The Slow Decline of Google Search

Google has long been the dominant search engine, but in recent years the quality of its search results has noticeably declined. What was once a source of highly relevant information has now become cluttered with low-quality results, adverts, and AI-generated answers of dubious accuracy. Even worse, Google search is now often being surpassed by LLMs in terms of useful answers.

There was a time when Google prided itself on delivering the best possible search experience, with a focus on surfacing high-quality web pages that directly answered user queries. But those days seem to be over. Increasingly, a Google search returns mostly results from a small number of mega-sites like Reddit and Pinterest, rather than the broader web. Relevant results from smaller independent blogs and websites are getting drowned out and buried. The “indie web” is withering away in Google results.

What’s more, Google appears to have stopped indexing source code from GitHub. In the past, you could search for code snippets and find relevant examples from GitHub projects. Now, those code results are missing, replaced by SEO-optimised junk. Granted, GitHub now has vastly improved code search functionality for logged-in users, diminishing the need for Google in this area. But it’s still a loss of a once valuable resource in Google search.

As if that weren’t bad enough, Google has recently started using its own AI systems to directly generate answers to some queries, similar to what Bing is doing with OpenAI’s technology. The problem is, these AI answers are frequently wrong or misleading. Google seems to be rushing to implement AI without proper safeguards.

Between the dominance of mega-sites, the disappearance of code results, the rise of AI-generated spam, and Google’s own flawed AI answers, the search experience has drastically degraded. Users are forced to wade through more and more junk to find truly reliable information. The tight relevance that was once the hallmark of Google is fading away.

Ironically, we seem to be better off asking an LLM directly rather than doing a Google search. Tools like ChatGPT and Anthropic’s Claude can synthesise information from across the web and provide direct, useful answers and relevant code examples, without all the SEO cruft. While not perfect, the trajectory of AI models points to them becoming better than traditional search.

Some argue that Google remains dominant simply due to inertia and lack of strong competition, not because they are still the best. Upstarts like Kagi and You.com show promise, but have an uphill battle against Google’s entrenched position.

Still, if the quality decline continues, more users will start seeking alternatives. Google may be headed down the same path as former tech giants like Nokia and Blackberry - overtaken not because they lacked resources, but because they grew complacent and lost touch with what users wanted. The future is looking more and more like it belongs to AI, not traditional search. And that may not be a bad thing. A shakeup of the search market is long overdue, and perhaps AI will be the force that finally dethrones the Google hegemony.