Recharge in the sun

Soon, nobody’s going to read my reviews.

A machine will.

When someone needs a plumber, they won’t scroll through star ratings and make their own mind up. They’ll ask an AI, and it’ll read the lot… every review, every mention, every scrap of data about me… and hand them an answer.

Here’s what keeps nagging at me. That AI has no way of knowing which of those reviews came from a real job. It can’t tell the customer I actually worked for from a paragraph someone generated in ten seconds.

We spent years worrying about fake reviews fooling people. The bigger problem is fake reviews training the thing that now decides who gets the work.

A human might smell something off. The machine just counts.

Unless the data itself can prove where it came from, we’re about to automate the trust problem instead of fixing it.

We already can’t tell whether a review is genuine. Most people just don’t think about it.
Now imagine what happens when generative AI makes it effortless to produce thousands of convincing, detailed fake reviews. Different writing styles. Specific details about jobs that never happened. Posted from accounts that look real.
The review systems we rely on were built for an era where faking it at least took effort. That era is ending.
And nobody seems to be talking about what comes next. We’re still trusting platforms to sort the real from the fake…. the same platforms whose business model depends on volume, not accuracy.
If the information itself can’t prove where it came from, we’ve got a problem that’s about to get significantly worse.

Derek's right. And this is a harder truth than it looks on the surface.
"Make awesome apps" sounds like a UX problem. Better design, smoother onboarding, match what the centralised platforms offer. All true, but I think the real barrier is deeper than polish.
The question underneath all of this is: who is freedom technology actually for?
If the answer is "for the people who already understand it," then we've built a very sophisticated club with excellent principles and few members. If the answer is "for everyone," then we have to build the bridge. And bridges, by definition, stand in both worlds. They have to touch the messy, imperfect ground that normal people walk on while still connecting to something better on the other side.
That means custodial key management for people who will never back up an nsec. Centralised databases where they make the product work. Invisible infrastructure that does its job without asking the user to care about relays, event kinds, or signing flows. Compromises that make protocol purists uncomfortable…. but that are the only way any of this reaches the people who need it most.
The best technology disappears into the background. You don't think about TCP/IP when you load a webpage. You don't think about TLS when you enter your card details. Nostr should be the same. The protocol works precisely when you never see it.
Legacy social had 20 years of polish, yes. But they also had something else - they never asked users to understand the infrastructure. Nobody signed up for Facebook because they were excited about PHP and MySQL. They signed up because their friends were there and it was easy.
We have better principles. We have better architecture. We even have AI to close the development gap. But I think we also need to let go of the idea that users should eventually "get" the protocol. They shouldn't have to. And if we build it right, they won't.
The cypherpunks wrote code. The next generation needs to ship products that people love to use…and never once ask them to care about what's underneath.