08 Nov 2026 Search

Previously: Notes and links from my talk at Texas Linux Fest

This is an extended version, with links, of the question and answer session from my Texas Linux Fest keynote yesterday.

First of all thank you great talk. A lot of this focused on Google and Facebook, or Meta. Is Apple any better? Apple has aspirations.

So the problem is enshittification, and this is Cory Doctorow’s term for when a service starts out being good for the customers and good for the companies that use it, then they start ripping off the customers while keeping the companies happy, and then surprise they start ripping off both sides.

Apple is earlier on the enshittification curve than other companies, but they’ve still got a bunch of sneaky stuff that they have no alternative but to turn on in order to keep “number go up,” right?

If you follow W3C inside baseball, you will see Apple participation in the attribution cartel, which is a project to put advertising tracking into the web browser. Apple Safari already has something like it. It’s a low-profile thing and it’s moving along fairly slowly. Apple obviously doesn’t need to squeeze on surveillance because they’ve got a revenue stream from devices that people really like to buy. But in order to keep number go up, they are eventually going to squeeze more and more.

I have a question about reviews, which they were supposed to be a guard against scam sites and all that, when you can look at user reviews, customer reviews, and get a signal that there’s product quality or brand trust, but it seems those have been co-opted as well. Is there anything in that space that’s fighting against fake reviews or bot reviews? Is there a trust emerging there?

The reviews problem is a big one because the original algorithm for Google was PageRank. You’ve got trusted site links to other trusted sites and you can calculate the trust level of various sites. And so that was a really promising approach because what it does is moves the reputation graph outside of one company’s black box. And the reputation graph, in this case it’s represented by linking relationships among sites and blogs.

In theory you could write your own PageRank-ish algorithm to check it as well. Now in in recent years, Google has moved away from that and they’ve moved toward more hidden black box signals within their own data center, and they’re actually discouraging websites from contributing to the reputation graph.

My website, I don’t really care about my search engine ranking. I will link to legit businesses in a lot of different categories. But one of the prime pieces of SEO advice that they give is if you want to do well in Google, well, purge your site of everything except for one subject. Google wants to keep that reputation graph in-house so that more people are making decisions based on a Google ad and fewer people are making decisions based on the reputation of a site.

The level to which Google is squeezing on stuff like review content and ability to figure out the reputation of a site, that is a massive opportunity for someone to build a PageRank-like search engine and start to recapture some of those those reputation relationships that that people are still actively sharing.

A site that’s good for understanding the reviews issue is HouseFresh. They evaluate air purifiers and they actually do the tests and they consistently lose out to slop sites in Google because of that reputation graph problem.

Were Internet ads a good idea? And if you were able to rewind time and change reality, would you be willing to share like a few things that you think would have made it different?

Alternate reality time.

A better timeline would be the one where lawyers dug up the California Invasion of Privacy Act and applied it to web tracking in the 1990s, and didn’t wait until the 2020s.

If that legal argument had been made when conversations about things like third-party cookies were first happening, and more people had realized that cross-context tracking is legally actionable wiretapping, then a lot of sites would have been motivated to invest more in the kind of control over audience data that enables a sustainable ad-supported site.

See McKinley Park News - Building a Trustworthy News Business.

Among the worst abuse of integrated third-party scripts occurs through advertising networks and advertising technology (or “adtech”) designed as much to profit from exploitation of user data as enabling relationships between buyers and sellers. Too many news publications have swallowed the poison pill of easy- to-implement, revenue-generating third-party ad systems, failing to consider how this can harm users, denude business prospects and destroy trust.

The McKinley Park News does not use third-party ad systems for a host of reasons:

Read the whole thing.

You talked about a lot of things I generally knew about but didn’t really understand how the whole advertising industry fit together and why it was so harmful. My question is, do you have a good resource that one might use to evangelize to a less technical person about what you’ve talked about here in this presentation?

Consumer Reports ran a feature called CR Data Bootcamp where there’s a series of lessons that you can go through to clean up your your privacy situation. I think that’s a really good one for normal users. But the problem is that putting all this extra work on normal users is not sustainable. And CSO Online really had the right idea, which is you’re the system administrator. You know this bad thing’s happening. Push out the the tooling that will protect your people rather than counting on everyone to keep up with all of this and make the right choices. So yes, regular users should be checking a few things, but a lot of it is on the open source scene at least in part because we got here because of the open source scene.

(I also have effective privacy tips, some of which the sysadmin can handle, and some of which would have to be done by the individual.)