We don’t go to the Sunday morning Larchmont Farmer’s Market as often as we used to, since Flywheel went bankrupt and we got our Peloton. When we do go, I like to stop by the produce stand and every once in a while I’ll pick up a bag of blood oranges and spend the morning juicing them. I love their intensity—for some reason the deep red color makes me think that they’d be harsher than your standard Valencia, but really they’re less acidic, sweeter, rounder in flavor.
One week I decided that blood oranges would make a tasty top note for a brew and I got to thinking how it would be structured. It felt a little like a Belgian and a little like a wheat beer. I also polled my Oregon text group, and Jeremy asked for something really hoppy. Collating all of this data, I landed on a white IPA, the bastard child of a Wit and an American IPA.
Copious amounts of blood orange juice were boiled down with sugar into a flavored simple syrup that would thin the body and raise the ABV. The grain bill would be an even split between pilsner and wheat, with a dash of oats to add a little mouth feel. Spiced with cardamom and, for a touch of the exotic, sage. Magnum hops for bittering and a bouquet of citrusy hops for flavor and an aggressive dry hop: Centennial, Sorachi Ace, and Mandarina Bavaria. Classic Witbier yeast.
This is just lovely: the blood orange held its own through the fermentation process and supports a really elegant Belgian. Delicious.
Back in 2008 or so we were doing a lot of financial modeling for Disney English to explain to management why the financials looked the way they did. At its base, we modeled out individual learning centers – each went through a predictable phase of heavy buildout investment, followed by a steady operational cost phase which broke even when the centers hit 70% student capacity. We stacked the models for these individual centers (each of which I called a “brick”) on top of each other in time sequence and added central costs to get to the overall financial model.
It was all a really complicated way to explain to Management that if they wanted us to hit a big revenue number quickly, we had to make a lot of investment fast. It was incredibly flexible—you could just throw in new center openings on the calendar and watch both the revenue and investment spike. But there was a bit of Rube Goldberg crossed with butterfly effect to it.
Andrew Sugerman, our unit head, started to ask what would come out of the sausage grinder. Or what the million monkeys typing away would come up with. One day these two concepts got confused in his head and he asked what was coming out of the Monkey Machine. And thus the concept was forged.
But the Monkey Machine really came into its own in 2023 as Generative AI came into the mainstream with the launch of Dall-E 2 and ChatGPT. The capabilities of these technologies are amusing and spooky. The initial hype—a lot of which persists—is that Gen AI is demonstrating that computers can think and the risk is that they could potentially replace humanity, a la The Matrix.
I’m not convinced by this—yes, what we’re seeing is a complicated leap in the capabilities of machine learning—but fundamentally what we’re seeing is impressively complicated math that’s using statistics to do an excellent job in predicting the most reasonable next word or pixel to add to an aggregate, based on an enormous corpus of information. What this exposes is that to some degree this is what humans do in day to day thinking—here’s an excellent article by Stephen Wolfram that makes this point—but it’s not underpinned by human-like understanding or internal model of the world. It’s literally harnessing a million monkeys’ typescripts—only God knows what’s going to come out.
The risk to humanity is that we put too much trust in these spooky statistics and treat what the computers as legitimate without verification. We see the risk of this in the lawyer who let the computer write his legal briefs for him; in the racial prejudice that algorithms confidently surface out of their embedded data sets; in the work that my employer, Mozilla, is doing to promote trustworthy AI.
To hearken back to my days at Disney, in this moment we’re the Sorcerer’s Apprentice and we’ve got to watch those broomsticks closely.
Social media is another kind of Monkey Machine; we’re all watching each other watching each other watching each other. Kevgusar from Indonesia came back to illustrate the nightmare of recursion in which we all live when we don’t lift ourselves up from the glowing rectangle to enjoy the sunshine.
It was my sophomore year of college and the CD rental agency had become a thing: rent a CD for a couple of days and maybe buy a blank cassette while you’re at it. Ah, those nostalgic early days of music piracy. We were all so innocent then.
Wire had just resurfaced with A Bell Is A Cup Until It Is Struck. The earlier postpunk albums that had made their name were available for rental, and they were on the Harvest label, which was a big selling point for a fan of early Pink Floyd. So I rented 154. I think I never got around to returning it, which blows the CD rental model. But whatever.
Wire’s music is not inviting. It tends to be cold. Off-putting. Like Sprockets without the monkey. This is about as sprightly as they get—and on the grey-toned 154 it shines out like a sunbeam.