Personanet and infonet, and the Semantic Web

I recently came across an article on someone's personal blog that spoke eloquently and plainitively about the reason they write online, and how a world of personal AI agents collecting and synthesizing information on-demand for users conflicts with the very reasons they write online in the first place:

There's a fair bit of talk about “Google Zero” at the moment: the day when website traffic referred from Google finally hits zero. If the AI search result tells you everything you need, why would you ever visit the actual website?

Well, I want you to visit my website. I want you to read an article from a search result, and then discover the other things I’ve written, the other people I link to, and explore the weird [blog UI] themes I’ve got. I want some of you to read my article then ask me to speak at your conferences. … I write the content on this website for people, not robots. I’m sharing my opinions and experiences so that you might identify with them and learn from them. I’m writing about things I care about because I like sharing and I like teaching. I spend hours writing these posts and AI spends seconds summarising them.

I'd much rather people read the whole thing, take it in, digest it and have opinions right back at me. I love it when people connect with what I’m writing (and sometimes they email me to tell me that, which is really delightful).

This sort of sparked a realization for me: there are two fundamentally different ways to use the internet, and although everyone uses it in both ways, we tend to use the internet in one of those two ways very distinctly, at separate times, depending on our goal and context. This means that running into things designed for or assuming the other use is frustrating. I call these two uses infonet and personanet.

Infonet is a source for on-demand information. Its purpose is learning specific things or solving specific problems. It's what we use forums, short technical blog posts, wikis, and now AI for.

Personanet, on the other hand, is a sort of proto-social network: it may convey useful technical information, but that's not the primary purpose; the primary purpose is to build human connections and convey personal expression. (This blog is 100% personanet).

On each of these internets, both infonet and personanet, there are people who write and read; however, outside platforms with explicit reward mechanisms for on-point technical answers, like Quora, StackOverflow, technical forums, and to some degree Reddit, or which were explicitly created to hold technical information as a reference like wikis, most writers on the internet are personanet writers, because that's the most obvious motivation to put that much effort into writing online: recognition and human connection. Humans are social animals, after all. Meanwhile, although all infonet surfers are personanet surfers at other times, current social media has subsumed most of this market, leaving most surfers of the internet at large, instead of locked down social media platforms, as infonet surfers.

The problem is that, fundamentally, the interests of those who come to the internet as infonet users and those who write for the internet as pesonanet writers are just in conflict here: when someone comes to the internet for infonet, they don't want to deal with personanet. But personanet writers want everyone, even infonet users, to be forced to engage with their sites directly, or not at all, because they want a social return on investment for their effort.

Previously, because personanet writers were the ones in control of the content — both in terms of where it was displayed and, perhaps more importantly, how it was expressed — they were able to coax infonet surfers to trawl through their 35 different personal blogs (with everyone telling the story of how to bake an apple pie starting with the universe) to find a useful technical detail, incidentally serving the personanet's interests. This was annoying for infonet users, and probably for the most part didn't lead to the types of interactions that personanet writers actually wanted (since human beings can open 35 tabs and skim each of them or Ctrl+F through them to find what they want, then close the tab immediately, never even seeing the navigation bar or anything else, too, not just AI), but it gave them some sense of satisfaction.

The problem that AI with web search is presenting for those who write primarily on the personanet is that it allows people to convert from the personanet to the infonet. As LocalGhost says, "I spend hours writing these posts and AI spends seconds summarising them." AI can convert a long post full of personal details and storytelling, or instructions with a particular taste and spin, on a unique website, into a short, to the point extraction of just the information that was needed by the user, synthesized from multiple places to answer their precise question, in a consistent centralized interface controlled by the user instead of by the writer, via an agent acting on their behalf. Essentially, it brings information to users, instead of users having to go to the information. For someone seeking the infonet, this is great: short, to the point, focused, general instructions — they don't care about personal taste and touches, or building human connection, they just want to get something done, and they want to get the information in a single consistent interface they control, instead of having to go out and trawl various places, all with unique interfaces that could be ugly, inaccessible, and difficult to read or navigate. This was the promise of the semantic web! But if your goal is to build human connection, to have someone see your unique expressions, writing style, website design, and to explore other things on your site and follow you on RSS, then this is an affront, especially if you're funding yourself with ads and speaking tours.

But although AI makes surfing the infonet easy, instead of being forced to engage with the personanet how personanent writers want to force infonet users to, and that makes them angry or sad, I'm not sure this is ultimately a bad thing.

First of all, because most infonet users, forced to interact with a personanet site, will definitely, 100%, never use the site like personanet users hope they will. If anything, it might even make the infonet users reset the personanet writer if they've buried useful information in too much personanet stuff. And if personanet writers want personanet users, well, those users will come find them when they're ready and in that frame of mind, and in that frame of mind, they're much more likely to interact in precisely the way personanet writers want. In fact, I think the vast majority of the total internet usage is, technically, personanet, so this is hardly actually a scarse resource — it only seems that way because of social media dominating the personanet, leaving little left for those not on social media; and that's a problem that might be fixable. So it isn't clear how much is lost by not forcing users of the infonet to interact directly with personanet.

Second, because while AI converting from the personanet to the infonet technically "goes against the author's wishes," but in my opinion, once you post information intentionally publicly, you lose the right to dictate and whine about how people are using it when it doesn't actually take anything away from you or hurt you in any way. Currently, there are economic concerns regarding lost traffic meaning lost ad revenue, but I don't think that's the core concern for personanet writers.

And third, because I think a more mindful, intentional split between the infonet and the personanet might actually be a good thing. Perhaps a semantic hypertext Wikipedia like internet designed for AI agents as well as more traditional org-mode-like hypertext database and lookup systems to crawl and correlate for users, that's mostly APIs and minimal plain text pages, where you pay a couple cents per lookup or something like Project Xanadu envisioned, and a neocities-like place full of digital gardens and personal posts and stories. And unlike previous attempts at an infonet (the aforementioned Semantic Web coined by Tim Berners-Lee), maybe this one can catch on, precisely because it isn't sequestered from the personanet, but can connect to and use the information there by transforming it using AI.

Crucially, AI web search based infonet still links back to the personanet sources, usually quite prominantly as well, providing ample on-ramps for users who genuinely want to engage with the personanet to do so. So again, it really isn't clear to me how much actually desired readership the personanet is losing due to all this, versus how much this is about just trying to control how people engage with what you publish publicly to try to lock down some sort of guaranteed return or make sure they're "thinking about it right."

The rise of AI as a way to access infonet seems like it might make the semantic web possible: where once it died because coming up with a data schema and ontology to represent wide varieties of fuzzy data was simply too difficult to be worth it, if it was even possible or the right paradigm in the first place, now the semantic web has the material conditions necessary to exist through AI, which can convert the fuzzy data that people actually post into data that's useful for the semantic web, combined with the switch from a melieu of XML and verbose standardized schemas to things like JSON documents, which don't require a fixed schema, that can be connected together with hypertext document references so that a rigid ontology isn't necessary, like what we see with Firebase and MongoDB; all of which is made more practical to query and traverse robustly through the use of AI to generate the queries and error handling needed to handle these schemaless hyperdatabases.