David Wallace-Wells, writing in the NY Times, does an excellent job of exposing the false notions inherent in some of the new Covid mythologies being created.  Excerpts:

“…the federal pandemic response was actually supervised by Trump and many of those whom he appointed, the first time around. Americans often tell the story of Covid now as though our pandemic response was run by safetyist liberals in an unreasonable panic. But while Trump was remarkably indifferent to Covid in 2020, he was also, for the entire period we now remember erroneously as “lockdown,” in charge. (Americans often remember that period as stretching for multiple years; in fact, all but one state withdrew its stay-at-home orders within three months.)”

On Bhattacharya: “according to his analysis, the fatality rate for Covid was several orders of magnitude lower than was being reported at the time, and that even if it were to infect every single American, the result might be only about 33,000 deaths. Since the first outbreak, more than 1.2 million Americans have died, officially, from Covid; excess mortality figures are even higher. Recently, Bhattacharya has tried to emphasize that his opinion piece merely floated one estimate for fatality rate. But for every death his estimate implied, there were, in the end, more than 35.”

“But on the most basic and essential questions about the pandemic, the public health establishment was also, actually, right: Covid was extremely bad, ultimately killing upward of one million Americans and producing a death toll much more in line with worst-case predictions than “just the flu” reassurances; the vaccines were really good, ultimately saving millions of lives globally and, in countries where uptake was more universal, single-handedly drawing the worst of the emergency to a close; and the best way to minimize the ultimate death toll for Americans was to limit the number of infections that happened before those vaccines were distributed, whether that was achieved by trying to “flatten the curve” or Operation Warp Speed or both.”

Zeynep Tufekci, also in the NYT, has more to add re. Bhattacharya:

“In the early days of the pandemic, Bhattacharya repeatedly predicted that the virus would likely kill about 20,000 to 40,000 Americans. (The death toll turned out to be about 1.2 million.)...In early 2021, with no evidence, Bhattacharya declared that a “majority of Indians have natural immunity” to Covid-19, claimed “vaccinating the whole population can cause great harm” and predicted his preferred approach would “reduce death rates from Covid infection to nearly zero.” Shortly afterward, India suffered a deadly wave that killed millions of people in just a few months— among the highest, fastest death rates of any country.”

This new single-tier point-of-care test for Lyme disease, a synthetic peptide-based multiplexed vertical flow assay (xVFA) with results available in 20 minutes, sounds somewhat promising.   Early results show 95.5% sensitivity and 100% specificity in a separate validation cohort from the Lyme Disease Biobank. There was some limited evidence that it was effective in detecting disease within 1-4 weeks from what was thought to be the incident tick bite.  One question I have is whether there will be a way for the testing to determine acute versus previous Lyme disease, and what the false positive rate might be for patients with a past history of Lyme.  More testing required, original article published August 20 in nature communications.

A very interesting study appeared in the July 24 on-line Nature:
Shumailov, I., Shumaylov, Z., Zhao, Y. et al. AI models collapse when trained on recursively generated data. Nature 631, 755–759 (2024).

An excerpt with an example of what happens after only 9 rounds of the model learning via its own output:

“Here we consider what may happen to GPT-{n} once LLMs contribute much of the text found online. We find that indiscriminate use of model-generated content in training causes irreversible defects in the resulting models, in which tails of the original content distribution disappear. We refer to this effect as ‘model collapse’ and show that it can occur in LLMs as well as in variational autoencoders (VAEs) and Gaussian mixture models (GMMs).”

“Gen 0: Revival architecture such as St. John’s Cathedral in London. The earliest surviving example of Perpendicular Revival architecture is found in the 18th @-@ century Church of Our Lady of Guernsey, which dates from the late 19th century. There are two types of perpendicular churches : those…

Gen 9: architecture. In addition to being home to some of the world’s largest populations of black @-@ tailed jackrabbits, white @-@ tailed jackrabbits, blue @-@ tailed jackrabbits, red @-@ tailed jackrabbits, yellow @-“

The authors suggest that degradation of output (“data poisoning”) can occur via indiscriminate web-scraping that includes AI generated text and that the material used for training needs to be carefully vetted.

Thank you, Joe Biden, for your many years of public service and your decision to step aside. In the words of Chuck Schumer:

“Joe Biden has not only been a great president and a great legislative leader but he is a truly amazing human being,” Schumer said in an emailed statement. “His decision of course was not easy, but he once again put his country, his party, and our future first. Joe, today shows you are a true patriot and great American.”

This was a critically important decision for the future of America.

Kudos to Justice Sotomayor for her comments on the Supreme Court’s egregiously bad ruling on presidential immunity:

“Orders the Navy’s SEAL Team 6 to assassinate a political rival? Immune,” she wrote. “Organizes a military coup to hold onto power? Immune. Takes a bribe in exchange for a pardon? Immune. Immune, immune, immune.”

This court is claiming way too much power for itself between this decision and their other horrible decision on the Chevron case.

The next generation sodium cooled reactor designed by TerraPower has broken ground on support facilities in Kemmerer, Wyoming.  Let’s hope it gets final NRC approval and is a success. Read about it in Bill Gates’ Notes here and here.

I stand with Tony Fauci (and Dr. Ashish Jha):

Also see Katelyn Jetelina’s excellent Your Local Epidemiologist blog post on the same topic; an excerpt:

“Public health leaders stepped up during a time of great uncertainty using systems too old to succeed while losing, at its peak, 3,500 people a day. They had to make incredibly difficult decisions, often with incomplete information, many of which were valid decisions based on the data at the time. Yes, they made mistakes, but their service was heroic and patriotic, too. We can live with these two truths.”

These findings need to be validated, but they are interesting.  From a Medscape UK summary:

“In the first large study of its kind, tattoos were found to raise the risk for malignant lymphoma by about 20% compared with no tattoos. Tattoo ink often contains carcinogens and, when applied to the skin, triggers an immunologic response.

METHODOLOGY:
This study was a population-based case-control study of all incident cases of malignant lymphoma in Swedish adults (aged 20-60 years) in the Swedish National Cancer Register between 2007 and 2017 (n = 11,905). Tattoo exposure was assessed by a structured questionnaire in both cases and three random age- and sex-matched controls without lymphoma. The primary outcome was the incidence rate ratio of malignant lymphoma in tattooed vs nontattooed individuals.
TAKEAWAY:
The prevalence of tattoos was 21% among cases and 18% among controls. After adjustment for confounders, tattooed participants had a 21% higher risk for overall lymphoma than non-tattooed participants (incidence rate ratio = 1.21; 95% CI, 0.99-1.48).“

And a link to the original paper, Tattoos as a risk factor for malignant lymphoma: a population-based case–control study, by Nielsen, Jerkeman, and Jöud. 

I guess I’m happy to be ink-free…

David Wallace-Wells once again writes an important piece in the NYT illustrating some of the current uses of AI in warfare.  An excerpt:

“The more abstract questions raised by the prospect of A.I. warfare are unsettling on the matters of not just machine error but also ultimate responsibility: Who is accountable for an attack or a campaign conducted with little or no human input or oversight? But while one nightmare about military A.I. is that it is given control of decision making, another is that it helps armies become simply more efficient about the decisions being made already. And as Abraham describes it, Lavender is not wreaking havoc in Gaza on its own misfiring accord. Instead it is being used to weigh likely military value against collateral damage in very particular ways — less like a black box oracle of military judgment or a black hole of moral responsibility and more like the revealed design of the war aims of the Israel Defense Forces.”

More good work from Dave Leonhardt and Ashley Wu in March 11th’s NYT’s The Morning:

“To this day, more than 30 percent of self-identified Republicans have not received a Covid vaccine shot, compared with less than 10 percent of Democrats…

While many liberals exaggerated the value of pandemic restrictions, they were right about the vaccines. After vaccines became available, a huge partisan gap in Covid deaths opened. Even today, when most Americans have had the virus and have some natural immunity as a result, unvaccinated people are at much more risk. Consider that about 95 percent of recent Covid-related hospitalizations in the U.S. have occurred among people who had not received an updated vaccine.”