Today’s Morning Buzz is brought to you by Raman Shah, an independent data scientist based in Providence, Rhode Island. Connect with Raman on LinkedIn or X/Twitter.
- What I’m working on: Workflows to help a team engage community members more consistently
- A hobby I enjoy: Car detailing
- What I’m watching: “Breaking Bad”
I’m a geriatric Millennial. I grew up during the search engine wars of the late 1990s and discovered the verb “to Google” at age 19. As a young nerd in the early-2010s, a job at Google in Search Quality was one of the most prestigious jobs a nerd could imagine, and the folks there certainly seemed to be doing their jobs well. We swam in relatively pristine waters of instant, trustworthy information.
Fast-forward to 2025, and we swim in an ocean of thick mud. We blurt half-formed thoughts into what used to be an address bar in our web browsers. In place of a tidy list of links to mostly trustworthy websites, we get back a pile of advertisements paired with a half-accurate essay produced by generative artificial intelligence. We’re supposed to fact-check this essay, but no longer have decent tools to do such a thing.
The increasing difficulty of finding accurate information is a story of corporate decay accelerated by generative AI. The work of local government leadership being so knowledge-intensive, it’s worth examining how best to navigate a muddy internet for the knowledge we need. There are, thankfully, some simple tricks, both across the open internet and within individual websites, that can take you a long way.
Tracking the decline of Google
It is not possible to talk much about searching the contemporary internet without talking about the dominant provider of search, Google. Google’s arc is a familiar, discouraging tale of Corporate America, in which essential infrastructure is controlled by a private-sector monopoly hungry to deliver ever-greater profits to shareholders. Numerous product decisions served this goal while gradually degrading the once-excellent search experience that made Google so rich and powerful in the first place.
The rise of generative AI accelerated the decline in search quality in two ways. First, its output displaces links to human-edited websites whose trustworthiness we can make judgments about. Instead, we contend with a statistical average co-mingling all of what the internet said, including untrustworthy parts of the internet. Second, generative AI made the internet itself far less trustworthy. Mass-produced synthetic content is a profitable form of information pollution that drowns out trustworthy information and makes search providers’ jobs far more difficult. This content also increasingly compromises the training data on which future AI models are trained, leading to a downward spiral in information quality.
So what do we do?
Searching the open internet
Google wielded its monopoly power so effectively that it’s easy to forget that there are alternatives. But with a little initiative, it’s possible to search the open internet in better ways.
I inconsistently use DuckDuckGo as one alternative. DuckDuckGo is a privacy-oriented search engine. My impression of its search quality varies. Actually paying for search as a subscription product is also possible. My ongoing experiments with Kagi are very encouraging; its search results at the time of writing are far higher-quality than Google’s. I used Kagi to research this article. Once dialed in, it’s possible to change your default search engine in all popular web browsers and assert some independence from what Google has decided you ought to do.
Ultimately, I think it’s untenable to expect accurate information in exchange for permission to be surveilled and manipulated. The conflicts of interest in this bargain are just too big. While the landscape of the corporate internet will keep changing, there’s room to keep looking for the rare corners where users’ incentives are better aligned with search providers’ incentives.
Searching within individual websites
In 2023, I wrote about the power of reading when robots can write. Collective human scrutiny is the biggest safe harbor of clear water in an ocean of dubious machine-generated mud. I wrote then about managing knowledge within your institution, but some websites manage knowledge in the same way.
It’s worth taking advantage of these websites! Sometimes, trustworthy knowledge lives on the websites of institutions such as federal agencies and universities. A lot of general knowledge exists on blogs of individual trusted writers, or on 2000s-era web forums. It also exists on Reddit, a sort of super-forum with thousands of discussion boards organized by interest and moderated by volunteers. What these have in common is that flesh-and-blood writers, editors, or moderators toil every day to root out bad actors and clean up information pollution.
The catch is that many of these safe harbors of good-faith knowledge don’t come with search features that work well enough. Thankfully, there’s a work-around, which may be the best remaining feature from Google’s heyday of being excellent at search: the “site:” qualifier. This uses Google Search to look within a specific website for information you need. For example, say you need to figure out your city website’s accessibility obligations under Title II of the Americans with Disabilities Act. It was a wrestle to find this information through Google with “title II web accessibility” as a search term. But search instead for “title II web accessibility site:ada.gov” — and you’re off to the races.
Finally, there is one of the singular philanthropic projects in the history of the Internet, Wikipedia. As a nonprofit institution independent of advertisers or even governments, Wikipedia massively crowd-sources collective human scrutiny to safeguard information quality. The upshot is that Wikipedia’s search feature is quite good! While it is not perfect, I believe that the search for general knowledge should often start not in any Internet search engine, but instead on Wikipedia.
Navigating the mud
It is not possible to be a successful analyst without efficiently finding accurate information about the world. Over the last generation, the infrastructure of the corporate internet declined and made this basic task much harder. By experimenting with different search engines and how they embed into your day-to-day, and by accessing reservoirs of high-quality knowledge with search engine tools, there remains an opportunity to find the information you need to deliver excellence for your residents.
When swimming in an ocean of mud, make sure you know how to navigate.