People love to talk about how artificial intelligence and machine learning is becoming more democratized. But it really is becoming a lot easier to work on these sorts of projects. I recently came across on Twitter that Robin Camille Davis had poured the Webster's Dictionary from 1913 into a neural network and generated new random words and dictionary entries. For example, Davis created these two oddities: "bortifying" and "tabernished." Delightful!
But then I looked at her project's guts, where she had modified a tutorial ML project from TensorFlow, changing the input data. And I discovered that it was really easy to do! No really, go and play with it, it's really straightforward.
So I decided to do the same thing, but instead with the Book of Psalms as the training data (for those playing at home, it was the JPS translation from 1917). And I got a whole bunch of weird stuff, a lot of which was very similar to actual verses. And some, not so much. A sampling of verses:
Who is the man who traineth my hands for war, So that mine arms do bend a bow of brass.
Therefore hath the LORD compassion upon them that fear Him, In those that wait for His mercy.
Glorify the LORD, O Jerusalem, And they that go down all our days.
I consider my soul from the desire of the hungry.
Behold, how great spoil.
For He spoke waste to sweeace before them.
Let subdunt: 'Return, ye contempt; .
So, what's the takeaway? Not the output, though that's certainly fun. Rather, it's that this kind of work is now so incredibly easy. So go forth and generate computational ML weirdness. (And if you need inspiration, check out Janelle Shane's AI Weirdness).
In our increasingly specialized world, we need more interdisciplinary and generalist-minded thinking. Which is something that many of you know I am obsessed with and have been hammering on for a long time. But part of this involves getting people to recognize the extent to which we are all embedded within complex systems: complex social systems, complex technological systems, complex ones of information flow, and much more. This attempt to convey complex systems thinking (along with a healthy dose of humility in how we think about these systems) is basically the message of my two books.
Well, as part of this I want to highlight several ideas related to systems thinking that I have been reading about recently. In other words, a sort of Systems Thinking Review (maybe this will become a regular thing?).
So here are a few articles that I'd like to mention:
The Infrastructure Tourist: We are embedded in massively complex infrastructural systems. Do we often think about them? Not really, unless something goes wrong and suddenly affects us. But here's one person who is trying to explore them in all of their wonderful physicality.
Let's Play War: Come for a discussion of games, simulation, and the complexities of reality, but stay for the wild notes in the margins of this article.
Roads and Bridges: The Unseen Labor Behind Our Digital Infrastructure: A fascinating report on the open-source ecosystem. Among other things, it highlights the fact that far fewer people are working on these projects than you might expect (and that open source is often poorly supported by its users), and yet they are the foundations for a vast technological infrastructure that touches basically everyone.
Read another article that demonstrates the complex systems around us? Please let me know.
One of my goals in life is to highlight the weird and the wonderful in the world around us (this is basically the only reason I'm on Twitter). And I love it when people read articles and books this way as well. That was why I was incredibly excited to read two recent things:
First, this short Twitter thread of weird facts about famous mathematicians, from Steve Strogatz’s upcoming book about calculus.
And, the second: Kathryn Schulz has a list of the best facts she learned from books in 2018. A sample: "The locust swarm of 1875, the largest in recorded history, affected a quarter of the contiguous United States."
Wow. Oh, and want to get a quick shot of the strange and delightful? Simply click on Wikipedia's Random Article.
Also, I recently appeared on the podcast So to Speak, where I discussed The Half-Life of Facts, changing knowledge, science as a rigorous means of querying the world, the nature of scientific publication, complexity science, and a whole lot more. It's a fun ride.
And some news, appearing here in this newsletter first, Dear Reader: I’m working on a novel! My next book is going to be a work of hyper-nerdy literary fiction (for lack of a better genre descriptor). I don’t want to spoil anything quite yet, but it ties together some fun themes: old technology, fictional maps, immortality, mathematics, ancient civilizations, satellites and rocket launches, philosophy, and humanity’s search for meaning. So, you know, the usual.
The novelist Richard Powers once described novels as "supreme connection machine[s]," and I'm hoping to use my writing as a way to explore not only the interesting and wonderful in the world, but the fundamental ways in which so many different ideas are all connected.
And, related to the novel, I will try to send out this newsletter more often (maybe once every month or so), and include some of the things I've discovered from my research for the new book.
Until next time.