How I actually use the web in 2026
I spent fifteen years building search engines. Inverted indexes, rankers, relevance tuning - systems whose entire job was to put the right ten links in front of you. I was good at it. I cared about it. I still do.
And at some point in the last year, without ever deciding to, I mostly stopped using one.
I noticed it properly this week. I was about to look something up - I genuinely can’t remember what - and I caught myself mid-action, fingers over the keyboard, about to type into the address bar. Instead I closed the tab, opened a different app, and asked the question there. It wasn’t a search engine. What unsettled me wasn’t that I’d done it. It was that I couldn’t remember the last time I hadn’t.
So I started paying attention.
A few weeks ago a light fitting in our hall started buzzing. The old me would have searched, landed on an electrician’s blog or a forum thread or three, read around, triangulated an answer. The new me asked Claude. I got four crisp paragraphs - the likely culprit was a dimmer switch not rated for LED bulbs, here’s how to confirm it, here’s the fix. I bought the part. It worked.
What strikes me, writing it down, is that I never learned whose knowledge that was. There was no electrician. No forum. No name, no date, no comments section of people with the same fitting arguing about the cause. The answer was simply there - fluent, confident, sourceless - and I acted on it without once stopping to ask whether I should. It happened to be right. I genuinely don’t know what my hit rate is, because I’ve stopped checking.
That’s the shift, in one small story. Not "I use AI now." It’s that the checking has quietly fallen away.
What I still reach for a search engine for
It isn’t nothing. The honest list:
- Navigation. When I want to land on a specific real-world page - a shop’s opening hours, a particular company’s site - rather than understand a concept.
- Buying things. A chat assistant can summarise reviews; it can’t put the item in a basket.
- Today’s news. Anything genuinely recent, where I want primary sources and know the model’s training data won’t have it.
That’s roughly it. Notice what those have in common: they’re the cases where I want a place to go, not an answer. Search has quietly narrowed, for me, into a directory.
What I’ve stopped doing
The other list is longer. Looking up syntax I half-remember. Reading a few blog posts to get the shape of an unfamiliar topic. "How does X work." "What’s the difference between X and Y." The quick fact that used to live in the snippet at the top of the page. Comparing two products on what they do rather than what they cost.
All of it now starts in a chat window. Sometimes the answer is wrong and I have to push back. But the round trip is fast, the reply is in plain language, and - this is the part that’s hard to convey to anyone who hasn’t made the switch - it simply feels like less work.
What it costs
I want to be straight about the price, because there is one.
I read less. A search used to hand me five tabs and I’d skim three; now I get a paragraph and move on. I’ve stopped bumping into things - the unexpected, slightly-too-long piece by someone I’d never heard of, three results down. That kind of accident is mostly gone.
I’ve lost the provenance, too. Five blog posts had five authors, five sets of assumptions, five biases I could feel against each other. One answer has one voice, and it sounds equally reasonable about all of it - the parts it’s sure of and the parts it’s guessing.
Here’s the thing I keep turning over, though. I notice all of this. I can name every loss precisely. And I haven’t switched back - not even a little. The convenience didn’t win a fair fight against the costs. It just quietly stopped putting them on the table.
Why I’m writing this down
It would be comfortable to file this under personal quirk. One man’s habits, mildly interesting, no larger claim.
I don’t think it’s a quirk. I think it’s most of us, and I think it happened faster than almost anyone planned for. A vast amount of the web - how it’s paid for, how things get found, how anyone discovers anything at all - was built on exactly the behaviour I’ve just described myself abandoning. The person who spent fifteen years building that machinery is now, in his own small way, routing around it.
That’s what this series is about. Not whether the change is good or bad - honestly, I don’t know yet - but what it breaks, who it costs, and what comes next. This first post is only me admitting the change is real, and that it reached even me. The rest is about what the change is doing.