo Jima to Okinawa to Inchon/Seoul to Chosin to I Corps (in Vietnam) to Kuwait and to Iraq and Afghanistan, the Marines have closed aggressively with the enemy and killed them, often at very high cost to themselves. Aggressive close quarter combat has been the Marine way of war for more than a century, and they think (with reason) that they do it better than anybody, ever.

The new commandant of the Marine Corps, General David Berger, has announced a radical new vision that is diametrically opposed to this historical tradition. Rather than close and kill, General Berger aims to reshape the force in order to permit it to operate in enemy held regions, and attrite the enemy’s air and naval forces with fires, primarily anti-ship and anti-aircraft missiles.

And for enemy number one, read “China.” Enemy number two? There is no enemy number two. This is all about China, and the South China Sea.

This would involve a dramatic change in Marine force structure. Fewer infantry battalions. Less tube artillery–but more missiles. No tanks. Zip. Zero. Nada.

Somewhat unaddressed in Berger’s vision document is how the “small Marine forces” that he envisions will “deploy” to islands “in close and confined seas in defiance of adversary long-range precision ‘stand-off capabilities.’” Presumably they will have to fight their way in–just like they did across the Pacific 1942-1945. How this is to be done remains very unclear, especially if close combat capability is reduced.

One of the most striking things about this document is the laser-like focus on China. In some respects, this is encouraging, because China is and will remain the primary threat to the US and US interests. And the Chinese anti-access/area denial plan of creating strategic depth by contesting a ring of island defenses (and indeed, even building the islands) bristling with missiles does require a major shift in US doctrine in response.

The document also demonstrates an admirable appreciation of the complementarity between air and naval forces, especially in the vast Pacific theater. Berger’s vision clearly entails close cooperation between the Navy and Marines to fight and dominate a powerful enemy in the western Pacific.

But one thing that history has demonstrated is that you often don’t get the war you plan for–in part because that’s not the war your enemy wants to fight precisely because that’s what you want him to do. Having a force specialized to execute a single operational concept provides little capability to fight other kinds of wars. And one thing that has helped the Marines survive the budget wars of the 20th and 21st centuries is its demonstrated flexibility in carrying out myriad different missions, from fighting furtive guerrillas in Central American jungles to digging out entrenched Japanese on Pacific islands to making hell-for-leather armor and infantry attacks across desert landscapes in the Middle East. Berger’s proposed force will not be able to perform such varied missions, or at least will have far less capability to do so. Is this a wise choice? Tough call.

One challenge Berger will face is in DC. I seriously doubt the appetite of the Pentagon, or especially the Congress, to embrace and fund such a dramatic transformation.

Another challenge will be from the Corps itself. This proposal is at odds with Marine tradition and self-image and culture. Closing with the enemy and killing him is what Marines do. More importantly, it’s what they believe they do better than anyone in the world, or anyone in history, for that matter. Remaking Marines from riflemen into missileers involves more than swapping out old weapons for new, or writing new operational manuals. It involves inculcating a whole new mindset. This is difficult to do in any organization, but is particularly difficult in one with such a deep commitment to and a well-justified pride in a traditional way of waging war.

You are seeing a lot of covid-19 numbers thrown around. Virtually all of those numbers are bullshit.

The death rates are bullshit. In a given country, there is considerable subjectivity regarding how deaths are qualified. The Great Scarfini* (Dr. Deborah Birx) pretty much let that cat out of the bag when she acknowledged that not only are the decedents who test positive (regardless of other co-morbidities) declared as covid-19 deaths, but those who have some colorable connection to covid-19 (clinical presentation, exposure to someone who tested positive) are declared to be covid-19 deaths.

It is likely that hospitals and physicians–and politicians–have an incentive to attribute deaths to covid-19. These incentives can be financial (a hospital could get greater compensation from covid victim than someone dying of something else) or power (death numbers are being used to justify draconian restrictions).

Further, different countries use different methods to count deaths.

What we are really interested in is people who would not have died but for covid-19. The official death statistics do NOT do this. And the fact that virtually all of the dead are aged and/or have multiple serious health problems, a but for attribution is dubious even in the presence of a positive test.

The only rigorous way to estimate these but for deaths is excess deaths (i.e., deaths in excess of expected deaths, conditioning on time of year, demographics, etc.). And preferably excess deaths from respiratory illness (or at least excess deaths from non-accidental causes). This is a good template for the analysis. This also presents some good cross-country data, which shows that in Italy and Spain there is evidence of excess deaths. Elsewhere? Not so much. Of particular interest is Sweden, which has implemented mainly voluntary social distancing measures, to the hysterical response of those deeply invested in mandatory lockdowns.

Do this for a variety of jurisdictions (countries, states in the US) and you would have enough cross-sectional and time series variation to do some real analysis that could provide reasonable support for policy decisions..

The case numbers are bullshit, at least if you want to measure infection rates. As I’ve been saying for weeks, there are so many selection biases that the numbers tell you NOTHING about the prevalence of the virus in the population, either at a point in time or crucially over time. Indeed, the CDC guidelines could be titled “How to Produce a Wildly Biased Sample”:

This testing protocol could be justified on clinical and diagnostic grounds, but it is a disaster from the perspective of generating data that is useful in shaping policy.

Further, trends in positive test numbers is driven to a considerable degree by . . . a greater number of tests.

The graphs that you see depicting trends deaths or cases across countries over time are bullshit. They are bullshit because the inherit all the flaws of the data discussed above (exacerbated by the fundamentally different data reporting methods across countries), and they almost fail to adjust for population size or demographic characteristics.

Chinese numbers are obviously bullshit. No need to elaborate this point.

The models that are being used to drive (or at least justify) lockdowns are bullshit. Their predictions went from apocalyptic to well, a small fraction of apocalyptic. Sometimes between one day and the next. Models should be evaluated on predictive accuracy. The predictions of these models have proved to be excessively pessimistic, i.e., bullshit.

And don’t buy the line that the lockdowns reduced the death tolls. For one thing, many of the models’ predictions included the effects of social distancing–and still came out way too high. For another, many countries’ death and case rates (above caveats apply) peaked before the lockdowns could have had any effect.

I keep hearing the IHME model referred to as the “top model.” Who says? On what basis? Basically because somebody else said it. And oh, Bill Gates is somehow involved. So that claim is bullshit too.

Also be very suspicious that the modelers are very opaque. We don’t see their assumptions or their methods. Notoriously, the most influential modeling team (at least initially) that did more than any to spark the panic, has not released its modeling code.

At least the honest modelers admit that social isolation and shutting down the economy doesn’t change the integral under the curve (i.e., the total number of deaths) but merely the time pattern of those deaths. And some epidemiologists claim that extending the period of time before the burnout may result in a higher number of total deaths.

But even putting that possibility for a higher total toll aside, the argument is made that it is necessary to “flatten the curve” in order to reduce the burden on the healthcare system. Well, one thing the models vastly overpredicted hospitalization/ICU visits as well. And I have yet to see any evidence of systematic shortages of ICU beds/ventilators. Yes, there are hotspots. But that just means that we need to understand the hotspots–and the non-hotspots–better.

Along those lines, I can’t say the numbers on ICU utilization are bullshit–because the numbers are largely non-existent. Instead we’ve had anecdotal journalistic (i.e., “if it bleeds it leads”) accounts that provide no objective quantitative standard by which to evaluate how binding the constraints are in the healthcare system.

But again the issue is cost-benefit. Basically what lockdowns do is discount future deaths/cases relative to present deaths/cases (since they accept an approximately equal number of future deaths for each death that does not occur today). And the discount rate is huge. We are losing trillions of dollars in lost output/income to push some deaths into the future. The interest rate is astronomical. Put differently, we are paying an immense price to kick the can down the road.

I understand the the supply of ICU beds, ventilators, physicians and nurses is pretty inelastic over the short run. But even given pretty substantial inelasticity, it would be far more efficient to throw billions at expanding capacity in the short run than to sacrifice ~25 percent of world income to reallocate the deaths over time. Capacity is not a fate. It is a choice.

And the fact that well into the crisis the foretold capacity disaster in hospitals has not been realized, the additional capacity required may well be quite small.

There is also the issue of how much the temporal pattern of deaths will really change. This depends on a variety of factors, including when the virus first spread and its virulence. The more we learn, the more likely it is that the virus has been spreading since late-fall/early-winter 2020. Which means that the lockdowns are reactive, not proactive, and that they have little impact even on the time pattern of deaths let alone the number: they are the proverbial locking the barn door after the horse done bolted.

In brief: our betters are destroying futures based on bullshit data. It’s as simple as that. And they are vastly increasing their power as a result, so they are destroying freedoms too.

In an earlier post I said that we have to grasp the nettle and decide what price are we willing to pay to save a life (usually of an aged, ill person). But it’s actually worse than that. It is very likely that the real question is: how much are we willing to pay to defer a death (of such a person) a few months? The cost that those who govern (or rule) us (and those who support them) are apparently willing to pay is astronomical.

*Dr. Birx is always adorned with a scarf. On my first trip to NYC in 1978, when NYC was near its nadir, I saw an obviously psychotic individual dancing on near Grand Central Station waving around a long scarf. Every once and a while he would shout “I AM THE GREAT SCARFINI” and then start dancing again.

Ex-Global Super-Regulator on Lunch Break asked me to opine on the relief of Captain Crozier from command of the USS Theodore Roosevelt, CVN-71. At present, I don’t have the information to make a firm judgment one way or the other. Depending on circumstances that I (and anyone else not in the Department of the Navy or Captain Crozier himself) cannot observe, I can see things going either way. (And I note, that even if I was a principal, the lessons of Rashomon should be kept in mind.)

Based on what is in the public record, the presumption must be that Captain Crozier’s actions in either leaking his letter, or disseminating it in a way that was almost certain to result in its leaking, were highly irregular, and prejudicial to good order and discipline.

But this is a rebuttable presumption. The problem is that the public does not have the information to rebut it, or to not do so. I can imagine realistic scenarios that cut either way.

I can imagine a scenario in which Captain Crozier had reasonable grounds to believe that the Navy Department (and the Pentagon generally) was insufficiently sensitive to the acute situation on the Roosevelt, and felt compelled to take extraordinary actions in order to make them aware. Bureaucracies frequently prefer to suppress information (cf. the CCP), or are so used to operating according to routine that they are incapable of handling the non-routine.

That said, if that was Captain Crozier’s belief, he should have resigned his command (and his commission) and stated forthrightly the reasons for doing so. Engineering a leak is unmilitary, and reeks of political calculation. I understand that is the way that DC plays the game, but that doesn’t make it honorable to play by corrupt rules.

I can also envision a scenario in which Captain Crozier’s understandable focus on his ship and his crew blinded him to measures that his superiors were taking to balance the interests of the crew and operational and strategic interests of the United States.

It is legendary that the first question that presidents ask during an international crisis is: “Where are the carriers?” They are the primary and most rapidly deployable asset in such crises. Making one unoperational is a weighty decision with major consequences, especially in fraught times: note China’s aggressiveness in the South China Sea at the very moment the world is gripped with the coronavirus panic.

Indeed, Captain Crozier alluded to this when he said in his letter that: “If required the USS Theodore Roosevelt would embark all assigned Sailors, set sail, and be ready to fight and beat any adversary that dares challenge the US or our allies.” If only it were that easy. It has taken days to disembark the crew and accommodate them in Guam. It would take that many days, or more, to reembark them and make the ship operational. Remember Kipling’s “unforgiving minute”? Patton did–and rightfully castigated his superiors for failing to do so.

Capt. Crozier then went on to say that conditions on the ship conflicted with CDC guidelines. But CDC (civilian) guidelines are created to address different situations with different trade-offs than the military must deal with.

And that is the fundamental tension (i.e., trade-off) here. Operational availability of one of the US’s most important military assets and a health risk to its crew. Captain Crozier understandably emphasized the latter. His superiors had to give much more weight to the former than did the Captain.

We don’t know what measures the Navy Department/Pentagon were taking to try to balance these very difficult considerations. Absent such information, I for one reserve judgment.

This episode implicates larger issues that I have mentioned several times.

The first is the shocking lack of data upon which decision makers have to act. What is the real health risk to the crew (which consists primarily of young, healthy individuals)? We don’t know. Why are we ignorant? In large part because of a failure to implement a random testing regime (which I outlined weeks ago) that would produce more reliable information. It is appalling that months after the risks began to crystalize that we still have to rely on data plagued with severe biases inherent in the method of collecting them.

The second is the grim calculus of trade-offs, never mind whether we have the information to make informed trade-offs. Of course we want to save lives–but not at infinite cost. To reprise a previous post, we don’t want to destroy the country to save it.

This is a grim calculus that wartime militaries must always face. They say there are no atheists in foxholes–but at the same time, there have to be utilitarians in headquarters.

Mind you that I do not write as someone who reflexively defers to command authority. The fact I chose not to remain in the Navy is the chief exhibit in that case: as I often say (including under oath in depositions) I left Navy because that’s where I learned I have issues with authority. Further, when a plebe at Navy, my final essay in the leadership course argued that disobedience of orders was justified in some circumstances: this did not go over well! (Which was a big part of my practical education that informed my decision to leave.) So I totally get that Captain Crozier may well be in the right. But I also get that he might not be. But what I get most is that I don’t know–nor do you–whether he is or not.

There is an ongoing outbreak of Covid-19 on the Nimitz class carrier, USS Roosevelt. The outbreak is severe, and today the CO, Capt. Brett Crozier, wrote an impassioned letter requesting onshore quarantine of the entire crew.

The first criticism that Captain Crozier raises is “Inappropriate focus on testing.” Crozier objects that tests provide little information: given the close proximity of those on board, they have presumptively been exposed, and should be isolated. Further, Crozier quantifies a relatively high rate of false negatives.

The captain is certainly correct regarding what is his primary responsibility–his ship and crew. But testing on the Roosevelt could provide invaluable information that could lead to far better policies in the United States, and the world at large. From a larger perspective, the opportunity for testing on the Roosevelt is something that cannot be allowed to slip away.

As I have noted repeatedly here, and on Twitter, policy is currently based on incredibly flawed data. In fact, the most useful piece of data is from a cruise ship Diamond Princess. The Roosevelt could provide a far bigger sample, and one that contains valuable information about the impact on non-elderly, relatively healthy individuals.

Even one of the things that Captain Crozier objects to–the presence of false negatives–is important. Quantifying that rate can provide information that greatly improves the inferences that can be drawn from other samples (apropos my earlier Bayes Rule post).

I understand that there are myriad competing considerations here. The health of the crew. The operational readiness of one of the most important combatants in the US Navy. Operational safety–e.g., who is going to operate the reactors and ensure that other systems are maintained properly even if the ship is not deployed? (You don’t leave a CVN parked in the driveway for a few weeks.)

Among those competing considerations, from Captain Crozier’s perspective, testing is indeed a near irrelevance. But it is extremely relevant for informing how we deal with the crisis around the world. The social value of this data is great indeed. I hope that those in the Pentagon, and in the administration, find a way to address Captain Crozier’s concerns while at the same time seizing on this opportunity to generate data that could save thousands of lives and trillions of dollars.

Coda: Another benefit here is that by the nature of the military, there would be excellent data at hand on virtually any interesting covariate you can think of–age, health conditions, socioeconomic background, etc. Combining BUPERS data with testing and clinical data from CVN-71 could provide a plethora of actionable insights.

In the aftermath of the last crisis, I played the role of Clearing Cassandra, warning that in the next crisis, supersizing of derivatives clearing would create systemic risks not because clearinghouses would fail, but because of the consequences of what they would do to survive: hike initial margins and collect huge variation margin payments that would suck liquidity out of the system at the same time liquidity supply contracted. This, in turn, would lead to asset fire sales, that would distort asset prices which would lead to further knock-on effects.

I wrote a lot about this 2008-2012, but here is a convenient link. Key quote from the abstract:

The author also believes that the larger collateral mandates and frequent marking‐to‐market will make the financial system more vulnerable since margin requirements tend to be “pro‐cyclical.” And more rigid collateralization mechanisms can restrict the supply of funding liquidity, and lead to spikes in funding liquidity demand that can reduce the liquidity of traded instruments and generate destabilizing feedback loops. 

Well, the next crisis is here, and these (conditional) predictions are being borne out. In spades.

Here’s what I wrote a few days ago as a contribution to the Regulatory Fundamentals Group newsletter:

In the aftermath of the last crisis of 2008-2009, G20 nations decided to mandate clearing of standardized OTC derivatives transactions.  The current coronavirus crisis is the first since those reforms were implemented (via Dodd-Frank in the US, for example), and this therefore gives the first opportunity to evaluate the performance of the supersized clearing ecosystem in “wartime” conditions.  

So far, despite the extreme price movements across the entire derivatives universe–equities, fixed income, currencies, and commodities (especially oil)–there have been no indications that clearinghouses have faced either financial or operational disruption.  No clearing members have defaulted, and as of now, there have been no serious concerns than any are on the verge of default. 

That said, there are two major reasons for concern.

First, the unprecedented volatility and uncertainty show no signs of dissipating, and as long as it continues, major financial institutions–including clearing firms–are at risk.  The present crisis did not originate in the banking/shadow banking sector (as the previous one did), but it is now demonstrably affecting it.  There are strong indicators of stress in the financial system, such as the blowouts in FRA-OIS spreads and dollar swap rates (both harbingers of the last crisis).  Central banks have intervened aggressively, but these worrying signs have eased only slightly.  

Second, as I wrote repeatedly during the debate over clearing mandates in the post-2008 crisis period, the most insidious systemic risk that supersized clearing creates is not the potential for the failure of a clearinghouse (triggered by the failure of one or more clearing members).  Instead, the biggest clearing-related systemic risk is that the very measures that clearinghouses take to ensure their integrity–specifically, frequent variation margining/marking-to-market–lead to large increases in the demand for liquidity precisely during circumstances when liquidity is evaporating.  Margin payments during the past several weeks have hit unprecedented–and indeed, previously unimaginable–levels.  The need to fund these payments has inevitably increased the demand for liquidity, and contributed to the extraordinary demand for liquidity and the concomitant indicators stressed liquidity conditions (e.g., the spreads and extraordinary central bank actions mentioned earlier).  It is impossible to quantify this impact at present, but it is plausibly large.  

In sum, the post-2008 Crisis clearing system is operating as designed during the 2020 Crisis, but it is unclear whether that is a feature, or a bug.  

It is becoming more clear: Bug, and the bugs are breeding. There have been multiple stories over the last couple of days of margin calls on hedging positions causing fire sales, with attendant price dislocations in markets like for mortgages. Like here, here, and here. I guarantee there are more than have been reported, and there will be still more. Indeed, I bet if you look at any pricing anomaly, it has been created by, or exacerbated by, margin calls. (Look at the muni market, for instance.)

But those in charge still don’t get it. CFTC chairman Heath Tarbert delivers happy talk in the WSJ, claiming that everything is hunky dory because all them margins bein’ paid! and as a result, derivatives markets are functioning, CCPs aren’t failing, etc.

This is exactly the kind of non-systemic thinking about systemic risk that I railed about a decade ago. Mr. Tarbert has a siloed view: he is assigned some authority over a subset of the financial system, sees that it is working fine, and concludes that rules regarding that subset are beneficial for the system as a whole.

Wrong. Wrong. Wrong. Wrong. WRONG.

You have to look at the system as a whole, and how the pieces of the system interact.

In the post-last-crisis period I wrote about the “Levee Effect”, namely, that measures designed to protect one part of the financial system would flood others, with ambiguous (at best) systemic consequences. The cascading margins and the effects of those margin calls are exactly what I warned about (to the accompaniment a collective shrug by those who mattered, which is why we are where we are).

What we are seeing is unintended consequences–unintended, but not unforeseeable.

Speaking of unintended consequences, perhaps one good effect of September’s repo market seizure was that it awoke the Fed to its actual job–providing liquidity in times of stress. The facilities put in place in the aftermath of the September SNAFU are being expanded–by orders of magnitude–to deal with the current spike in liquidity demand (including the part of the spike due to margin issues). Thank God the Fed didn’t have to think this up on the fly.

It also appears that either (a) the restrictions on the Fed imposed by Frankendodd are not operative now, or (b) the Fed is saying IDGAF so sue me and blowing through them. Either way, such liquidity seizure are what the Fed was created to address.

I’ve expressed considerable skepticism about relying on test data to craft COVID-19 (AKA CCPVID-19) policy responses. This note formalizes the basis for my skepticism. Testing data would provide an accurate measure of the prevalence of severe infection if (a) the tests had low rates of false positives and false negatives, and (b) testing was random. Neither condition is remotely correct. Meaning that the test-based statistics are an extremely poor guide for policymakers, and a particularly dubious basis for driving the world economy into a depression, at the cost of trillions of dollars.

So what should we look at? If this is a particularly prevalent, virulent, and deadly respiratory disease, it will result in elevated levels of hospital admissions or physician visits for respiratory illness, and elevated levels of death from respiratory causes. That’s what we should be looking at. Or more to the point, what policymakers should be looking at. Is this a particularly deadly and widespread disease? If it is, it will have measurable effects on mortality and hospital admissions.

The CDC does collect data on influenza. Unfortunately, many of the statistics condition on a positive influenza test. For example, hospital admissions with a positive influenza test. That is not helpful, because we are focused on something other than the influenzas the CDC tracks. But the CDC does report deaths from influenza and pneumonia. That is more useful, as a deadly new respiratory illness should lead to higher pneumonia death rates.

Through last week, these data demonstrate little elevation on a national or regional basis. There was a spike in deaths above the “threshold” level early in 2020 (where the threshold basically is at the 5 percent significance level above the seasonally adjusted baseline), but subsequently it converged almost back to the baseline:

It is particularly interesting to compare 2019-20 with 2017-18. Heretofore, 2019-20 compares very favorably to that year, and even to less extreme years 2016-17 and 2018-19. Through now, in other words, 2019-20 does not look at all unusual.

The CDC also tracks data on those seeking medical treatment for flu-like symptoms: these data do not require a positive influenza test, and thus should reflect people suffering flu-like symptoms caused by something other than the flu. These data show somewhat higher levels for 2019-20 compared to previous years (except for 2017-18, which was much higher), but not extremely so. The main worrying aspect to the 2019-20 data is that they do not appear to be declining as rapidly with the approach of spring as in prior years. But the data do not exhibit a huge spike–they are just declining less rapidly than in prior years.

Yes, these data are backwards looking. I can imagine scenarios, such as the late introduction of CCPVID-19 into the US, which would mean that the wave of deaths/illness would not be manifest in the data, as it is still to come. But there are indications that the virus has been on the loose in the US at least since mid-January, and given its existence in China no later than mid-November, it could have been present in the US prior to mid-January. If it is indeed highly contagious and deadly, it should be leaving tracks in the mortality data.

It would be highly informative to have such data for other countries. I am not aware of it in as accessible a form as is provided by the CDC. If anyone can point me to it, that would be greatly appreciated.

You might argue that I am whistling past the graveyard. All I can say is that the data that alarmists point to is highly unreliable (and inherently so), and the reliable data as of yet demonstrate nothing out of the ordinary on the dimension that really matters–people dying from respiratory ailments.

What I can say with considerable confidence is that policymaking is driven by flawed data, and that there are types of data that would be more informative, and which are not infected by (deliberate choice of words) the problems inherent in the flawed data that dominates public discourse, and apparently dominates public policymaking. Produce that data. Disseminate that data. Make sure policymakers are aware of it, and are aware of the deficiencies of the data we hear about 24/7.