Predicting the future is a tough racket. If people were any good at it, we’d all be multiple lottery winners. Since we aren’t, that’s a pretty good indication that we have the ability to guess what will happen, but possess no clairvoyance. That brings us to today, when the country and the world are shut down over what is expected to happen with the coronavirus. Like everything else, the predictions aren’t panning out. Were this just $2 wasted on a lottery ticket it wouldn’t matter. But are we hurting millions of people and spending trillions of dollars on nothing more than a hunch based on computer models that are wildly inaccurate?
There’s no question the coronavirus is deadly and dangerous, but is it going to kill hundreds of thousands in the US and millions around the world? We have no idea. All we have are guesses. But if you’re going to destroy the economy and severely restrict people’s lives, you’d better have some damn good information backing up your actions. There are questions about the computer models being used to impose just that.
One prominent model in the shutdown is from the University of Washington, from their Institute for Health Metrics and Evaluation (IHME), funded by Bill and Melinda Gates. How closely does what they predicted track with reality so far? Turns out, not so well.
While the reporting data from some states are lagging, others have provided information that calls into question the validity of the whole model, and with it, all the actions taken by government.
On April 4th, for example, the IHME model predicted there would be between 120,963 and 203,436 Americans requiring hospitalization, with the average of that range being 164,745. In reality, there were 18,998.
Before your jaw goes through the floor, it must be noted that not all states have reported their numbers, so the actual total is incomplete. The largest states have reported, so the missing numbers from places like Kansas or even Michigan will add to that total, but still will not get it anywhere near the projections.
That’s just hospitalizations, when it comes to intensive care beds, the difference between the projections and reality show an equally large gap. The average projected ICU beds needed on April 4th was 31,057; the reality was 4,686. That’s across the country, not just New York. Even after factoring in the caveats about missing states, that’s a miss akin to swinging and striking out on a pickoff to second base.
So what other models are federal and state authorities using that justify the actions they’ve taken? The simple, horrifying answer is we just don’t know. The government isn’t sharing their models and data with the public, which is odd. Mitigation efforts like social distancing and the like may be a factor, but these models supposedly took that into consideration.
Are the modelers making assumptions based on the horrific numbers out of places like Italy? That doesn’t make much sense when you consider the difference between our countries – quality of care, number of ICU beds, average age of citizens, etc.
As of this writing, there are 7,810 deaths in the United States attributed to the pandemic. More than 7,000 Americans die every single day, on average, of everything. Over the course of this pandemic, this has amounted to a noticeable but still mild uptick above that average over the course of it. Add to that the directive from the Centers for Disease Control and Prevention to count any death tangentially related to COVID-19, and you have to wonder if that number actually holds. That directive reads, “COVID-19 should be reported on the death certificate for all decedents where the disease caused or is assumed to have caused or contributed to death.” (The emphasis in bolding is theirs.)
So what is going on? Where is the truth? Are we on the verge of a massive crush of cases and deaths, or are we destroying ourselves over the equivalent of tarot card reading gone wrong? I don’t know, but I do know we need answers and we need them now.
If we had a media capable of doing their jobs, Dr. Anthony Fauci and Dr. Deborah Birx would be pressed on all of this, hard, at the White House daily briefing. Since we don’t, we get Jim Acosta pulling months old quotes from President Trump attempting a “gotcha” moment that CNN’s audience and Jim’s ego craves. So-called journalists are attacking Fox News for allegedly spreading misinformation, even though they were sounding warning bells back in January, while the rest of the media were busy calling the president a racist for banning travel from China and wasted a week arguing with Republicans over how it was supposedly racist to point out a virus that originated in China originated in China.
Predicting the future is near-impossible. Shutting down a planet based on predictions is crazy, especially when the predictions aren’t coming true in real-time. If in the next week, these numbers from the models don’t start to jive with reality, President Trump needs to seriously consider not only reopening the country but also firing everyone involved in advising him based on those models. If they do start to line up, then may God have mercy on everyone.
Coronavirus is serious and deadly, especially to vulnerable people, but just how serious and deadly is one of many questions we need an answer to, and we need it now.
1 comment:
After lies, damn lies & statistics, what might serve as an overwhelming encore? That’s easy: Predictive Modeling in the sociopolitical or pseudoscientific arenas— and there just about anyone can roll their own nowadays. Accordingly, predictive modeling is basic component in the ideologue’s toolkit. In ‘best’ practices, such modeling relies heavily on modeler’s rectal database. Hence a given model can readily yield a broad range of conclusions with facile model/data tweaking - in order to maintain consistency with shifting elitist intent. Sort of outcome-driven prediction, as typically camouflaged with ritualistic emotionalism and commotionalism.
Happily, it is universally the case that you can fool many of the people virtually all the time, not that most people are actually interested. especially in a systematically dumbed-down populace. Lest this account seem overly harsh, it may be stated in defense of the modelers that they often seem to mean well, and many even take their craft seriously. Moreover, the preposterous scatter of results on any given topic can usually be rectified by after-the-fact tweaking. And each modeler can then pompously if selectively assert: “see I told you so.”
So do not be distracted by any reliance on broad perspective, in-depth domain knowledge, experiential grounding, or expert intuition - these notions are not ‘scientific’. So one can properly ignore the likes of Michael Polanyi and Russell Ackoff.
Of course, rigorous modeling has long been routinely practiced in aerospace engineering for example. Rather than mere prediction, the purpose of such modeling is the consequent realization of a desired artifactual product. In effect, engineers then bet their reputations on implementing serviceable products, proceeding from analytical models to deployed implementations. That leaves no wiggle room to rationalize deficient modeling. Any associated shortfall in deployed products is all too evident to virtually anyone interested, perhaps disastrously so.
Post a Comment