Every kind of Google Search result was bad, and results for the names of products were the *worst*, an endless cesspit of affiliate link-strewn puffery and scam sites.

It's not that the internet lacks for high-quality, reliable reviews. There are plenty of experts out there who subject a wide range of products to careful assessment, laboratory tests, and extensive comparisons. The sites where these reviews appear are instantly recognizable, and it's a great relief to find them.

3/

One such site is Housefresh.com, whose proprietor, Giselle Navarro, runs a team that produces extremely detailed, objective, high-quality reviews of air purifiers. This is an important product category: if you're someone with bad allergies or an immunocompromising condition, finding the right air purifier can exert enormous influence on your health outcomes.

4/

As good as Housefresh are at reviewing air purifiers, they are far less skilled at tricking Google. The world champions of this are spammers, content farms that produce garbage summaries of Amazon reviews and shovel them into massive, hidden sections of once-reputable websites like Forbes.com and Better *Homes and Gardens*, and thus dominate the Google results for product review searches:

pluralistic.net/2024/05/03/key…

5/

Google calls this "site reputation abuse" and has repeatedly vowed to put a stop to it, and has repeatedly, totally failed to do so. What's more, Google has laid of more than 10,000 workers, including "core teams," even while spending tens of billions of dollars on stock manipulation through "buyback" schemes:

qz.com/google-is-laying-off-hu…

6/

Of course, the Housefresh team are smart cookies - hence the high caliber of their air purifier reviews - and they *could* apply that intelligence to figuring out how to use SEO to trick Google's algorithm. Rather than doing so, they took the high road: they applied all that prodigious analytical talent to researching and publishing on *Google's* systematic failures - and even collusion - with the spammers who are destroying the web.

7/

This month, Housefresh released its latest report on Google's enshittification, this time with an emphasis on the "AI Overviews" that now surmount every search results page. Google has widely touted these as the future of search, a way to bypass the ad-strewn, popup-obscured, AI-sloppified (!) pages that it is seemingly powerless to filter out of its search corpus:

housefresh.com/beware-of-the-g…

8/

Beware of the Google AI salesman and its cronies

Exposing the overly salesy AI Overviews that will push you to buy bad products and exploring the system making it possible.

^{Gisele Navarro (HouseFresh)}

Rather than hunting through these SEO-winning garbage pages, you can simply refer to Google's AI Overview, which will summarize the best the internet has to offer, in hyperlegibile black sans-serif type on a white background, with key phrases helpfully highlighted in bold.

9/

Most critiques of AI Overview have focused on how these AI Overviews are a betrayal of the underlying bargain between the web and its monopoly search engine, whereby we all write the web and let Google index it for free, and in exchange, Google will send us traffic in proportion to the quality of our work:

bbc.com/future/article/2025061…

10/

Is Google about to destroy the web?

Google says adding more AI to its search engine will rejuvenate the internet. Others predict an apocalypse for websites. One thing is clear: this era of online history is closing.

^{Thomas Germain (BBC)}

This is true, as far as it goes, but it doesn't go far enough. Google is a platform, which is to say, a two-sided marketplace that brings together readers and publishers (along with advertisers). The bargain with *publishers* is that Google will send them traffic in exchange for access to their content. But the deal with *readers* is that Google will help them answer their questions quickly and accurately.

11/

If Google's marketing pitch for AI Overviews is to be believed, then Google is only shafting publishers in order to double down on its bargain with readers: to give us better access to high-quality information (recall Google's mission statement, "To organize the world's information and make it useful"). If that's true, then Google is the champion of readers in their long battle with publishers, a battle in which they are nearly helpless before publishers' abusive excesses.

12/

This is a very canny move on Google's part. Publishers and advertisers have more concentrated money than readers, but the dominant theory of antitrust since the Reagan administration is something called "consumer welfare," which holds that monopolistic conduct is only to be condemned if it makes consumers worse off.

13/

If a company screws its workers or suppliers to deliver better products and/or prices, then "consumer welfare" holds that the government should celebrate and protect the monopolist for improving "efficiency."

But all that is true only if Google AI Overviews are good. And they are very, very bad.

In the Housefresh report, titled "Beware of the Google AI salesman and its cronies," Navarro documents how Google's AI Overview is *wildly* bad at surfacing high-quality information.

14/

Indeed, Google's Gemini chatbot seems to *prefer* the lowest-quality sources of information on the web, and to actively suppress negative information about products, *even when* that negative information comes from its favorite information source.

Indeed, Navarro identifies a kind of madlibs template that Gemini uses to assemble an AI overview in response to the query "Is the [name of air purifier] worth it?"

15/

> The [model] air purifier is [a worthwhile investment/generally considered a good value for its price/a worthwhile purchase]. It's [praised/well-regarded] for its ability to [clean the air/remove particles/clean large rooms]. Whether the [product] is worth it depends on individual needs and priorities.

16/

This is the shape of the response that Google's AI Overview shits out when you ask about *any* air purifier, including a model that Wirecutter called "the worst air purifier ever tested":

nytimes.com/wirecutter/blog/wo…

What's more, AI Overview will produce a response like this one even when you ask it about *air purifiers that don't exist*, like the "Levoit Core 5510," the "Winnix Airmega" and the "Coy Mega 700."

17/

The Worst Air Purifier We’ve Ever Tested

In this week's newsletter: The worst air purifier we've ever tested, plus the best new iPhone 11 cases.

^{Ganda Suthivarakom (The New York Times)}

It gets worse, though. Even when you ask Google "What are the cons of [model of air purifier]?" AI Overview simply ignores them. If you persist, AI Overview will give you a result couched in sleazy sales patter, like "While it excels at removing viruses and bacteria, it is not as effective with dust, pet hair, pollen or other common allergens."

18/

Sometimes, AI Overview "hallucinates" imaginary cons that don't appear on the pages it cites, like warnings about the dangers of UV lights in purifiers that don't actually have UV lights.

Google argues that AI Overview won't displace traffic to the sites it summarizes. The company points to the fact that the statements in an AI Overview are each linked to the web-page they come from.

19/

This is a dubious proposition, predicated on the idea that people looking up a quick answer on a search engine will go on to follow all the footnotes and compare them to the results (this is something that peer reviewers for major scientific journals often fail at, after all).

20/