Building a wiki for nyc psychoanalytic institutes

A note on my quirky "institute wiki" is the title — and yes, I'm aware that's a strange thing to lead with.

I've written here a few times about my ongoing project: an online directory of New York City institutes for analytic training. The directory has been up, online, and accessible for a couple of months now. It gets a steady trickle of visitors — I know this courtesy of Google Analytics. But except for one institute, a staff member of which submitted a tranche of information about their program, there haven't been any reader contributions. No data added by anyone.

I've been coming to terms with some competing facts about my fantasy here. On the one hand, it's a genuinely nice idea: something like Wikipedia, but for psychoanalytic institutes. On the other hand, for it to actually work like Wikipedia, it would need what Wikipedia has — an army of users committed to its upkeep, maintenance, and quality assurance. In the absence of that, it is instead something more like an Old West ghost town.

I heard a podcast recently about Grokipedia — an interview with Renee DiResta about the initial project, its current liminal suspended state, and the implications of all of that. I'm no fan of Elon Musk. No fan of X, or Twitter, or whatever the fuck it's called, or any of his businesses. But I also can't help thinking that he occasionally has genuinely interesting, perhaps transformative ideas. As I heard DiResta describe Grokipedia, I thought to myself: this is one of them. Not in a this is going to change the world and save it kind of way — but in a wow, this raises all sorts of interesting questions and opens up all kinds of interesting opportunities kind of way.

The idea that stuck with me: instead of setting up a wiki as essentially a notebook with titled pages and all the data to come later, you use artificial intelligence and the internet to populate those pages initially — and then make the wiki function a correction function, rather than an initial data-gathering function.

When I first conceived of this project, I took a crack at populating the pages using AI, and it was laughable. Those were the days of real hallucination by LLMs and very limited tool use. A lot has changed since then. I've been able — I think — to do a much better job on this second pass of populating my directory with accurate, meaningful information. Still some distance from the functionality I crave, from the ease of use that Wikipedia boasts. But I'm not Jimmy Wales, and I don't have an army of Wikipedians who can translate clever ideas into authoritative listings.

So I'm left with a clunky, idiosyncratic, nonetheless useful, preliminary, gestural directory.

Take a look. Let me know your thoughts. I'm receptive to whatever they may be — whether with regard to design, aesthetics, functionality, or content.