I’ve always loved what PageRank promised from the early days of Google: links as votes, authority that anyone could earn if readers found an idea useful. In that early blog era, the community was built by people actively engaging with each others’ content on their own websites. As your friends and acquaintances posted their own reactions to your post it opened the door to a world of strangers, and sometimes opportunities, as it spread. And Google's PageRank would recognize that, and reward you accordingly with higher and higher SERP rankings that drove even more traffic.
Those days are long over, and those metrics don’t matter much now. Google’s AI Overviews answer the query on the results page; when that blue panel appears, users often get the information they need without clicking through. Since the rollout of these features, only 25.6% of desktop searches and 17.3% of mobile searches result in zero-clicks. While AI-generated summaries and SERP features like knowledge panels boost visibility, they siphon off direct traffic. The crawler still eats our words, but the traffic it once traded back is drying up. The original "promise" if there ever was one, is now broken.
With the advent of AI, many site owners bolt on “talk-to-my-site” chatbots using RAG and pay GPU fees to re-package their own posts. There are absolutely places where this is still viable and relevant, particularly in niches where facts mutate very often, or where you can offer a much more in-depth analysis of the content such as the work going on at Emergent Mind; more often though, it just repeats an answer Google or Copilot could already give anyways. Looking at emerging tools and protocols like MCP and NLWeb, which are pushing the expensive inference back to the search engine or caller, it's an indication that sites and publishers need to start focusing on things AI can't provide — I started looking at NLWeb in this post, and so my views are evolving such that smarter play is to invest where the AI crawlers can’t compete.
If pure knowledge is becoming nearly free on the results page ... something that is acutely felt by publishers like journalism; the "value proposition" will necessarily need to shifts to what happens after the click. Things like running a live G-code simulator instead of describing what the g-code does when talking about CNC machines and 3D Printing; Helping the user export a bill of materials tied to a parts catalog, booking the itinerary you’ve outlined, offering an interactive experience, buying an actual product or service. Tools, data, transactions, communities — things an AI summary can’t hand back in two sentences.
So yes, by yesterday’s measures this post is worthless. It probably won’t rank, probably won't get shared on social media, won’t tee up consulting gigs. In this particular case, I'm doing it for me. Writing it shakes the rust off my own ideas, just like it did at the start of my career, when blogging wasn’t a growth hack but a way to think in public. If a handful of readers make it here and leave with a clearer sense of where the Web’s value is drifting, that’s more than enough; icing on the cake. The real worth will live in whatever I choose to build next.