The two-class platform

Then:

“I would be ashamed to admit that I had risen from the ranks. When I rise it will be with the ranks, and not from the ranks.” —Eugene V. Debs

Now:

“[site name redacted] is unusable without an ad blocker.” — employee of a web advertising company, overheard at [tech event redacted]

The World Wide Web of today is almost two platforms. Back when “Wintel” (that was Microsoft Windows on Intel x86, kids) dominated the market, a Microsoft Excel developer who brought home a copy of Microsoft Excel was getting the same experience as an ordinaryMicroWarehouse shopper. People certainly had better and worse PCs, but there was no productive release of Excel for the elite and shitty Excel for the regular customers.

The web, though, is a two-class environment: a tolerable business class version for those with the right tools, and amess for everyone else. The upper class is harder to define, and to get into, than it looks. About half the peoplesay they have some kind of ad blocker, but in practice the fraction of web traffic that’s affected at all is below that, and the fraction that has full protection includingsearch andYouTube advertising is a lot less. Many of the “ad blockers” that are easiest to find using search engines and browser extension directories provide only partial protection, or are outright adware or worse. Reasonable people disagree on how to measure ad blocking, and it has only gotten harder to measure as the proportion of “AI” scraper bots goes up. A lot of those look realistic, and bots differ in their ad blocking behavior. Some want to load the ads for adfraud, while others just grab the content they need.

But I do have some better numbers that I think are starting to show how we are now on a two-class web. Hope to be able to share when FIXME [add site info and link] puts up a page about some user research. The privileged class here doesn’t necessarily match up with the10% of the people who buy 50% of the stuff in the US economy as a whole, but there’s some overlap.

The inconvenient situation here is that some kind of element hiding and/or other page finagling has gotten to be a necessity. About as much as an email spam filter. Obviously you need to block certain ads, especially Google Search (checks RSS reader for a fresh search ad scam story…here:Search Ads as a Vector for Travel Scams) and you need to block the script that trains Meta’s AI toscam you, but browsers have other reasons to modify the page, too. Den Odell points out thatBrowsers Treat Big Sites Differently—when a site only tests on Google Chrome, the other browsers have to work around it.

Of course, not all the stuff that’s bad about the web is fixable with tools alone.Pulitzer-winning newsrooms are quietly publishing mountains of gambling slop, and in order to keep a productive experience with search you need tofix it which includes some habit changes as well. If we’re lucky we can get the web to the level of protection that email has—most email users are reasonably well protected, for historical reasons. Email spam filters fortunately squeaked through back before the big enshittification trend. If spam filtering were newly invented today, the same companies that form the web’sattribution cartel would be standardizing on an approach that lets their own spam through.

Some good news. It has been about a week since I put up next steps on tinyMyTerms, in which I added a hella annoying Accept Terms of Service dialog to this site—but made it invisible to anyone with a decent element-hiding ad blocker configuration. And zero complaints so far. Element-hiding ad blockers working, I’m (sniff) so proud of all of you.

We still need an answer to the problem of the two-class web. And making the web more fair by ending the maltreatment of the “underclass” is going to work better than the alternative. I think it’s going to take some humility, recognizing folk privacy practices and bottom-up applied behavioral economics, and working with them. Sometimes just hide the annoying thing is the productive starting point, then you can do the necessary paperwork on top of that. This doesn’t mean abandoning personalized advertising for those who want it—Rewarded Interest is a thing, and works. Eugene Debs again: I would not lead you into the promised land if I could, because if I led you in, some one else would lead you out.

Bonus links

The old world of tech is dying and the new cannot be born by Baldur Bjarnason. The ground is shifting underneath every industry that was built on the assumption that the US would protect and preserve the globalised status quo.

How Dangerous Is Anthropic’s Mythos AI? by Bruce Schneier. Even more interesting are the broader implications. The same searching, pattern-matching and reasoning capabilities that make these models so good at analyzing software almost certainly apply to similar systems.

OpenAI Sued Over ChatGPT Medical Advice That Allegedly Killed College Student by Maggie Harrison Dupré. ChatGPT recommended a dangerous combination of drugs without offering even the most basic warning that the mix could be fatal, added Matthew Bergman of the Social Media Victims Law Center.

Large Study Finds That Replacing Workers With AI Is Backfiring Badly By Krystle Vermes. The Gartner survey found that execs who slashed staff to invest in AI have seen the same financial gains as those who held onto their employees. In other words, attempting to replace workers with AI isn’t showing any detectable returns for these companies.

An Approaching Crisis by Brian Jacobs. Our trade bodies talk of advertising being legal, decent, honest and truthful. And yet they stay remarkably quiet when some of their vendor members behave illegally (running fake ads), indecently (undressing apps), dishonestly (designing addictive algorithms then denying having done so) and deceitfully (inflating ad metrics).

Spending Just 10 Minutes With AI Can Fry Your Brain, Researchers Find By AJ Dellinger. According to a new study by researchers at Carnegie Mellon, MIT, Oxford, and UCLA, just a 10-minute session with an AI assistant can lead to users significantly abandoning their own capacity for reasoning.

“Stanford report highlights growing disconnect between AI insiders and everyone else” by Sarah Perez. The U.S. also reported the lowest trust in its government to regulate AI responsibly, compared with other nations, at 31%.