The thing about predicting the future is that if you wait too long, then you have another set of training data for your model.

I was going to get to work predicting the future this week. But then some practical operational issues came up. The practical operational issues always come up because the people who designed the database to do some particular thing were unaware (or willfully ignoring) some future in which the database would be asked to do some different thing. And so in order to create some sort of entirely reasonable document I need to (metaphorically) integrate over all of the changes that are stored in the database in order to figure out where things are now.

We all know that we can’t be the first people to try to predict the future. All sorts of people who work in the College of Business Administration must have written scholarly papers about ways to predict the future. These papers are probably filled with equations and algorithms; they probably warn (in technical language) about what sorts of futures you can predict with any reasonable certainty and what sorts of error bounds you can put on your future. Someone has probably gotten a graduate degree in future prediction by writing some code that predicts the future in newer, faster, easier, or more accurate ways. With access to a university library (which I can get—I know a guy), you can have the future for free.

But most of us are pure mathematicians, so all we can do is prove that the future must exist. In the land of the blind, the one-eyed man is king, so I feel like I had a decent shot of figuring out how to calculate the future. Because I have this terrible habit of attending colleges (four, if you must know), I have amassed credits that roughly approximate about one semester’s worth of graduate study in business analytics (mostly statistics, some management science).

At first this seemed like a hopeless task. I was trying very hard to google the right keywords and not rely on emailing statistics professors that I know asking them for better keywords. There are a lot of people who are trying to sell their proprietary methods for predicting the future, so there are a lot of SEO-optimized pages that come up in my searches. And they all read pretty much like business plans written by South Park’s underpants gnomes.

  1. Collect data
  2. ????????
  3. Future!

But then I stumbled upon the names and articles and textbooks and software and websites and everything of some of the world’s experts in predicting the future! (They call it “forecasting.”) Once I am done with short-term operational stuff, I will feed my data into their future-predicting algorithms!

It is too bad that my migraine auras are mostly disruptive and do not function like prophetic visions. It hardly seems fair that I do not get any benefit out of seeing things that other people do not see. Yesterday I worked on neither future prediction nor on operational details because I was at home, freaked out about my latest encounter with the flashing rainbow crystal.

This time the flashing rainbow crystal was close enough to the center of my vision that I was able to see that it was also spinning. It looked like a glowing octahedral twisty puzzle, but where the pieces weren’t all identical, so if you twisted it, it would shapeshift or something. And it was positively glowing, as if the pieces were cladded with tiny little LED screens. Possibly related, I am tapering off one medication in anticipation of starting another (IF THE DRUG COMPANY EVER CALLS ME).

The aura seems to be gone for now, but I would really prefer not to have uncommon neurological symptoms. I mean, the headaches are annoying, but they are remarkably consistent in how they develop, how long they last, and how very well they go away when I take medication. My usual auras are so fleeting (and so consistent) as to be really quite boring. But the flashing rainbow crystal is not something that I see often. I’m told that most people with auras will see the aura for an amount of time best measured in minutes. Mine will come and go, several times a day, over a period of weeks. And I do not have enough data to predict when the flashing rainbow crystal will return.