Climate Models and COVID-19 Models

From American Thinker.com (April 15):

Computer models are seductive even though they are very often completely wrong. The more complicated they are the greater chance that they are wrong. Like children, they copy their parents -- the model architects. Confirmation bias involves cherry picking facts to obtain a result consistent with preexisting beliefs. A complicated computer model with many degrees of freedom is a perfect environment for confirmation bias to have its way. The investigator usually will believe, or at least claim, that his model is objectively setup without bias entering into the effort.

Not all computer models are wrong. Sometimes they produce good predictions. But all too often they fail and the failures are not acknowledged because the modelers are emotionally or ideologically attached to their creation.

The claim of objectivity by academic modelers contrasts with the standard leftist or academic belief that practically everyone is a racist, driven by unconscious motives. Google “implicit bias” if you want to know more. When professors are pushing racial justice theories everybody is driven by unconscious forces. But when constructing computer models all is well.

Complicated models are always full of escape hatches that can explain away any failure. Climate models still enjoy support in spite of 30 years of failure. The failures are alleged to be due to things like chaotic variation or data that has to be adjusted because it does not agree with the model. COVID-19 models are new but have had notable failures. For example, the IMHE model predicted up to two million deaths but has been repeatedly adjusted and now is down to 60,000 deaths. Usually it is claimed that the model is not wrong, but deaths are lower because the American people have been good boys and girls.

In both climate models and disease models, the associated scientific establishment has a vested interest in the validity of the model. If the climate models are a waste of time, then climate science and its practitioners have been wasting everyone’s time. The scientific establishment behind COVID-19 modeling is wedded to a particular method of dealing with an epidemic. Their approach is to constrain the spread of the disease with social control until a vaccine is available. [read more]

Other articles dealing with the subject:

• The Challenges of Forecasting the Spread and Mortality of COVID-19
• Flawed Models Show Why COVID-19 Policies Must Consider Total Mortality
• COVID-19: Can We Estimate Infection Speed and Fatalities?
• What Happens When the Coronavirus Models are Wrong?
• Computer Simulations in Science

It’s interesting that climate models and the COVID-19 spread models are a mystery. Why do we have to social distance six feet? Why that number? Why not seven feet? Five feet? It would be helpful if a reporter asked questions about the model during Trump’s press conference. Maybe the press isn’t smart enough to ask the question or maybe they didn’t want to put Dr. Fauci on the spot.

All computer models are just theories about a specific phenomena. Therefore data inputted into the model has to be valid and the model has to be calibrated correctly. The model should be ran more than one time to get a consistent result. Also, the models should be ran with different scenarios if possible. For instance, in this case of the coronavirus, the model should contain agents that social distance in one scenario and agents that don’t social distance in another scenario. Ones that wear masks and ones that don’t, etc. I am using the word “agent” because I am guessing this is an agent-based model the epidemiologists are using. If the agents have variability the better off the model is. I understand their model only used a model that used self-quaratine-- that was part of their assumptions. That’s not good. Meteorologists use different computer models to better grasp what’s going on with the weather. I would hope epidemiologists would do the same. Disease spread is complex too. I can imagine it is hard to predict how a virus would mutate for example. Would it be more or less deadly? More or less spreadable? But still to try to have any variability is better than none. Anyway, just understand computer models are just a tool—not the final answer.

Also, computer models, especially ones where the agents are represented as people, never take into account side-effects or what economist Thomas Sowell calls “stage two thinking.” The models are usually never holistic. They never take into account the psychological, sociological, economical, and yes the spiritual side-effects of their models.