Business Crystal ball

The Perils of Predicting COVID-19

Predicting the future is hard. Predicting the future of COVID-19 is a case in point.

“Trying to predict the future is like trying to drive down a country road at night with no lights while looking out the back window. ” – Peter F. Drucker

On March 16, 2020 Neil Ferguson, an esteemed epidemiologist at Imperial College London, confidently predicted that if the U.K. did nothing, more than 500,000 U.K. citizens would die from coronavirus, and that even with some mitigation efforts it “would still result in about 250,000 deaths and completely overwhelm intensive care in the NHS.” In an abrupt about-face on March 26 (10 days later) in parliamentary testimony he revised his estimate to “less than 20,000” and stated that the U.K. should have sufficient intensive care units to handle it. All of his statistics are derived from computer modeling.

Poor Mr. Ferguson, if only he’d read Peter Drucker! The media and politicians pounced on the initial estimates and forged full steam ahead with frightening headlines, multi-trillion dollar spending packages, recriminations and pious pronouncements, only to say “Oops” 10 days later. Except that many never said oops or even acknowledged the revised forecast.

In truth forecasts of what’s ahead with COVID-19 are all over the map. There are two problems:

  1. We don’t have the data we need regarding the status-quo.
  2. Even if we did, we don’t have reliable models to predict the future.

Other than that, we’re in great shape!

True, we need to act decisively in times of crisis. But, as the adage says, many unscrupulous politicians, journalists and hacks in general won’t let this crisis go to waste. And the pandemic is the biggest crisis faced in my nearly 72 years. I fear that it will be used by some to take our money and our freedoms. As professionals who know the hazards of predicting the future we owe to ourselves and our posterity to be watchful. Don’t let fear cause you to suspend skepticism. Question the numbers. What are the assumptions that underlie the predictions? How are the models validated? What data were used? What are the shortcomings? What are the possible biases of the researchers? These are just a few of the crucial questions you need to ask when you read those scary headlines.

Fancy computer models and impressive statistics are no substitute for common sense. In fact, red flags should fly when you see them used. So when a Harvard researcher predicts the death of millions, take it with a grain of salt. Take a deep breath, and ask a lot of questions.

* To date (April 20, 2020) there have been 16,060 coronavirus deaths in the U.K., a tragedy for sure, but no where near the initial predicted number. The 20,000 deaths prediction may be close.


  • Manny Barriger April 20, 2020 at 5:58 pm Reply

    Very well written Tom through this chaos we have to think clearly and be very accurate with what is published. I have received many calls regarding virus and trully state with each dialogue, “I am not trained in medicine and therefore unqualified to answer that question.”

    • Thomas Pyzdek April 23, 2020 at 2:52 pm Reply

      Great comment, Manny. “I am not trained in medicine and therefore unqualified to answer that question.”

      Actually, MDs are also not qualified to answer many of the questions surrounding this pandemic. For example, its distribution and spread. Most of the questions surrounding this novel coronavirus involve the statistical specialty of epidemiology. MDs can credibly discuss the progress of coronavirus infections and its effects on the human body. In this age of specialists what we need are some generalists who can authoritatively comment on forecasts, systems and plans, and on the pronouncements of “experts” who may have biases and agendas to promote.

  • Mike Sawicki April 22, 2020 at 12:53 pm Reply

    Good read Tom, now the “experts” are predicting a second wave coming this fall that will be worse than the current one. They didn’t get this one right and we are suppose to believe their prediction for the coming fall season. Lots of fear mongering in my opinion.

    • Thomas Pyzdek April 22, 2020 at 10:41 pm Reply

      Experts say our actions will just spread the pandemic out so we don’t overwhelm our healthcare system. We will still need to reach herd immunity by paying our dues.

  • T M Kubiak April 22, 2020 at 10:22 pm Reply

    Outstanding article. Well-written and concise and it hits the key points.

    Who was it that said something similar to “All models are bad. Some are useful.”?

    I hope all is well with your family. Be well. Bell safe.

    Best regards,

    T. M. Kubiak

    • Thomas Pyzdek April 23, 2020 at 2:54 pm Reply

      That was George Box. How right he was!

    • Thomas Pyzdek May 22, 2020 at 9:48 am Reply

      Let’s not forget Yogi Berra “”It’s tough to make predictions, especially about the future.”

  • Bobby Wainscott April 25, 2020 at 5:40 am Reply

    Well written and straight forward. Thank you for another one “from the gut” Tom!

  • John L Ware May 15, 2020 at 1:32 pm Reply

    Thanks Tom – this reminds me of our yearly hurricane models…affectionately known as “spaghetti models,” as each predictive model represents some colorful strand of spaghetti usually reaching from the Eastern Caribbean to the Gulf of Mexico or up the Atlantic Coast.

    You may remember Hurricane Irma a years ago – predictive models had it going into the Gulf of Mexico, “possibly” crossing the lower tip of Florida in doing so. As it happened, the hurricane ended up literally splitting the state in two from the Everglades on up to the Georgia border, and beyond. No model predicted that.

    The past doesn’t predict the present and, with storms getting stronger and stronger due to climate warming, that trope will only prove to be more true going forward. I work with businesses every day improving their operations; while the data and information collected is a boon to the DMAIC process, ultimately success or failure hinges on changing business behaviors. Imagine that.

    • Thomas Pyzdek May 15, 2020 at 4:45 pm Reply

      It helps to remember George Box “All models are wrong. Some models are useful.” And the corollary to this, some models are not useful.

Leave a Reply

Your email address will not be published. Required fields are marked *

FREE Sample Lesson