Other What Makes Your Life, Meh?

Last year Revolut discontinued its "Freelancer" business accounts, so I got moved onto the "Basic" tier. Little insulting, but whatever, the fee was still £0.

Naturally, the "Basic" account is now getting a fee: £10 a month. But for that, they'll double my fee-free transfers from five to 10 per month! Hooray!

...I use one. Actually, I don't even use that, 'cos it's a Revolut-to-Revolut transfer, so always free. Joy.

EDIT:
And I've had to send a Corsair 128GB rugged flash drive back, 'cos it was getting 8MB/s sustained write. Eight megabytes a second, in this the Year of Our Lord 2025. It'd have taken four and a half hours to fill the drive!

 

Was looking at those EiA magazines @Yaka recommended, and found an interesting article on fractal image compression - I used to save *everything* as FIFs back in the day, and of course nothing opens 'em now.

Trouble is:



...the scan's too low quality to actually see any differences. Heck, it looks to me light the right-hand side is blurrier than the left - but apparently that's the least compressed slice.

Hence why I'm keen on scanning my own mags!

 

Fifteen quid's worth of plant food, currently feeding the cardboard box it came in. Can you believe the delivery oik had the temerity to dump it in the parcel box and run?!



'course, it's an Amazon Marketplace seller (I know, I know...) so no nice easy refund. Can't even send it back easily, 'cos it's apparently a postal-restricted item(?). <sigh>

 

The slow, but rapidly accelerating, death of the World Wide Web.

I remember first getting online. The web was absolutely amazing to me, after years of dialling up to BBSes. Graphics! Sounds! Tiny videos that took you an hour to download! Hyperlinks! Email! Directories! Search engines!

It felt like the world's information was at your fingertips.

Now, we've all seen what's happened. The growth of invasive advertising. Registration walls, and later paywalls. Silos. But readily available large language models, spewing nonsense as a service? That may be the straw that breaks the camel's back.

This morning, my youngest asked me a simple question: are there any famous German Shepherd dogs in books. The sort of thing the World Wide Web would have been fantastic for in the past. So, I searched - using DuckDuckGo, 'cos we've all seen what's happened - and continues to happen - to Google.

The first hit? "20 Famous German Shepherds on TV and Movies - Pet Reader." Close enough, let's have a look! Starts with the telltale fluff indicative of ChatGPT then starts naming dogs. Rin Tin Tin: a German Shepherd! Good start! Strongheart! Yes! Roy Rogers' Bullet, still on track! Thank you, wonderful website! Superman's Krypto! Wait, that's... that's not a German Shepherd. Inspector Rex! Okay, yes, good, that's a German Shepherd. We're back on track. Ace of Hearts (2008, not the Lon Chaney one), yup! Hooch, from Turner & Hooch!

HOOCH FROM TURNER & HOOCH? CHUFFIN' HOOCH?

THIS IS HOOCH:



(Hooch, right, pictured with some actor or other.)

Other suggestions include Duke, from the Secret Life of Pets and its sequel:



Brain from Inspector Gadget:



And Kaleb from Game of Thrones. First, there are no German Shepherds in Game of Thrones, as far as I can recall. Second, this is Kaleb:



Now, this is the top ranking hit for my search. The best of the best, says DuckDuckGo (well, Bing, really.) And the results pages are absolutely filled with this dreck. Some are obvious, some more subtle - but you end up second-guessing everything, and then trying to figure out what's real and what's not almost always relies on going to Wikipedia 'cos you can't trust anything else to not also be LLM slop.

It's awful.

And it's getting worse. And will continue to do so, until the bubble bursts. Then it's unlikely to get better, as Pandora is sat there with an empty box and a guilty expression on her face, but at least it might stop getting worse.

For now? I'm having to go back to using a list of known-good sites as sources, or the old trick of searching for date ranges before the LLMocalpyse - which doesn't even work properly any more, 'cos a lot of sites are backdating posts to make it look like they've been established longer than a week...

Ugh.

 

I remember being with an ISP - maybe Freeserve, I forget - where you paid a flat monthly fee and could stay on as long as you liked, on an 0800 dial-in number! The catch: you'd get disconnected after two hours.

I wanted to download the Kung-Fu Mod for Max Payne. Approximate download time at 56k? Two hours two minutes. No, the server did not support resume. Ten hours later I lucked into a connection just slightly faster than average and snagged the file before the timeout.

Worth it.

But back to the topic of the death of the web. What can we do about it? Well, at the risk of using my passion for computing history as the hammer for every problem's nail-like appearance, I believe the solution can be found in the past. We didn't always have algorithmic search engines, and we certainly didn't always have ones that used an LLM to "summarise" sites instead of sending you there¹.

Skipping Gopher and the like, the first majorly successful commercial search engine² was Yahoo!. Except it wasn't called Yahoo!, it was called Jerry [Yang] and David [Filo]'s Guide to the World Wide Web. It was not a search engine at first, it was a web directory. Even when it became a search engine, the search function only searched the directory. What's a web directory? What it sounds like: a list of sites on the web, typically categorised by topic. The list was curated by hand. Sites that wanted to be in it paid $600 a year to be listed. There was no web crawling, there was no algorithm, there was just Jerry and David typing in links. You knew a site on it was halfway decent, 'cos otherwise Jerry and David wouldn't have put it on there.

Yahoo! wasn't the first web directory, and there's even some still out there: Curlie is based on the ashes of the DMOZ project, run by Mozilla before it went... <gestures at modern Mozilla>. Web directories can't be gamed by "search engine optimisation" tricks (but can be gamed by the old-fashioned method of throwing a brown envelope stuffed with twenties at the person running 'em). They don't use AI to badly "summarise" stuff and steal clicks from sites. They don't list crap, or ephemeral stuff, or the like. Yeah, they're not "web scale" - the reason we switched to algorithmic search engines and crawlers is 'cos new pages were appearing faster than Jerry and David could type. But who cares? I search for "key lime pie recipe" on Google, I get 116 million results. I don't want 116 million recipes, I want one. A web directory will give me that.

Crawlers and full-text search have their place, but for deeper research and fast-breaking topics; they absolutely should not be used to find a particular website, and these days that's just as well 'cos they can't.

What else can history offer? Web rings! You've got a site all about Hornby model trains? Join the Hornby Model Trains Webring! You put a little graphic at the bottom of your page, and people can click through to visit other pages by other fans of Hornby model trains. Everyone gains a little bit of traffic, it's easier to find related sites and get second opinions, nobody loses, and no money changes hands.

Forums! I don't need to tell you how great they are, you're on one.

But what can you do, right now, to help? Easy: stop giving this shite the shine of legitimacy. Every time you go to Grok and have it generate a funny picture, ironically or otherwise, that goes down as a user in Pedo Guy's investment deck. Every time you try to goad Gemini into telling you to eat rocks, it's recorded as more usage - and chews through the same energy and water resources as any other. Every time you ask Claude a question only to have to search anyway to validate the answer? Every time you autocomplete subtly-different boilerplate in your latest GitHub project instead of having a boilerplate library? Yup, you're hastening the demise of the web. You monster.

Just... don't. Switch to a search engine that either doesn't have "AI" or lets you turn it off. Stop using Gemini, Claude, Perplexity, all the gen-AI crap. Turn Copilot off. Yes, all the Copilots: the Windows one, the GitHub one, the one that used to be called Microsoft 365 and was called Microsoft Office before that and you could buy in an actual box on an actual disc (or bunch of disks) in an actual shop for a one-time, non-recurring payment. Turn 'em off. 86 the domain names in your DNS, if you can. Don't let them use you as evidence people actually want this.

Err... didn't mean to write that much. I've got words to write, sure, but different words, and I get paid for those. Whoops!


1: This is a problem for more reasons than "it hallucinates and tells you to glue the toppings to your pizza and eat three small rocks a day." The modern web is, like it or not, funded by advertising. Advertising requires eyeballs, which requires visitors. When Google et al take your page's content, remix it, and present it to their user, the user doesn't click through to your site. You get nothing. Nada. Bupkis. A big fat zilch. Wanna know why all the paper magazines have folded (pun entirely intended)? It's 'cos nobody wanted to advertise in 'em any more. Wanna know what'll happen when nobody wants to advertise on websites, either? Give you three guesses, first two don't count.

2: Aided in its success, in no small part, by Netscape, fact-fans, which put a honkin' gert link to it in the toolbar.

 

Also, and I should (and will) post this in Firefox thread, beware if you install LibreWolf and sign in to Firefox Sync: LibreWolf is configured to clear everything on exit, and will synchronise that setting to all your other Firefox browsers through Firefox Sync. Go ahead, ask me how I know(!)

 

Yeppers. Which is why I'm calling for a return to web directories. I've been thinking it for years - I've got a folder in my RSS reader called "Curated Web" full of stuff about why we need it. I even have a workable plan for launching one (elevator pitch: trusted staff contributors, broader contributions in partnership with library reference desks who get a cut of revenue 'cos libraries are chuffin' struggling too, potential for broader contribution using a Reddit-style voting system, category-specific advertising which does *not* track individual users.)

Short-term, though, it's about collapsing the AI bubble sooner rather than later (so turn off AI in your search engine!) and doing the best you can with what's out there. If you want more control, I can recommend SearXNG - self-hosted works, or there are public installations you can use. It's a metasearch engine and still relies on Google/Bing/Yandex/and so on, but that's the best you're going to get right now.

You could use Curlie, but most of the content on there is from DMOZ and a decade or two old - and it's constantly timing out on me.

Yup, &udm=14. Takes you to a list of paged search results. No summaries, no client-facing "AI," nothing but search results. If you'd prefer not to send all your search queries to some randomer, you can find instructions for setting the default search in the browser of your choice to append the magic string here.

But, yeah, Google could turn the "web" tab off at any time. Which is why I don't use Google (directly) any more...