Pandemics Data & Analytics questioned the projections of the Covid-19 epidemiology models on which government based its lockdown decisions. Nechama Brodie wrongfully tries to dismiss it as a ‘narrow interest group’ that is ‘dishonest’ and ‘revisionist’.

In early May, I wrote a column highlighting the flaws, bugs and gross overestimates in Neil ‘Professor Lockdown’ Ferguson’s pandemic models developed at Imperial College, London, and wondered why South Africa’s epidemiological models were not subjected to academic peer review and public scrutiny.

Some time earlier (and unbeknownst to me at the time), a group of actuaries, data analysts, statisticians, medical doctors and economists formed a research group named PANDA, or Pandemics Data & Analytics.

On 4 May, they wrote to president Cyril Ramaphosa, warning him that life-years lost to the economic fallout of lockdown would likely dwarf the life-years saved from Covid-19 by the lockdown. Their full report was published a week later, estimating that South Africa’s lockdown will cause a loss of life-years 30 times greater than the loss of life-years it stands to prevent. The group subsequently published an open letter calling for the models used to arrive at Covid-19 death projections to be made public.

In mid-April, the government predicted that a two-week extension of the initial hard lockdown would ‘shift the peak to September instead of June or July, as was expected earlier’.

That has not happened. The peak, both of the number of cases and of Covid-19 deaths, occurred in mid-July. If the lockdown made any impact on the timing of the peak, it wasn’t much at all, even though that was a primary purpose of shutting down the economy and throwing millions out of work and into penury.

By mid-May, the government projected that compliance with lockdown regulations could push the peak out from early June to at worst mid-July, or perhaps to late-August. If you ignore the fact that still projecting a peak of early June, seven weeks into lockdown, is tantamount to an admission that those seven weeks achieved nothing at all.

Saying, with hindsight, that the peak did occur in mid-July hardly vindicates the government’s projections, which were all over the place. Reality was not inconsistent with the government’s early estimate of when a no-lockdown peak would occur.

The original death toll projections of between 87 900 and 351 000 deaths was quickly revised to between 120 000 and 150 000 deaths, and then to a lower bound of 45 000. The president from mid-April right up until the July peak maintained that between 40 000 and 50 000 would die by the end of the year.

Curiously, three months’ worth of new data, as infections, hospitalisations and deaths actually started to roll in, did nothing to budge the projected death toll.

With three months to go until the end of the year, and well over two months after the peak, the documented death figures stand at a third of that projection. It would require another one and a half times the number of deaths we have had in the first six months of the pandemic to reach even the lower bound of 40 000 deaths by the end of the year.

PANDA initially estimated that deaths were unlikely to exceed 10 000, and soon produced a projection of between 10 000 and 20 000 deaths. They criticised the government’s models for ‘grossly overestimating’ the likely impact of the pandemic.

Throughout, the real-world data followed PANDA’s projections far more closely than it followed the government’s projections.

Papers, please!

The fact that PANDA’s critique of the government’s statistics appear to have been correct irked Nechama Brodie, prompting her to pen a ponderous, 35-minute read entitled 2020, hindsight, which comes in at a hefty 8 600 words.

Brodie is a journalist who frequently touts her Ph.D. in journalism and media studies, and identifies herself as ‘Dr. N. Brodie’ on social media, even though it isn’t customary to do so outside an academic setting unless you’re a medical doctor. Her doctoral dissertation formed the basis for one of her books, Femicide in South Africa.

I mention all this to establish her credentials to be writing about Covid-19 and the epidemiological models upon which the government based its lockdown policy. That is, she has none. She is just another journalist.

And I mention that, because she makes a terribly big deal of the credentials of members of the PANDA group, which includes three actuaries, a statistician, a big data analyst, two medical doctors, a data scientist, a lawyer, an economist, an engineer, a geophysicist, and an expert in fundamental analysis and machine learning.

None of these fields, Brodie argues, qualifies them to critique the work of the real experts, who work for the government building models that are wrong.

Epidemiology, of course, is largely the application of mathematical methods of analysis to data. What the underlying data represents is neither here nor there. The methods of analysis and the techniques for constructing mathematical models are the same.

I would imagine actuaries, statisticians economists and data scientists are eminently suited to critique such models. Models are their bread and butter. It’s what they do.

But no, Brodie wants us to bow to the ‘experts’, and those experts apparently exclude anyone who is not a professional epidemiologist and working for the government.

Given that her deference is to government-selected experts, whose work was long shrouded in secrecy, hers is a particularly obsequious instance of the appeal to authority fallacy. While she is correct to say it is advisable to take the opinions of experts seriously, it doesn’t seem to occur to her that merely being an ‘expert’ does not place everything a person says above questioning by mere mortals.

Experts frequently disagree, and that is certainly the case here.

Expert opinion

Her only substantial claim to support the validity of government’s models is that some substantial share of the 42 000 excess deaths reported between early May and the start of September must have been attributable to Covid-19. She notes the correlation between Covid-19 infections and the excess mortality curve, without recognising that an equally suggestive correlation exists between excess mortality and lockdown.

In her expert opinion, it is implausible that these deaths could be attributable to any other cause, such as people being afraid to seek medical attention for non-Covid-related illnesses, or plain ol’ starvation.

Therefore, by claiming as fact something which is only highly speculative, she confidently concludes that the government projections were right all along.

The irony of this conclusion is all the more pointed given that she spends an inordinate amount of time on explaining why Covid-19 data is not reliable in the first place.

She launches into lengthy explanations of matters that are not ‘remotely new or surprising to anyone who has ever worked with mortality data’, as if she herself is an expert. Which, we have established, she is most certainly not.

Given her reverence for experts, perhaps she would be interested in the view of Professor Shabir Mahdi, former head of the National Institute for Communicable Diseases, who has a world-class CV as long as my arm.

Contrary to Brodie’s view that the early models were merely hamstrung by a lack of information, in late April he declared them ‘flawed’, ‘back-of-the-envelope calculations’, ‘based on wild assumptions’. According to Katharine Child, a journalist with an excellent reputation for critical thought and a skeptical approach to official propaganda, he noted that nobody was modelling  the collateral damage of the lockdown, and the impact of the economic meltdown on disease and death.

The fragility of white male opinions

Brodie also makes much of the fact that the most prominent social media voices criticising lockdown policies are, in her opinion, ‘middle-class white males’ with ‘poorly informed and self-important opinions’, and that PANDA consists of ‘mostly white males’.

Why someone’s race or gender has any bearing at all upon the validity of their work is a mystery to me. Imagine if I told you to ignore Brodie’s article not because it is has no substance, but because she is a woman or a Jew. Imagine if I critiqued the National Coronavirus Command Council not on the grounds that it was authoritarian and based irrational decisions on questionable data, but because its members were ‘mostly black’.

There would be outrage. I’d never work again.

Somehow, naked racism is okay when it is directed at white males. In academic circles, they call that ‘anti-racism’. When the Democratic Alliance embraces non-racialism, journalists of Brodie’s ilk call it ‘racism’.

No, judging someone’s work by the colour of their skin is racism, plain and simple. Judging their work by their gender is sexism. Brodie reveals herself as both a sexist and a racist, and in the same breath betrays that there is little of substance to her criticism of PANDA.

Self-contradiction

Brodie writes: ‘What do we know or learn from [revised government projections]? That the initial estimate or model outcome was changed because it was based on limited information – which was updated and amended once newer and better information became available. This is how estimates and modelling work, and there is nothing inherently ominous about a projection being updated or revised.’

When PANDA updates an early guesstimate of at most 10 000 deaths to a more properly modelled 10,000 to 20 000, however, Brodie considers this to be not just ‘revisionist’, but ‘dishonest’.

This is a self-contradiction. When one group revises their estimates based on newer, better data, there’s nothing to see, but when another does so, it’s dishonest. Really?

She points to the government’s mid-May hopes of pushing an early-June peak to between mid-July and late-August as evidence that the lockdown worked, but while accusing PANDA of cherry-picking, pointedly ignores the government’s mid-April hopes of pushing a June-July peak out to September, which suggests the lockdown achieved nothing.

Her headline about ‘2020 hindsight’, and her claim that ‘[w]e have already forgotten how bad it was’, is aimed at PANDA, as if they are only now making the claim that the government’s projections were exaggerated and the lockdown cure would be worse than the disease, instead of having been at it since the start of May, at least.

For Brodie to claim that PANDA’s projections amount to 20/20 hindsight is, not to put too fine a point on it, dishonest. So is her claim that PANDA calls the pandemic ‘some kind of hoax,’ or is somehow acting in bad faith.

Brodie claims that ‘it is important that people are allowed to criticise the government,’ but spends 8 000-plus words explaining why nobody other than the experts who work for the government have any credibility.

Presumably she’ll concede, then, that her own piece is just partisan rhetoric, that it is little more than a prejudiced smear against people who are trying to hold government accountable, and that she does not have ‘truth’ or even the ‘facts’ on her side.

The views of the writer are not necessarily the views of the Daily Friend or the IRR

If you like what you have just read, subscribe to the Daily Friend