Welcome to Colligo

Toward a humanistic theory in an age of data

Colligo /ˈkol.li.ɡo/ Latin “to bring together, gather, collect”

collecting paintings together

Greetings and welcome to Colligo. Colligo is Latin for gathering things together, which is admittedly a broad category of action. Gathering what together? Why? But I think this big tent action verb, to collect or gather, perfectly fits my project on these pages. In my 2021 book, The Myth of Artificial Intelligence, I argue that computer pioneer Alan Turing made an “intelligence error” in adopting or at least tentatively accepting a problem solving view of intelligence. Treating intelligence as problem solving is what gives us “Narrow AI,” artificial intelligence applications that can solve a particular problem, like playing Chess or Go or various Atari games, but can’t generalize to grasp commonsense thinking necessary for true intelligence, like ours. A better view of intelligence emerges from studying the process Turing and his colleagues underwent at Bletchley Park, during those anxious months in the Second World War, when German U-Boats were devastating shipping lanes on the English Channel and preventing needed supplies from reaching a beleaguered Britain.

The mathematicians and champion chess players at Bletchley were not really engaged in problem solving—or I should say not only engaged in problem solving. They were collecting clues to bring to bear on a larger mystery, that of cracking the “uncrackable” Enigma Machine codes, the cipher device used by the Nazis to encrypt war communications. In a brute force problem solving world, the Enigma was uncrackable. But Turing and company used a sleuthing procedure which in mathematical circles might be called “weight of evidence.” Human intelligence sources from captured or destroyed U-Boats also played a role. Fragments of partially deciphered text were discovered by Polish forces, for instance. Mere observations became clues in context. The Bletchley team thus gathered together the rag tag bits and pieces of evidence they could bring to bear on the core task of deciphering. This more open-ended, non-computational procedure, this sleuthing by investigating and collecting clues, is a key feature of general intelligence. That’s why I named my project here Colligo. I’ll be gathering ideas together to get a Gestalt. I’ll be sleuthing toward what I hope’s a better understanding of the world we live in. Speaking of…

What is the World We Live In?

My starting point is that the world is screwed up. Really, most of the 21st century has been screwed up. It’s not like the Cold War (but it is), it’s not like the previous century (but it is). I’ll unpack all this as we go along (as we collect our weight of evidence…). For now, we can say the world we live in is dominated by data, algorithms for analyzing data, and even an understanding of ourselves as data. The popular historian Yuval Harari, for instance, declared that our modern worldview is “Dataism.” Dataism is the view that everything is data, including ourselves. As a factual statement this is either vacuously true or almost certainly false. As a worldview it’s positively anemic. A key problem with the world today is data. We modern bipeds are awash in data, and like the proverbial fellow with a hammer who sees everything as a nail, we see intelligence, progress, innovation, and even happiness as a question of getting the right data and manipulating it in the right way. This is a bad view. It’s no wonder we feel isolated, set adrift, anxious, depressed, and well, crappy. Dataism is at odds with human flourishing. Data is just input into something else. If we are just data, than we are inputs into something else. It’s difficult to find a Renaissance moment in this ruinous reductionism.

Enter Colligo. Colligo is about collecting clues. It’s not just about gathering things but about gathering them together to paint a picture of what’s wrong and how to fix it. Human smarts and not just data crunching is what matters. I want to show the problems with our data-driven world and show or assemble a richer humanistic picture. I think we have to do it this way, because the task is too complex to martial academic arguments and try to deduce everything from first principles. Escaping the data-driven cave into the light of humanism is not a “problem solving” task. It’s not debate class. It’s a Bletchley task. Weight of evidence from myriad sources. Observations. Think Montaigne rather than early Wittgenstein. That’s Colligo.

Who I Am

My name is Erik J. Larson. I’m probably best known for my book The Myth of Artificial Intelligence: Why Computers Can’t Think the Way We Do, published in 2021 by Harvard University Press. I also write shorter pieces for The Atlantic, Wired, The Los Angeles Review of Books and other pubs. Most of my focus is on Artificial Intelligence (AI) and the social and cultural impact of AI, but I’ve also written about plane crashes, burning cathedrals, and literary theory. I was awarded a two-year grant by the Thiel Foundation to work on a second book, titled The Return: Why the 21st Century Looks Like the Past (so far), Not the Future We Wanted. I’m also a Fellow with The Institute for Advanced Studies in Culture at the University of Virginia, and in November last year I was a visiting researcher at The Santa Fe Institute.

Who I’m Not

I’m not an academic. I have a Ph.D. from UT Austin, but other than some brief forays into online teaching, I’ve never been interested in employment as a professor. For the last twenty years I’ve been working as a research scientist in industry, where I specialize in text processing and the subfield of AI known as Natural Language Processing (NLP). I’ve worked in Fortune 500 companies and garage startups. I’m also the founder and co-founder of two DARPA-funded companies, the first in the 2000s and the second in the 2010s.

Why Substack? Why Now?

I admit I was initially reluctant to start writing here, as I have this old-fashioned notion that frequent colloquial writing might dull the sheer genius that surfaces when sweating over a piece for The Atlantic. This may (or may not) have some truth in it, but I’ve come to believe it’s more important to get thoughts out than swing for home runs a handful of times a year at most. Most of my year I spend reading and not writing. I write for publication when I’m inspired, which may be weeks or months between projects. Book writing is playing the long game indeed.

I’ve spent the last eighteen months, give or take, researching and writing a proposal for The Return. If and when publishers contract the writing, that’ll be another eighteen months give or take. That’s three years, and honestly I think my timeline is a bit ambitious. As a general rule, writing a book (or I guess any big creative project) takes longer than you expect. It’s harder than you wanted. And it pays the bills a lot less than you’d hoped.

I’ve also heard from colleagues about their frustrations trying to craft prose that entices people in suits somewhere on the East Coast (editors and publishers). I did not notice this with Harvard U Press—they were in general easy and reasonable to work with—but I have noticed the creeping pressure on longform writers to write web-digestible chunks of prose on topics that can’t be too philosophical or too deep. A writer can end up chasing the news cycle and playing a popularity game, which can make you feel a bit like Bill Murray’s character in Lost in Translation, shilling Japanese whiskey (make it Suntory time!). All of this to say: it’s time for a venue like Substack. The pros far out way the cons. AND… I’ve got lots to say.

What To Expect

I aim to post once a week on Colligo. I may not meet this goal perfectly, but I won’t be far off this mark either. Once a week works for me. I hope it works for you.

Also, on content, for techies: I’ll say plenty about the nuts and bolts of particular technologies. I intend to do more than one post just on ChatGPT and the large language model craze. For non-techies (or techies who are not just techies), I’m working out a larger theory of culture and history, so there will be lots of non-technical content to chew on. I’m trying to put together a picture of the world today, you might say with a technology centered focus but reaching out further.

If you find anything here useful for your own journey, your own questions about the world today, what’s wrong with it and how it might be improved, please consider subscribing. You can subscribe for free, or you can do a paid subscription (I am reader supported, so paid subscriptions are much appreciated). Paid subscriptions will be $7 a month, or $70 a year. If you do a paid subscrition, you’ll get all the content with no restrictions whatsoever. Free subscriptions will get more limited content and won’t be eligible for perks, like the ability to comment on certain posts, private chats, and others TBD. About 50% of the posts will be locked, so if you like Colligo, consider the paid option so you don’t miss anything.

Want to be a writer? You can’t. You’re not good enough. Kidding. Hit the button below to start your own Substack and connect your newsletter to Colligo.

Love ya, man.

Colligo is not just a newsletter. What I’m starting here is really a community of like-minded people who share similar values and interests. I’m grateful to be able to start Colligo community, to gather together all of you who share a passion for truth, for good writing, and for human flourishing. Colligo means “gather together,” and along with the human mind’s gathering together of “the rag tag bits and pieces of evidence” to peer more deeply into the mystery of the present, like-minded searchers gather together in communities to share ideas, knowledge, hope and inspiration. It’s not a community of machines. It’s a community of people. Colligo—and you all—are part of a new humanism. Let’s get on with it.

---