Alan Zendell, August 6, 2022
When historians write about the last few years, “forecasts and outcomes” might well be the tag line. As marketing and government systems, social networks, and statisticians collected mountains of data, and computers crunched them, we experienced a massive change in the way we process information. You almost have to be a meteorologist to understand a weather forecast, dozens of software packages help investors predict stock prices, and we’re fed so much quantitative information during sports events we barely have time to watch the games.
Recently, while I was watching a baseball game on television, a ball was hit over the center fielder’s head. He took off, leapt through the air, and snagged the ball just as he hit the ground. Within seconds, my television screen informed me that the fielder had had less than an eight percent chance of catching the ball, and the route he reflexively took to intercept its path was ninety-seven percent accurate. No one even questions how they know that stuff. People seem to trust everything they read on a computer or TV screen, unless they have a vested interest in believing an election was rigged.
Mathematical forecasting models have become extremely reliable as the amount of data and computer speeds grew, but they’re far from perfect. Last week, my town’s weather forecast predicted rain every day. We never got a drop. The reasons predictions fail fall into two categories: insufficient or inaccurate data and a faulty model. When a model fails, it’s usually not the math that’s in error. Forecasting technology and data modeling have been around forever, and they’ve been vetted and tested countless times.
What goes wrong is that models become obsolete, It can happen quickly, and even when the best minds in the business sound alarms, many people ignore them because they have too much invested in the status quo. A good forecasting model has a database of many years of historical information from which to identify and model trends. If the data are good and the underlying forces that produce the outcomes we track remain constant, predictions improve over time. But if the fundamentals change, the whole basis of the model can be undermined. That’s happened, notably, with climate change. Global warming changed the way energy flows in our atmosphere, making predicted outcomes (weather forecasts) less accurate because the new conditions weren’t included in historical data.
The same is true for economic data, politics, and social trends. The emergence of a political force like Donald Trump is totally unpredictable. An economic dislocation due to a temporary shutdown of the economy because of world-wide health crisis makes accurate forecasting nearly impossible because, nothing like it exists in our historical database.
All this matters a lot right now. Economists use historical trends to predict future GDP rates, inflation, job growth, and unemployment. Even determining when our economy is in recession is based on arbitrary definitions and past trends. Political forecasters spent 2022 predicting the outcomes of the midterm elections, and it’s important, in our sound-bite culture, that we rememeber it’s our votes that control the outcome, no matter what the polls say.
No model or expert could have anticipated how moving most of our manufacturing overeas would result in broken supply chains when COVID struck. No model could have anticipated the political divisions over a president leading an insurrection that threatened the basis of our democracy. Nor could it have predicted how the actions of a reactionary Supreme Court would affect how Americans cast their votes. The result, for forecasters, is chaos.
Seven weeks ago, the average price of a gallon of regular gasoline peaked at $5.016, triggering panicked forecasts that inflation would cause a catastrophe for Democrats in November, yet today, the average cost of the same gallon of gas is $4.084, down 18.6% and dropping every week. Despite predictions to the contrary, our economy added nearly a million new jobs in June and July, with modest increases in wage growth. And last week was the biggest shock of all, when Kansas, one of the reddest states in the country voted to preserve abortion rights in its state constitution.
We don’t know whether the Trump or Cheney factions of the Republican Party will dominate, or how independent voters will react in November. The Democrats’ success this week in passing the CHIPS and PACT bills, and reaching an agreement that will allow passing the Inflation Reduction Act in the Senate under reconciliation have thrown yet another monkey wrench into forecaster’s predictions. And there is still a huge unknown that could change everything. If former president Trump were to be indicted either by the state of Georgia or the Department of Justice, how would that affect the November vote count?
The truth is, no one knows. Americans need to do their own thinking for a change. Turn off your damn social media, and if you must let your favorite news channel fill your head with propaganda, at least listen to the other side and decide for yourselves what’s best for our country.