Don't Believe the Hype

At the same time, I imagine he could make a lot of accurate guesses on where the industry is moving. People are not expecting him to be exact so, to me, his response reads like someone who just doesn't want to be wrong. Even if Brad doesn't invest in the short term, he is perfectly capable of giving generalized short term predictions. There are superior less evasive answers from other investors for anyone who is looking.

If you record them and follow up on them to see how well you did, predictions are also an excellent way to calibrate your mental model of the world. If you make predictions that you feel pretty certain about based on how you think the world works, and those predictions turn out to be false, that gives you an opportunity to adjust your point of view and figure out why you might have gotten it wrong.

At a more general level, making and checking predictions is really the only way to get a good sense of whether your certainty about your beliefs in different areas should be upgraded or downgraded.

The problem is that if you don't predict anything, then your brain will automatically fit whatever happens to your existing model. We're frighteningly good at witnessing an event and then recalling the beliefs we have that would predict that event, and we're absolutely terrible at witnessing an event and recalling the beliefs we have that would have ruled it out. This gives us the sense that we have a firm grasp of what's going on, when in reality we're holding onto beliefs that should have been discarded long ago.

Making a prediction is a way of trapping yourself so that your brain can't sweep your faulty beliefs under the rug. Once a group of beliefs have been organized into a hypothesis, and that hypothesis has been falsified, it's harder to ignore the reality that something in your model is broken. And since those beliefs have been organized into a hypothesis, you then have the ability to analyze and narrow down what you need to rethink.

So we should all be trying to make predictions. But don't just start imagining stuff that sounds cool. Actually look at what you believe and figure out what you should expect from those beliefs and why.

And if you can't come up with any expectations, you need to scrap your model and start over. Because predicting is what beliefs are for.