All nonsense thus... / Shall in thee centre, from thee circulate

December 29, 2024 1:59 AM

We are at a moment in the history of the web in which the link itself – the countless connections made by website creators, the endless tapestry of ideas woven together throughout the web – is in danger of going extinct. So it’s pertinent to ask: how did links come to represent information in the first place? And what’s at stake in the movement away from links toward AI chat interfaces? from A linkless internet [Aeon; ungated]
posted by chavenet (14 comments total)

links make sense as early victims of the AI drevolution

ftfy
posted by HearHere at 3:53 AM on December 29 [2 favorites]

The trouble with people is that in the aggregate we aren't smart. In the aggregate, we're lazy, easiest-path types and we're not smart.

Individually, sure, people might think "perhaps relying on random AI hallucinations is a bad idea and perhaps it is good for me to maintain the ability to read original documents, select a variety of sources for myself, take my own notes rather than daydream and otherwise keep from becoming a human tuber", but in the aggregate it's all "sign me up for AI summaries and chatGPT".

Individual humans sit around developing absolute garbage murder poison to unleash on the stupid aggregate so that the individuals can get stock options and bonuses; the aggregate just takes in the garbage murder poison because it's quick and cheap.

And that is why, if I had not bought into a lot of happy crap about how people were wonderful and I was just an introverted weirdo who should learn to fit in, I would have gone off the grid twenty five years ago and would be sitting at least moderately pretty now instead of waiting for some fatal disaster to strike.
posted by Frowner at 6:36 AM on December 29 [10 favorites]

And, of course, for how long will we have access to the original stuff? What happens if free search engines are just shopping and AI summaries? (I assume there'll be a paid option for those with interest and disposable income.) Will it just be dueling nonsense AI summaries?

Before the internet, of course, we had libraries - but even if you have access to a library, the human tendency is going to be to accept the AI trash because it's right there rather than go to the library, which takes substantial time. We're all used to at least semi-reliable immediate answers to an endless variety of questions, so people aren't exactly likely to go back to using the library. And newspapers and magazines aren't reliable either.
posted by Frowner at 7:05 AM on December 29 [3 favorites]

Most libraries around here seemed to have dumped physical reference books, periodicals, etc. I think it’s going to be AI doofuses all the way down.
posted by njohnson23 at 7:19 AM on December 29 [2 favorites]

While these ["Storyspace schools"] works achieved critical attention in the mid to late 1990s, they haven’t had a lasting impact.

Depends who you ask.
posted by Lemkin at 7:31 AM on December 29 [1 favorite]

Hypertext never made any sense, and we saw this especially in the early days of the web, but you can see it in Wikipedia now, or in one of those editions of the Dunciad with all the footnotes: Text wants to be linear, it wants to be sequential, little clumps of letters or words digested at a time. It's a tiny keyhole in a door, and on either side of the door are these great masses of complexity--the world on one side, your associative mind on the other. Hypertext didn't help represent the complexity of either, it only interrupted the flow of text. It made your mind skip from one task to another. Reading a book with end-notes is like being told a story by a little child who keeps pointing out everything else that catches his eye, and won't get to the point.

In a way, LLMs aren't too different from hypertext links, but automated, with the logic deeply hidden from you. They are an extreme form of the same interruption, the same reasoning that what should interest you is not the text itself but what can be done to or said about the text.
posted by mittens at 8:14 AM on December 29 [2 favorites]

Text wants to be linear, it wants to be sequential, little clumps of letters or words digested at a time.

And yet you have tables of contents, chapter numbers, sections, page numbers, indices cross-referenced to those page numbers, notes, lists of characters, map sections. It's almost as if people have wanted to facilitate semantic access to information from the very freaking time they started writing and put what the text was about in the first line of the clay tablet.
posted by sukeban at 8:30 AM on December 29 [12 favorites]

In some ways, AI feels like a distillation of the old search engine approach. Ideas only become reified within a model if cross-referenced enough times across the web, leading to a sort of conceptual PageRank -- the more niche or obscure the idea, the less likely you are to get anything useful. Knowing what sorts of questions are likely to be productive is akin to the "Google-fu" of efficient queries and operators. Leading providers are trying to better ground their models (and mollify publishers) by linking to sources and providing only capsule summaries of articles. And of course the real internet has no shortage of misinformation and delusion that requires critical thinking to evaluate.

IMHO, the biggest current threat to the link model is the dominance of social media apps that endeavor to keep users siloed inside proprietary networks, with the concept of permalinks and outside domains hidden behind the UI and temporary, stripped-down browser windows. Sometimes I get the feeling that the average grade schooler thinks of a URL the same way I might have thought of a card catalog at that age -- I knew they existed and what they were for, may have even been required to learn them, but they were antiques irrelevant to my information diet.

(See also: how the ubiquity of the closed, simplified, user-friendly smartphone and tablet OS is making younger users illiterate at basic stuff like software installation, file management, and troubleshooting.)
posted by Rhaomi at 8:35 AM on December 29 [4 favorites]

(Recommended reading: Shady Characters by Keith Houston, Index, A History of the: A Bookish Adventure from Medieval Manuscripts to the Digital Age by Dennis Duncan, A Place for Everything: The Curious History of Alphabetical Order by Judith Flanders)
posted by sukeban at 8:35 AM on December 29 [4 favorites]

Rhaomi - I spent the first half my working career in tech. There, when talking about product plans, and someone proposes some new thing, people would respond with, “Yeah but… people in the real world don’t…” In the early 90’s, I started working in that real world, and was shocked to see how ignorant people were about using the computers they all had on their desks. The basic stuff you listed were all beyond their skills. The IT dept where I was wasn’t much better. They also didn’t believe in user training, possibly to give reason to their existence.

At Apple in the 80’s, hyperlinks aka HyperCard, were going to undermine the old systems and change people’s relationship to computers since with HyperCard, everyone can build their own model/system of data, etc. Computers were to become the tools to make everyone a creator, since everyone had the undying need to make things. Yes, the WWW did bring links across computers everywhere at first, but with a system that was too complex for the average person, who really just wanted to be fed information. Computers, etc. became just TVs, with a bajillion channels all sponsored with ads. Just like old time TVs. That promise from 40 years ago never happened. Links are just another ephemeral digital object destined for the bit bucket, as is a lot of content that now resides in just digital form.
posted by njohnson23 at 9:11 AM on December 29 [4 favorites]

People didn't need closed simplified phone UIs in order to not learn file management skills

Most people did that just fine with general-purpose computers
posted by Aardvark Cheeselog at 10:24 AM on December 29 [2 favorites]

>Links are just another ephemeral digital object destined for the bit bucket, as is a lot of content that now resides in just digital form.

That seems akin to saying TCP/IP or https is a dinosaur and soon to be discarded. I could see format or convention around links changing, but I can't see it being discarded. Half of everything I see on the social websites is just a link plus a snippet of comment.

Or is this meant more like the linked things are being discarded? I have seen a number of younger users delete whole accounts they found embarrassing, older ones when it gets them in trouble, and every once in awhile there's another grand self purge when people are reminded that Big Brother is watching.
posted by cult_url_bias at 1:28 PM on December 29 [1 favorite]

I know someone working at a university helping faculty manage their online educational material. Most of the time is spent identifying and trying to fix broken links. And there are a lot of them, some can be relinked but others are left dangling. And we are not talking about links from a decade or more ago. And a lot of these links are to other academic postings. Yes, a book may have a table of contents and an index, but if the pages of a book are gone, they are worthless.
posted by njohnson23 at 3:46 PM on December 29 [1 favorite]

Hypertext never made any sense...
I assume you're referring to the Web, as that is the subject of the article, but my first contact with hypertext was IBM's Bookmaster documents, which was how we got our first electronic service manuals.
(Nowadays, these would be PDFs)

It was extremely convenient, and not just in the index and TOC, but within sections as well. A repair manual is not a linear narrative, and being able to jump from one repair section instruction like 'first remove the outer cover' to those instructions and back easily made a lot of sense.
(I still prefer paper documents, though)
posted by MtDewd at 5:30 PM on December 29 [2 favorites]