How the internet lost its curiosity
There’s been a sea change in how we interface with the internet - we've devolved from interest-based browsing to being spoon-fed content.
My first flashbulb memory was discovering a giant PVC longbow that could propel a Nerf dart 120 feet.1
I was sprawled out on an unmistakably suburban carpet, laptop on the floor; this must’ve been in sixth grade (the heyday of Nerf wars). The Windows Aero circle had been spinning for a good ten seconds, when the crappy UI that characterized every early forum finally loaded, and nerfhaven.com came into view.
I had just clicked a link on the Outback Nerf blog while trying to learn more about modding Nerf guns. I wanted my Nerf Raider, an unreliable peashooter, to become the foam equivalent of Master Chief’s assault rifle. It took me to NerfHaven, the most elitist Nerf forum at the time. There was a section on the home page titled Modifications, but before I could click it my attention was distracted by the hottest post on the whole forum, “The Snap-7 Microbow.”
A Nerf bow? That was way too cool not to click.
And it was then that I realized just how magical the world of internet forums and blogs was.
The golden age
In the early 2010s, there were a wealth of websites around the world dedicated to this funny foam dart hobby; the first one I discovered was Outback Nerf, a blogspot site created by an Australian bloke who set out to review every Nerf gun he could get his hands on. That was my gateway drug to the Nerf internet, a surprisingly dense web of blogs and forums discussing tournaments, gun mods, and which Nerf guns to buy. I learned everything there was to know about stock gun ranges, dart-per-second firerates, and which springs worked with which guns. I even acquired a substantial amount of niche knowledge - for example, how Australia had very strict toy gun laws that affected the performance of certain Nerf guns, especially the fully automatic ones like the Swarmfire.
Life was great.
That was the golden age of the internet. Curiosity was the best way to find things, and with enough interest you'd stumble across whole communities like this, a patchwork collection of people and sites all branching off their love for the same thing. It was an authenticity and unpretentiousness that is hard to find now.2
The gilded age
Contrast that to today, the age of the algorithmic feed.
We are in the age of instant gratification for zero effort. For most people now, their entryway to the web is not Google. Instead, they open TikTok and start scrolling.
There’s a huge difference between those two models of interaction. When you start from Google, browsing the web is a very active affair; you have to type something into the search bar, and then make a conscious choice of which site to visit. When you do, you’re usually using the “will this site show me something cool” algorithm.
Not so much for the feed. The feed plates, serves, and forcefeeds you content as soon as you scroll. The expectation now is that some powerful algorithm will know exactly what to give you to stimulate your dopamine production. There is no curiosity driving your engagement with Instagram or TikTok - no one gets off these apps after two hours of scrolling and says, “wow, I learned something cool.” Interaction with the internet has become purposeless, something done by force of habit rather than any conscious decision.
In that sense, the smartphone is the cigarette of devices - if you aren't paying attention, you instinctively reach for a puff of that Reels algorithm. Instead of using nicotine to hijack your dopamine response, social media hijacks your brain with algorithmic power.
All hail our new reality
I could rant about the feed for a lot longer, but the New Yorker does a better job than I ever could.
The Internet actually limits attention, in the sense of a deep aesthetic experience that changes the person who is engaging. The business model of digital advertising incentivizes only brief, shallow interactions—the gaze of a consumer primed to absorb a logo or brand name and not much else. Our feeds are designed to prod the would-be attender ever onward from one monetizable object to the next.
-Kyle Chayka, The New Yorker
They highlight my biggest issue with the feed model of content delivery - how there’s no opportunity to dive deeper if I see something cool.3
On a forum or Wikipedia, a cool topic can be explored by clicking the next link or going to the next page of discussion ad infinitum, but that’s just not possible on the feed. We all know that repetition and active engagement are the most important factors in retention and learning; neither are present in the feed, probably by design. The comments are never substantial. The next post is completely unrelated. There is no History subsection, no Technology subsection, no chance to explore and meander and hit pause on the content firehose, and so right after you see something cool, it gets erased from your short term memory by the very next thing the algorithm brings up.
Is the future of the internet more hopeful?
Okay, enough cynicism for one post.
I think a lot of people have realized how unfortunate the state of the internet is. Quite a few of my friends use their phones in black and white. Some have even gone phone free for a few days. Being conscious of the way feeds hijack your brain definitely helps a lot.
I’ve recently been trying to replace and replicate the feeling of discovery, like stumbling across a fun website. Unfortunately, Google is genuinely bad at discovery these days (I think this is becoming a common sentiment), so since the Google results are just SEO spam and LLM-written content, we’re back at square one.
The only good way to find cool content is the old fashioned way, turning one blue link purple at a time.
And that is kind of what I've been doing - digging up blogs and Substacks and reading Substacks recommend by other Substacks4. And I have to say, I enjoy it. Substack is pretty great - people are just writing whatever, but the nature of the platform requires effort. You might get 40 pages of insanity or delusion, but there's a beauty to the madness that can't be replicated by Facebook posts or Linkedin cringe.
Aside from Substack, there's also the sending articles to friends thing - some of the content I've enjoyed the most in the past year has been word-of-mouth recommendations (shoutout to Nathan for recommending the Acquired podcast).
A bit of speculation
So what's next for the internet?
We're currently in the middle of a generative AI boom, and it might forecast a better era for the internet. As a mental exercise, we can consider two timelines of AI evolution.
The first case is that this is it; we will never find a way past the limitations of generative AI. In that case, we would simply have a group of LLM services that do a very good approximate summary of the internet. Essentially, Wikipedia, but instead of curated, it's generated on the spot. Instead of maintainers doing the hard work of keeping a website factual and accurate, it's algorithms, mathematicians, and engineers building and training complex models to resynthesize information with as little hallucination as possible.
This is basically status quo, but with a workaround for how bad Google is now. ChatGPT and Claude Opus are already infinitely better at answering how-to questions than the garbage that Google surfaces.
It’s not hard to envision a world in which they replace a lot of other Google functions, like looking for reviews or product recommendations.
The second case is the extremely optimistic case: generative AI takes over everything, and we start to be able to generate anything. Games, movies, etc. would just be a few natural language sentences away from existing.
I forget where I heard it - maybe a NYTimes event, or an Acquired podcast, or maybe GTC - but Jensen Huang (of course) spoke about how the future of computing was going to move away from retrieval-based computing to retrieval and generation-based. We’re already seeing some of that - Qualcomm and Microsoft are buying in with their NPU and “Windows AI PC” certifications, and we’ll see what Apple has to say this WWDC as well, but it’s not hard to imagine a future where half of our interaction with computers is asking them to create, not retrieve, something.
This would be a different kind of internet - some kind of active-passive hybrid, because I doubt feeds will ever go away, but it would be an internet where creativity is unleashed. We could live in a world where "all you need is an idea" to create and release something of high quality and exceptional production value, and where your entire interaction with the internet looks more like a two-way street, where you request content to be made for you. Because there’d be no need to make content for everyone, we might be able to skip the shallow fluff that gets a passing like from everyone like that one egg that got upvoted on Instagram and go back to an age of genuinely cool shit.
Imagine being on Claude Epic (or whatever it’s called five years from now) and it’s like a personal history professor, answering every question you have about how America turned the tide of WWII in the Pacific, and it even creates a whole animation that showcases the Battle of Midway. Wouldn’t that be incredible?
Thanks to Eric and Mehul for reading early versions of this piece.
For those unfamiliar with foam dart blasters, 30 feet was considered a decent range. Modded blasters that your delinquent neighborhood engineering child might have could do 60, maybe 70 feet. 100+ feet was (almost) as crazy as Michael Phelps winning a bajillion gold medals.
Google was a bit more useful back then, too.
Like that time when I learned way too much about the difference between inkjet and laser printers and started talking about it to like five people at a party three weeks later. Not sure this is a point in Wikipedia’s favor, but I did retain that information for three weeks so I could remember it at the most (in)opportune time.
I hear RSS is making a comeback.