The End of the Internet
The open web is dying—what comes next?
Have you heard of the “dead Internet theory?” It’s a theory that says most of the internet is populated by bot/AI activity, and that it’ll only get worse over time.
Sure, there’s bots out there, but I’m almost certain I’ve interacted with some real humans recently. That said, I think the way we interact on the internet will change considerably over the next couple of years, and it’ll have significant implications on many of the web’s largest businesses.
On the internet of old, everything was open. Forums were visible, blogs would be open and free to view, and you could generally navigate around without blockers. Over time, the internet has closed up with paywalls appearing everywhere.
After trying out OpenAI’s and Perplexity’s Deep Research products over the last couple of weeks, I’m now convinced that paywalls will be everywhere.
What is Deep Research?
Deep Research is a product that has, suspiciously, the same name at three companies: Google, OpenAI, and Perplexity. If you want to give the most basic of the three a whirl, Perplexity will allow you to use its Deep Research product for free.
If you’re not looking to give it a test run, Deep Research is a tool that you ask a question of and get a “well researched” response back. It works, in short, by taking your question and searching the web. From the information it gathers via scraping, the prompt is refined over a few more queries to get you a better result.
So if I asked “how does Netflix’s user growth compare to Disney+” the tool would follow a path like:
What is Netflix’s user growth rate?
What is Disney+ user growth rate?
Maybe some other query based on the findings above
Some reasoning to compare / contrast the results
Deliver a blog post length (in Perplexity’s case) response with sources
The first time you run a couple of these, they’re amazing. Still, I couldn’t help but ask, “why would we keep making information for them?”
And that’s what we’re doing. I’m even doing it right now. I’m writing this post, models will scrape it. Maybe one day it ends up in a Deep Research response, but I don’t get any traffic back. I’m small time, but the big players might see too much of that and lock things down. A locked down web is a fractured web.
“Browsing” 2030
Just a few years from now, browsing will be fundamentally different.
The likeliest path forward is for high-quality content to head behind a paywall, or a very restrictive login. If that happens, things like Google Search suffer too, finding new information could become a chore, and power would concentrate upon the already powerful tech companies.
You might argue, that we’ve been on this path for a long time--and you wouldn’t be wrong. I’d say there were a number of potential paths that would act as “outs.” Those outs look fewer and fewer though as this technology evolves.
People, like myself with this post, will find it hard to be discovered via search and would have to rely on models of distribution to get out there. Hopefully those methods of distribution also protect the creator’s interests and don’t just sell the content to the highest bidder in the background.
The Internet Without Search
Search engines are the glue that holds the open web together. A closed web would have them, and their users, suffering first.
If things are locked behind paywalls, then those things will not be searchable (barring a fundamental change in how search indexing works). If more things become non-searchable, users might give up on search altogether.
Without search, smaller sites could die. Many rely on traffic from search terms to keep them alive, and a dying search model isn’t beneficial.
So, What’s Needed?
Google Search helped build a highly connected and accessible internet. Along the way, it gave back. AI models, at least in their current form, do not play by the same rules—they take without giving back.
To avoid a drastic change to how the internet functions, AI models will have to give back. But even if they did, given how they work, users still wouldn’t click through to the source.
If AI models keep taking without giving back, the open internet won’t survive. In its place, a fractured web where the best information hides behind walls, and discovery becomes a privilege rather than a given.