Models are indispensable. Reality is insanely complex. Mapping every possible interaction would be computationally impossible and utterly useless for understanding the world. Instead, we flatten things down to key causal variables and use them to help us make predictions and decisions.
But models come in different shapes and sizes. Which model is useful depends on what one is trying to do.
Take, for example, the following two maps. Both of these maps are of the same area: Nicholls State University in Thibodaux, LA, my employer. The first of these is the campus map we have on our website. The second is a topographical map of the same area. Let me ask you, dear reader: if you wanted to get to Powell Hall, which of these two maps would you prefer? Obviously, the campus map would be most helpful. The topographical map, while it contains useful information, would be useless for navigating the campus. Likewise, the campus map would be useless if one wanted to, say, hike in Thibodaux.
Using an incorrect map (an incorrect model) can lead to disaster.
One such example comes from the US Invasion of Grenada in 1983. The US invasion was planned, not with military maps, but with tourist maps with military grids superimposed, bought earlier that day from a shop in Fayetteville, NC. The maps were not helpful: 4 SEALs died due to water hazards not on the maps, coordination between soldiers and their air cover could not occur as they were using different maps, and so on. It almost became a disaster on the scale of the Iran Hostage Crisis. All because the military used inappropriate models.
The same is true of economic models. As we have seen over the course of this second Trump Administration, their models of trade have been disastrous. Absolutely nothing has gone right: interest rates are rising, countries are retaliating, the dollar is weakening, prices are set to rise once these stockpiles run out, American firms are laying off or slowing hiring, major factory plans have been cancelled, the stock market tanked, and China is stepping into American markets. It’s so bad that the president and his team now admit that prosperity is no longer a goal (but it’s ok, because prosperity is bad). Indeed, these tariffs were supposed to bring in almost $60 billion in revenue per month. April’s tariff revenue (which is high because of stockpiling) was just $17 billion.
When the Trump Administration gives models, they give bad models. It’s like invading Grenada with a tourist map or using a topographical map to navigate a college campus. Their models have unrealistic, often contradictory, and usually unsupported assumptions. Their models give unrealistic, often contradictory expectations. And, consequently, the American people pay the price.
What is important to note is that most models are at least mathematically coherent. But that mathematical cohesion does not imply the model is useful for the purpose. A model’s use is not determined by the mathematical sophistication, but rather its application and ability to provide useful insight. Consequently, a model that might be more “realistic” by incorporating more elements of reality or mathematical sophistication may be all but useless to a model less “realistic” for being more simple. If the more complex model provides limited (or incorrect) insights, then it is not useful, no matter how mathematically coherent it is. Science is not merely about manipulating models. Much of it is having the knowledge to choose the right model.
