So let's assume, just for the sake of argument, that you were foolish enough to try to make a living by writing high-quality, well-researched, technical articles about Linux and free-software development. I know that's crazy, but bear with me. In such a scenario, how does one succeed in a world increasingly full of stuff like this?

https://www.webpronews.com/linux-7-0-looms-large-inside-the-landmark-kernel-release-that-could-reshape-open-source-computing/

(I'll post no more links to that site, I promise).

These folks appear to take the stuff we humans write, inject a bunch of errors, then slop it out to the world.

If you were to engage in the silly quest described above, you would find that what you do is increasingly buried in the flood of this kind of material. Does anybody have any bright ideas about how one might survive in such an environment?

@corbet

@corbet I suspect it was really written by John Mastodon, not the John Marshall of the byline.

@corbet I don't know, but it feels like the same and potentially quixotic quest all of us trying to draw the line between "this is a tool" and "this is slop" face.

@corbet My subscription to LWN is entirely based on your reputation.

Curated, high-quality information, written by technical folks at the heart of Linux / FOSS communities for decades.

An island of quality, in a sea of regurgitated slop from stochastic parrots.

@corbet
An established name (such as you already have) helps. The subscriber base is unlikely to flee to return to re-read such fluff.

Keeping it relevant and not just a reproduction of e.g. press releases is important. But then lwn has never gone there that I recall.

I recall (perhaps incorrectly) linux magazine (or something similar) (horror of horrors, a paper based medium) copying a SuSe press release more or less word for word for their English version, where the German version actually

@corbet sometimes I wonder if we should create a browser extension with a list of websites that do shady stuff like this – which then shows a big fat "you are better than this, don't go there" warning if you accidentally try to open one. That way they would not get revenue von ads, so it might be in the interest of websites to not land on this blocklist.

But well, creating that extension and motivating users to install it is quite a bit of work – and maintaining the list is a hell of controversial work. Which is why a concept like this likely won't fly, unless we really really want that to happen.

@corbet IMHO, I don't see the problem: lwn is for a (slowly growing) group that takes open source seriously and/or professionally. I very much doubt any lwn subscribers would exchange it for "AI" generated crap. They're not competition. Did lwn had a drop in subscriptions after this LLM fever started? If so, I'm willing to step up my subscription tier to help make sure you folks are around forever.

@corbet
did a useful review. Guess which one I continued to purchase going forward.

(They ended up irritating me by pretending I had signed up for auto-renewal, when I needed a pause, but that had nothing to do with the content).

@ruivo Subscriptions are down a little bit (not massively) from a year ago. Determining causes is hard; I suspect that the slop flood, economic challenges, and geopolitical factors all play into it.

@corbet paywall the fuck out of the real content and just publish model poisoning garbage to non logged in accounts.
(This clearly won't work, but I've come to believe the web is fucked and that's not a very productive angle)

@corbet Blocked and reported to @kagihq SlopStop: https://help.kagi.com/kagi/features/slopstop.html

@corbet I think it's an environment we all need to deal with. AI seems to be a way for the uneducated to put more garbage into the interwebs.

It's up to the educated to find their way to valuable information and guide the to be educated.

@corbet
Given the question, I have to assume this is happenning.
And that lwn is losing either subscribers, or ad revenue from the site/non-subscriber access.

For non-renewing subscribers - try asking. You might not like the resposes you get, but there may be a grain of truth to find.

For ads, not sure. Maybe there are paths through making it more difficult to access content without ads. That said, my current employer has installed something which breaks https security (dope).

@corbet
So I have the problem of wanting to access, and maybe share, content, without giving away my login details.

@gezza The subscriber link feature is likely to be the droid you are looking for: https://lwn.net/op/FAQ.lwn#slinks

@corbet
That was the one I had in mind. But I have to log in to get it - which will not happen away from my home computer (at the end of an already long day).

@corbet Don't post links to the slop? Post links to your original writing instead?

@JustinDerrick Good advice but inadequate. We do definitely operate that way most of the time; here I thought it was important to call out what is going on.

But if I search Google for "Linux 7.0", I see that two out of seven of the "top stories" are from that site, including the one I posted. This stuff is being strongly amplified in ways that we're not well placed to counter.

@corbet Best answer I know of: Find everyone you can doing such work and promote each other.

If the web's major institutions are promoting inferior reporting, you'll need all the promotion you can get!

@corbet Very tough.

But such content mills can be reported in some places - e.g., https://blog.kagi.com/slopstop

How realistic it is to keep up with the slop? Unclear.

But validated "small web" sites (which is another layer Kagi supports) might be a way. Back to curated Internet site lists as if it's the 90s again!

@kernellogger Kagi has been mentioned a few times here.

This would also be a very good integration in a browser, either via Kagi's extension or maybe a selling advantage for something like Vivaldi?

@corbet
For most searches the internet is now broken. SEO made it pretty unusable, LLMs killed it.
I'm glad I subscribed to LWN a long time ago, because now I might never find it again.

@corbet
Bugger.
I would prefer not to transfer money to the US, but I'm ok with some exceptions.
I assume that a lot of people rely on AI summaries when searching and never see the source sites.
This future is awful.
@ruivo

@kernellogger @corbet There is at least one blocklist for uBlock Origin:

https://github.com/laylavish/uBlockOrigin-HUGE-AI-Blocklist

However, it appears to have been abandoned, already.

@kernellogger @corbet At some point I found someone made a list for Ublock you could sub to that did block a lot of AI sites.....naturally, someone posted it while I weote this.

@corbet I did not know what LWN was until today thank you.

I dont want to sound Pollyanna-ish, but cream rises. We learned to ignore listcicles, content written for SEO, and other poor quality content. I clock AI in a heartbeat (or human-authored but written like AI, in which case shame on them) and move right past. Not because its bad, just untrustworthy. We don't trust search results anymore because it's so infected. We'll come to a new equilibrium, at least until the next disruptor

@corbet @JustinDerrick DuckDuckGo does not show that website, but MSN (!) is near the top. To me, the awfulness under discussion is a small part of the wider problem of where people seek information. Using Google Search is bad, but at least it's not social media.

@corbet Perhaps increasing the window when LWN articles are publicly available should be increased from a week to maybe a month? Unfortunately, I feel that people engaging in plagiarism and slop generation is inevitable once they have access to LWN articles but delaying that slop generation may redirect traffic to LWN during that window and people may consider subscribing rather than waiting for a month.

Reporting this domain to some uBlock blocklists should also help. I'll raise a PR to add that domain to the Huge AI Blocklists project. Maybe there are similar adblock lists out there which are popular enough to make a difference once that domain is added to them.

@corbet I have almost no idea what would be involved, but if I were in this hypothetical situation I would consider having a business oriented separate, but affiliated, newsletter which is subscriber only and priced for business customers. As far as I can tell that is what @crankyflier.com is doing with https://crankyflier.com/ and https://crankynetworkweekly.com/

@corbet "Help us improve our content by reporting any issues you find." lol. no.

@hexchain yeah, after posting I already suspected that something like that existed already. 😆

But as I said: getting it to really fly is the bigger problem. To have an effect, a whole lot of people would need to have it installed.

@ayushnix @corbet "increased from a week to maybe a month": won't help much or maybe nothing at all, at least here in Germany I've seen websites that published articles written by some AI that were clearly based on texts that my former colleagues (I used to be writer/editor at heise.de) had published 24 hours earlier behind a paywall.

@kernellogger @corbet ah, in that case, getting that domain on popular adblock lists seems like the best course of action at the moment

@pchestek I was about to write something similar. Nobody wants to read slop, and if anybody actually does, they're not the kind of readers you'd win with factual writing anyway.

@ayushnix The length of the paywall period has been a concern since its inception in 2002... We want to be a part of the development community, and, I fear, walling off the content for too long would interfere with that.

@corbet that's going to sound quaint, but the answer, in my mind, is still this old internet we're on right now... #mastodon and the #federation is not exactly "old" but it certainly matches the old school usenet/email models of old, warts and all, but it feels mostly insulated from this madness... search engines are dying because of AI, but we still have, thanks to you, trusted sources of news, and those will stay places we can rely on, not by stumbling on them randomly, but by directly seeking content there, because we know it's there.

@corbet i'll be giving a workshop on "digital sovereignty, federation and mesh networks" and i realized that one of the alternatives to "google" i need to suggest is not only, say, Kagi or Qwant, but mostly Wikipedia and "your public library", because about a large majority of what people search for can be answered by studying there... it's more work than asking a LLM or SEO crap, but it's more reliable, makes you smarter, and it will be more likely to survive peak oil and the AI bubble crash

@corbet i just had another thought on this: i think search engines are dead. they were dying even before AI slop started hitting harder, SEO is a class attack against the founding principles of Page Rank that made search so good back then. instead, perhaps what we need to build is an open, "allow-list", topic-specific search engine, one that would index all sites (like LWN, Wikipedia, and so on) *and what they link to*... perhaps we could find good contents there!

@corbet I got one. We do some sort of trust based network of good work and _human non-AI authors_ and add a search engine to that.

@corbet We could have some democratic group, which gets us human curated list of good websites - not articles, sites - and they take different topics each. Everyone gets a corner, and then we pull some sort of community early yahoo combined with PageRank. DuckDuckGo got a crawler, maybe they would even help out and get us an exclusive corner away from their other results. And we would have every decision go back to a person, so if someone is really iffy - or an AI - we roll back what they did.

@corbet Of course we can't have just one person on the subgroup who up and leaves. We should do democratic groups.

@corbet And yes, authors are welcome.