For a fifth-grade science experiment, Tim tested whether the internet was faster than the library. The library won two out of three times. One search took twenty-seven minutes. This was before Google existed, and yes, Tim is aware of how that makes him sound. Episode 2 is about what happened between then and now — which is that we built a search engine so good it owns 90% of the global market, and somehow the thing it’s worst at is finding stuff.
Is Google Actually Worse, or Just Different?
James drew the distinction that most people feel but can’t quite articulate: in 1998, Google’s job was to get you off of Google as fast as possible. The faster you found what you needed and left, the better it was doing. That’s not what Google is anymore. Now it’s a destination — AI summaries, quick answers, featured snippets, maps, shopping panels. The goal now is to keep you there, which is roughly the opposite of the original goal.
Tim’s take is simpler: there’s just too much stuff. Back when Yahoo had a category directory for the internet, it worked because the automotive section had maybe 400 sites in it. You could sort through that. Type “automotive” into Google now and you’re looking at four billion results, and the top ones are all from Honda because only Honda has the budget to play the SEO game at that level.
Which leads to the Reddit hack — the thing everyone does now and nobody talks about out loud. When you actually want to know what people think about something, you add “Reddit” to the end of your search query. The organic results are sponsored. The Reddit results are anecdotal, ranging from “I love this product” to “this product killed my father,” but at least they’re real opinions from humans. Consumer Reports exists somewhere in the middle, but finding it has become its own project.
What Tim actually misses is lawnmowerdad.com. Some guy with seven lawnmowers and a weird dedication to metrics nobody asked for, like — and this is a direct quote — “ability to hold a can while mowing.” That’s the kind of content the early internet was full of: specific, passionate, and completely unprofitable. Wire Cutter exists now and it’s fine. It’s not the same.
The Part That’s Actually a Tech Problem
The week’s news hook was a React security vulnerability that turned out to be a Next.js vulnerability, which turned out to be the predictable result of taking JavaScript — a language designed to run in a browser where a page reload fixed literally everything — and running it on a server, where a page reload is just a prayer.
Tim put it well: when you had a clear division between front-end and back-end, the front-end developers didn’t need to know about Java threads and the JVM, and the Java developers didn’t know how CSS worked, and that was fine. Both sides had specialized knowledge for their domain. The “full-stack JavaScript” era collapsed that distinction without necessarily filling in the gaps. Someone who went to a bootcamp and learned React can absolutely build something that looks and works great. Whether they know which parts of it are running on the server, what that means for security, or how to debug something that isn’t in the browser — that’s a different question.
James framed it as an abstraction problem. He got to see every layer on the way up: assembly, then C++, then the web, then self-hosted hardware, then the cloud, then containerization. Each layer made sense because he’d seen what was underneath it. Someone starting today skips straight to Next.js and has no particular reason to wonder what’s beneath it until something goes wrong. And then they’re console-logging into a terminal they barely know exists.
“It’s like magic,” James said. “When it works, it’s impressive. When it doesn’t, you hate on the magic.”
They both caught themselves sliding into the “two old men yelling at cloud” register around this point. James confirmed they’d crossed that line several episodes ago and there was no going back.
Would You Rather: The Google Trilogy
Round 1: Pay a monthly subscription for ad-free search — OR deal with ads and search for free?
Both chose free, without much hesitation. James then disclosed that he runs a Pi-hole on his home network, which blocks ads at the DNS level. This means he personally sees very few ads and pays nothing. It also means that when his wife clicks on Google Shopping ads, the links don’t go anywhere — she just sees the product and can’t buy it. This has become a recurring household argument about whether shopping should be allowed, which James characterized as “a little friction isn’t a bad thing, especially when it comes to shopping.” His wife’s characterization was not recorded.
Tim noted that this is basically just Yahoo’s old category directory with extra steps.
Round 2: 2005 Google — no Gmail, no Android, no YouTube, no Maps — OR 2025 Google as it is today?
Both chose 2025, reluctantly. Tim compared trying to de-Google your life to trying to eliminate plastic: it sounds achievable until you discover your prescription bottle is plastic, the ice cream container is plastic, the cereal bag is plastic, and you’ve become miserable at parties explaining your mission. James noted that YouTube is full of videos about how to de-Google your life, and the people who make those videos almost always end up going back to Google. If the people who dedicated a video to leaving can’t leave, that tells you something.
Round 3: One large external dependency in your Node project — OR many small ones?
This one split them. James wants many small packages: Lego blocks, mix and match, and there’s always one that does the specific thing you need. Tim wants one large one with everything included and someone responsible for it: less surface area, less dependency hell, and when things break there’s a person to call. Tim also pointed out that one “minor” package he recently installed silently added nine other packages. James’s response was “only nine?”
The conversation landed somewhere personal: James has a multi-brand desktop setup with Sony headphones and various components that sometimes conflict. Tim has a MacBook and AirPods and everything from one company that works together. Tim switched to Apple from Android after his Bluetooth stopped connecting in his car. He cannot drive in silence. The music stopped, he made a decision, he never looked back.
The Unresolved Issue of Shuffle
Briefly: shuffle doesn’t work. Nobody’s shuffle works. They play the same five songs on rotation and you know it’s not random but you can’t prove it. Tim’s theory is that Audio John works at Spotify and is personally curating everyone’s queue to match their vibe. James’s theory is that humanity mastered shuffle at some point and then lost the knowledge, like Roman concrete.
Both are equally plausible.
The Takeaway
The internet got big, and then it got too big, and now finding the specific thing you want requires knowing the right workaround — add Reddit, use a different browser for recipe sites, build your own DNS filter. Web development went the same direction: incredibly powerful abstractions that let you build faster than ever, on top of a foundation fewer and fewer people understand. These are probably related problems.
The solution, per Tim: reboot the internet. Set a date. Whatever comes back is important. The rest can go.
He was, he clarified, mostly joking.
Tim’s mom does listen to podcasts. She has not yet followed the show. The subscriber count remains a work in progress.