After my last post expanding on some of the ways to best evaluate information and ways of knowing, people asked for more such discussions.

So let’s talk more about metaepistemology.

One of the things I noticed throughout the past year has been that a lot of my friends who had grown up in authoritarian or poor countries had a much easier time adjusting to our new pandemic reality. My childhood was intermittently full of shortages of various things. We developed a corresponding reflex for stocking up on things when they were available, anticipating what might be gone soon. That was quite useful for the pandemic. So was trying to read between the lines of official statements—what was said and what was not, who was sitting with whom on the TV, and evaluating what the rumor networks brought in. It turns out those are really useful skills when authorities are lying at all levels.

I commented on this:

zeynep tufekci @zeynep

It's been a weird year where a lot of my "Third World" habits and skills have been more relevant than I'd have liked: trying figuring out how to import medical masks for health-care workers, reading between the lines during press conferences of a misinformer-in-chief, shortages..

Others with similar backgrounds started reporting a similar “authoritarian muscle memory.”

Karol Karpinski @karolkarpinski

IRL friends know I make little of my origins, to say the least, but let's be honest: I've been feeling exactly the same way this past year. It's as if you have a little "failed state" switch in your head that gives you those superpowers when you turn it on.

zeynep tufekci @zeynep

It's been a weird year where a lot of my "Third World" habits and skills have been more relevant than I'd have liked: trying figuring out how to import medical masks for health-care workers, reading between the lines during press conferences of a misinformer-in-chief, shortages..

Karol Karpinski @karolkarpinski

@zeynep very weird! I don't even know how to properly call it... like authoritarian muscle memory!

🐝Readers on transit for Secretary Mayor Pete🐝 @Readers4Pete

I second this. My American husband marveled at how quickly I went into planning mode for filling the freezer and how to categorize and store food for a long term quarantine. Also how to plan 5-6 meals ahead because you have to defrost something for dinner tomorrow.

zeynep tufekci @zeynep

It's been a weird year where a lot of my "Third World" habits and skills have been more relevant than I'd have liked: trying figuring out how to import medical masks for health-care workers, reading between the lines during press conferences of a misinformer-in-chief, shortages..

When Trump got sick with COVID last October, there was a lot of speculation about how sick he was. We knew he was taken to the hospital, but that may have been out of an abundance of caution. He was, after all, the president, and there was indeed footage of him walking to the helicopter. When his medical team held a press conference, one detail stood out: he had been given dexamethasone—a steroid that has been shown to greatly reduce mortality, but only when the patient was severely ill. In the early stages of the disease, the result was the opposite: it increased risk and negative outcomes.

The reason for this flip is that the disease has two phases: the early phase is dominated by the “virus pathology,” and in the latter phase, it’s the immune system response that’s doing the damage.

There is a growing movement to approach COVID-19 as a two-phase disease: in the early phase, virus pathology dominates; and in the later phase, immunopathology drives disease. In thinking about COVID-19 this way, perhaps it is not surprising that dexamethasone offered no benefit to patients in the RECOVERY trial whose disease had not progressed to a stage necessitating respiratory support; indeed, the immunosuppressive effects of glucocorticoids at this stage of disease might hamper antiviral responses. It is in the later, hyperinflammatory phase of COVID-19 that the immunomodulatory effects of glucocorticoids are beneficial, perhaps by breaking the inflammatory feedforward loop, at least in some patients.

Dexamethasone is a synthetic glucocorticoid that’s used to suppress the immune system—a very useful drug if it’s the immune system response that’s driving the disease, but obviously terrible if it’s in the early phase, where we want the immune system to work hard in suppressing the virus replication. Hence, the guidelines recommend against using it for patients who do not require supplemental oxygen—something that happens around low 90s percentage oxygen saturation.

So what had actually happened to Trump? One obvious hypothesis was that he was given dexa because he was sicker than his doctors were saying, and it was an appropriate medical response.

Bob Wachter @Bob_Wachter

My @UCSF colleague Vinay makes great point – if T just had mild hypoxia (truly low of 93%), could support argument to hold steroids given data below that giving it too early may be bad (which makes sense physiologically). A guess: O2 was <90 & worrisome chest CT led to decision.

Vinay Prasad MD MPH @VPrasadMDMPH

Not only was there a pre-specified interaction between severity of hypoxia & dexamethasone benefit 👇; there was also interaction with timing of symptoms Both suggesting dex works in immune phase of disease Giving many Rx without good data not good med It can be detrimental https://t.co/Z6W6VWA0xo

The alternative idea was that he was a victim of “VIP medicine” syndrome where important personalities get aggressive “kitchen-sink” treatments: everything gets thrown at them, even if it's not to their medical advantage. Many raised this possibility on Twitter:

Vinay Prasad MD MPH @VPrasadMDMPH

Not only was there a pre-specified interaction between severity of hypoxia & dexamethasone benefit 👇; there was also interaction with timing of symptoms Both suggesting dex works in immune phase of disease Giving many Rx without good data not good med It can be detrimental

And the New York Times story about Trump’s COVID suggested this possibility as well:

Some experts raised an additional possibility: that the president is directing his own care, and demanding intense treatment despite risks he may not fully understand. The pattern even has a name: V.I.P. syndrome, which describes prominent figures who receive poor medical care because doctors are too zealous in treating them — or defer too readily to their instructions.
“You think you’re helping,” said Dr. Céline Gounder, a clinical assistant professor of medicine and infectious diseases at the N.Y.U. Grossman School of Medicine. “But this is really a data-free zone, and you just don’t know that.”
Still, based on the doctors’ account, Mr. Trump’s symptoms appear to have rapidly progressed since he announced early Friday morning that he had tested positive for the coronavirus.

Other outlets raised the same idea.

So which was it?

I listened to the news conference delivered by the doctors. Here’s what they said:

Aaron Rupar @atrupar

Q: Was Trump's oxygen level ever below 90? CONLEY: We don't have any recordings here of that Q: But was it ever below 90, here or at the White House? CONLEY: No, it was below 94 percent. It wasn't down in the low 80s or anything.

So here’s the text again:

Q: Was Trump's oxygen level ever below 90?

CONLEY: We don't have any recordings here of that.

Q: But was it ever below 90, here or at the White House?

CONLEY: No, it was below 94 percent. It wasn't down in the low 80s or anything.

So, here’s what I deduced at the time from the information we had:

zeynep tufekci @zeynep

@Bob_Wachter @UCSF Dr. Conley said it "never reached low 80s" and that they don't have recording below 90s "here" so via Kremlinology, a sadly appropriate method now, we can probably infer that it was probably mid-to-high 80s Friday night which sparkedgiving him oxygen and transport to Walter Reed.

Notice the pattern. He clearly adds the word “here” to answer whether Trump’s oxygen level had fallen below 90. So it must have fallen below that somewhere. He’s then asked whether it fell below that at the White House. He says that it was below 94 percent but "never reached low 80s."

Metaepistemology may be a fancy term, but it’s actually a mundane skill. Let’s think through its applications. The statements made by the doctor in charge of Trump’s treatment contained more information than he claimed--if one read them correctly., If one interpreted them by “reading between the lines” in a way that took into account who the person was, what their motivations were, and what their incentives and “borders” were.

A principle that’s often useful in these situations is that most deliberate misinformation from authorities—especially in places that are mid-range in terms of institutional trust and strict licensing—comes from omission, not saying the truth, rather than outright lying. That offers a way to get at the truth by trying to detect a picture, and looking at the parts that have been obscured, to make out the actual shape.

It’s like looking at this picture (which is actually an optical illusion from the San Francisco Exploratorium Museum) and using the columns to figure out that the relevant information: the people.

In other words, the “Baghdad Bob” syndrome—where a high-level official will outright lie about something that almost anyone can see for themselves is an obvious lie—is not just rare, it is not the best way to lie:

Sahaf's nickname, "Baghdad Bob," now denotes someone who confidently declares what everyone else can see is false--someone so wrong, it's funny. But when read beside the eventual cost of America's decade in Iraq, "Baghdad Bob" isn't so funny anymore.

And perhaps a key point here is that the difference between lies of omission—misleading by skipping relevant information—and lies of commission—outright lying—is not just that the latter is weak, it’s also that it’s harder for the person doing the misleading. It deprives them of their self-respect. And in countries like the United States, it’s not easy for a medical doctor at a respectable institution to be outright lying.

In fact, whenever cornered by reporters, where he could not directly mislead, Dr. Conley started invoking HIPAA—the law that protects patient privacy.

Aaron Rupar @atrupar

"There are HIPAA rules and regulations that restrict me in sharing certain things for his safety and his own health and reasons" -- Conley won't talk about what Trump's lung scans show

This was another obvious tell: privacy rules did not prevent Dr. Conley from sharing positive impressions of the president’s health, but were invoked when it came to questions that would have revealed the extent of the severity of the illness.

Hence, I concluded that the most likely explanation was not “VIP syndrome” but that, indeed, the President had faced severe illness.

Yesterday, we finally got actual reporting on this. Here’s the New York Times:

The people familiar with Mr. Trump’s health said he was found to have lung infiltrates, which occur when the lungs are inflamed and contain substances such as fluid or bacteria. Their presence, especially when a patient is exhibiting other symptoms, can be a sign of an acute case of the disease. They can be easily spotted on an X-ray or scan, when parts of the lungs appear opaque, or white.
Mr. Trump’s blood oxygen level alone was cause for extreme concern, dipping into the 80s, according to the people familiar with his evaluation. The disease is considered severe when the blood oxygen level falls to the low 90s.

Emphasis mine: and so there it was. Indeed, it was in the 80s.

There is often talk of teaching people “critical thinking” thinking skills, and that’s certainly something worth doing. A mistake, though, is to think that such critical thinking skills are independent of knowledge: that there is a recipe, or a way of interrogating conclusions, that can turn into “critical thinking.” In reality, the process by itself isn’t where the magic happens.

These do not seem complicated skills in some sense—and especially not in retrospect, once the actual answer is known. But they require more than parsing of words. The institutional operation, and the status and psychological incentives of the people, matter greatly to discerning the truth. Like most knowledge, this is more than “word games.” It is a mixture of sociology and psychology—if we are putting them into fields—but also involve probability: what’s the most likely outcome? What types of evidence would help tip the balance in which direction? How do these institutions operate? What are the personal and professional incentives of this particular person? And so on.

Critical thinking is not just formulas to be taught but knowledge and experience to be acquired and tested and re-examined, along with habits and skills that can be demonstrated and practiced. But there is no separating the “process” from the “substance”.

It may be a privilege to live in a society that does not always need official statements to be interrogated as such.But if the past few years have shown anything, that privilege is not something to be taken for granted.

Insight

Critical Thinking isn't Just a Process