Models are not evidence

Seems sensible to me.

Models are not scientific evidence. Evidence is proof of something that has already happened. Model predictions have not happened.

Repeat it with me:

Models are not evidence.

Models are not evidence.

Models are not evidence.

This is true about lots of things, specially in economics. First the conception, then the model, then the real world, then the application, then the failure and then the forgetting about what happened so that the same mistakes can be made all over again.

Models are not evidence was a comment on this post at Instapundit. Along with these.

If there is a silver lining to this event, perhaps it will be that more people will come to understand why the models for climate change should not be taken as gospel. Models are based on assumptions, and the more complex the model, the greater the number of assumptions. When our so-called experts were so wrong with the models for a virus that acts much like many other similar viruses, how the heck can anyone expect experts to correctly model the interaction of everything in the earth’s atmosphere? It makes no sense.

Everyone’s got a cold, let’s shut the economy down. The hyperbole is strongly on the side of the jackboots.

If you want to live in fear, stay in your stinking lair and obey your government masters, you quivering slave.

It’s going to be fun watching the Blue City State folk impoverish themselves in an attempt to damage Trump’s re-election. While the rest of us are at work, restaurants, gyms, shopping, etc. At some point, they will have to throw in the towel and grudgingly rejoin the rest of the country, with the (delicious, to me) knowledge that this will enable Trump 2020.

I always ask them “why not lower the speed limit to 25mph too? If it saves one life…”

Don’t mean to be argumentative, but GIGO is only part of the problem. In all modeling software the data inputs are usually “tuned” by weighting factors. These factors are sometimes based on experiential data matrices or historical observations and collective norming, but at times they are little more than “knowledgeable estimates.” With experience – a whole lot of detailed experience – a model designer can fine tune the formulation and weighting factors to increase accuracy but this process is often somewhat less than rigorous “scientific method.” Consequently, in some instances, “modeling” is little more than a fancy word for guessing.

When they drew the line in the sand back at the start of March, Sarah was the only mod here who backed a free people and supported individual liberty. The rest went apocalyptic, demanding we sacrifice our rights for the common good (Yes, Glenn Reynolds wrote a USA Today article using that phrase), and screaming that the authorities weren’t locking us down fast enough. When push comes to shove, I know Sarah will support me and my freedoms, and that the rest of the mods here will happily sacrifice me. Don’t ignore that very important data point.

2 thoughts on “Models are not evidence

  1. Pingback: Models are not evidence - The Rabbit Hole

  2. Good article. Having created 100’s of models professionally, I will add the following. Models are mathematical representations of reality. F=ma may be the simplest and best known example. Reality is complicated and as a result almost all models are a simplification of reality – otherwise the models quickly become intractable. The models used by epidemiologists are very simple when compared to reality of the spread of a virus. The same is true for modelling hurricanes, the stock market, and climate.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.