From Nate Silver:
The “prediction paradox”: The more humility we have about our ability to make predictions, the more successful we can be in planning for the future.
I’ve read this quote many times and it’s not really making sense to me. If “humility” is replaced with “confidence” then it seems true, but “confidence” seems nearly the opposite of “humility.” It doesn’t seem like a paradox to me, it just seems wrong.
I think the quote makes sense. Simple models are more robust. Also, humility would mean constantly keeping in mind that your model is incomplete and evaluating how well it performs when perturbed, analogous to shaking a ladder before you climb it in order to test how solidly it is positioned.
I guess it depends on what is meant by humility and successful planning. Consider the headlines “Flash Floods Very Likely Today” and “Some Weather Event Possible In the Near Future.” Which is the more humble prediction? Which allows you to be better prepared?
It’s not humble predictions. It’s being humble about your ability to make predictions. It’s about knowing your models are crap, like all models, and knowing the predictions might be bad, but knowing this you’ve done what you could to improve them.
Jon hit it; it doesn’t mean making such a general and amorphous prediction so as to cover all possibilities and elucidate none. Rather, understand that models have limitations, and within the limitations they have value, but are not infallible. Knowing how to balance the predictive value of the model with the uncertainty that comes along with it will prevent “going off the deep end” following the model.
Nate Silver’s quote reminds me of the suggestion that one should have “strong opinions, weakly held”. That is, act on your beliefs with conviction but be willing to revise them in the face of conflicting evidence.
Your email address will not be published. Required fields are marked *
You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>
<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>
Notify me of followup comments via e-mail
John D. Cook
Subscribe by Email