In my mind, knowledge has a color. It has a shine. It has a weight.
It looks like this:
These were the encyclopedias that lined one of my bookshelves in younger years. (I don’t think we bothered with the dictionary and yearbooks.) I remember them less for the facts they contained than for the confidence they inspired. They had what I needed to know.
In practical terms, I consulted them less and less as I passed out of elementary school and needed more contemporary sources to do reports. The 1986 article on the Soviet Union was pretty dated by 1992. Still, I remember how the volumes felt in my room, like a stronghold, like a literal knowledge base. I’d heft five of them over to my bed now and then, just to show that I could.
Time has not been kind to the heaviness of World Book, or the Encyclopedia Britannica, or the Oxford English Dictionary. The World Book continues as a prestige item, and the other two have succeeded online at different levels, but the idea of an essential print encyclopedia is in the past. Once, conservative scholars treated the digital world as if it didn’t exist. Nowadays, it’s the non-digital world that might as well not exist for many who seek information. I'm not here to argue that the age of Wikipedia is better or worse, just that it's here. In the last twenty years, the encyclopedia has changed. In the next twenty, it will surely change again.
The shift toward AI is giving us a pretty good idea of what that change might look like. Less than a year after Dall-E and ChatGPT first hit the internet, both Google and Bing are using generative AI to answer almost any question about the world that you’d care to ask. Pointing out the flaws in generative AI’s offerings won’t stop people using it, any more than pointing out design flaws stopped them from using Wikipedia and Google. The task therefore becomes making those offerings better.
The profit motive can be helpful here, because a cachet as a resource of reliable information is highly valuable. Search engines and other information sources generally live and die by how much trust they engender in their audiences. And such trust isn’t as easy to come by as it was in the younger days of cable news and websites. That’s good; it needs to be worked for.
That said, this isn’t a mere “horse race.” The stakes are too high for that. From pampered billionaires to authoritarian politicians, there are many who seek to erase the facts when those facts threaten their power. I’ve been reading The Infernal Library, an exploration of dictator literature. From Stalin and Hitler to Hussein and Khomeni, dictators have not only censored literature that threatened them, they have produced works of their own to instruct their people how to think—guides to history, the arts, the sciences, and anything else one could ask about.
Once, these sorts of “ideological encyclopedias” were mostly an amusing curiosity in free nations. But today, conservativism leans more and more on “its own” media, “its own” reality, and tries to enforce party loyalty by shouting that any non-conservative sources are broken and corrupt. The tactics of dictator literature are very much in play at URLs like fox.com and conservapedia.com.
Generative AI represents a rare opportunity to reassert facts in the media landscape, but it can’t be left to fend for itself. The times call for a new generation of editors, as devoted to their work as those waging edit wars on Wikipedia.
They must be passionate, but that is not enough. They must also be paid.
Can you elaborate at some point about time not being kind to Britannica, OED, and World Book? OED still seems robust, but online only. (I've subscribed at $100/year for OED for a while and use it pretty often, and it still is revised.) Britannica still exists, but online. I saw a note in Wikipedia saying that World Book still is ordered but has mostly institutional customers.