Alexa Can’t See Air Quotes

The following is a true story. Including the part where Alexa rolls her eyes at me.

ME: Hey Alexa, play the Yes Album by Yes.

ALEXA: Playing Yes by Yes.


ME: Wait, what? No. Alexa, play “The Yes Album” by Yes.

ALEXA: Playing Yes by Yes.


ME: Dammit, I know there’s a… Oh, I see what’s going on.

ME: Alexa, play “The Yes Album” album by Yes.

ALEXA: Playing the Yes Album by Yes.


ME: Yes. That’s the one. *smug chuckle*

ALEXA: *rolls eyes*

So THAT’S why they call it a Nor’easter

I live in Massachusetts, and big storms here are often referred to as Nor’easters. Why? Because they are associated with northeasterly winds that blow in from the sea ahead of the storm. These winds are so severe that they can blow the letters TH right out of the word NORTHEASTER, leaving behind nothing but a limp apostrophe dangling from the ceiling.

I’ve recently fallen in love with, a weather site that vividly animates winds. It’s particularly gripping during hurricane season, and this hurricane season has made for some eye-popping imagery. Hurricane Jose (officially it’s just a tropical storm now, but I’m not saying that to Jose) is currently rumbling off our shores, and the picture from Windy really illustrates the Nor’easter phenomenon. Look.


The air is being sucked down the low-pressure drain of Jose’s eye, dragging over Massachusetts’ soggy sleeves along the way. A lovely painting of a terrifying creature.

Peak Twilight

Happy Crepusculus! Tonight is the earliest sunset of the year: 4:12:02 PM. At least it is for me and everybody else at my latitude. This image, taken last week, shows the exact moment of every sunset for the week preceding and following today’s early sunset.


Almost every sunset falls between 4:12 and 4:13. It’s like the sun is standing still! We should give this season a special name to honor this remarkable observation. We’ll call it Sun-Still. No, how about Sun-No-Go? No. How about something fancy and Latin sounding, something derived from sun (sol) and standing still (sistere): solsistere. Sol-sister? Okay fine, let’s just shorten that to solstice. I’m sure everyone will figure out what it means.

If you already know about the solstice but are surprised that it’s happening as early as December 8th, I should point out that this is merely the earliest sunset. The latest sunrise is in January, leaving the shortest day on December 21st where it belongs. If it seems surprising that the earliest sunset and the latest sunrise don’t coincide, you can blame the earth’s slightly elliptical orbit around the sun.

In the meantime, I’m more than happy to celebrate the slow retreat of sunset. Today may not be the actual solstice, but it’s worth observing for its own merits, so I’ve given it the name Crepusculus (more Latin: twilight = crepusculum).

Happy Crepusculus!

The Objective Function of the Good Life

I read an interview with baseball stats guru Bill James in which he said something like this: we know a lot about how to optimize the play for a single team. But we don’t know how to optimize a league. For one team, the goal is simple: win the championship. Anyone good with stats, optimization, and machine learning, if given enough data, can help you solve that problem. But what about an entire sport? Suppose you want to optimize Major League Baseball? What do you optimize? Do you want every single game to be a 50/50 toss-up? Probably not. Do you want one or two teams to dominate season after season? Probably not. Should you try to maximize revenue? Happy owners? Happy fans? Happy wealthy fans? Happy advertisers? It’s easy to see how any of these might have nasty consequences if sufficiently amplified.

In general it’s easier to describe specific undesirable outcomes than universal desirable ones.

In an age of machine intelligence, this becomes increasingly important. Machines and data can help you achieve marvelous things, but only if you have a clearly defined goal, a test to tell you if any given outcome is better than another. This puzzle is the idea behind Nick Bostrom’s Paperclip Syndrome. If you give a sufficiently powerful artificial intelligence the goal of making paperclips, it will chew through the galaxy grinding matter into paperclips, humanity be damned. Bostrom’s scenario sounds silly, but the idea behind it is serious. If you have the power to optimize the human condition, what are you optimizing for? Okay, so we’re not going to make paperclips. But we’re going to make something. What?


I recently listened to a Long Now talk by Brian Christian. The topic was Algorithms to Live By. It’s really good. It does a good job addressing the increasingly fraught intersection (or collision) of computer science and the real world. One gets the sense that computer science isn’t ready for it and neither is the real world. Christian takes on several topics, but the most profound one was related to Bill James’s question about baseball: At the highest levels in life, what is the objective function of the Good? It’s clear the answer isn’t to maximize quarterly profits for big corporations. But that’s the world we’re busy building, because we know how.

We’re amazingly good at answering questions. But we’re not so good at coming up with good questions.