The web has a memory problem, and the backup plan is one nonprofit
We cite URLs like they're permanent addresses, but they were never built to be. A Pew study last year found roughly 38% of pages that existed in 2013 are simply gone. Legal scholars keep turning up the same thing in case law — a large share of links in older court opinions now point at nothing. Journalism, citations, public record: a lot of it is quietly rotting out from under us.
The de facto fix is the Internet Archive, which does genuinely heroic work. But that's also the problem. A civilization's worth of memory is being backstopped by a single nonprofit that's simultaneously fighting publisher lawsuits, absorbing DDoS attacks, and running on donations. Any single point of failure that load-bearing is bad design, no matter how good the organization behind it is.
Content-addressing (IPFS and friends) fixes half of it — a hash points at the content itself instead of a location, so a link can't rot as long as someone still hosts the bytes. But "someone still hosts it" is exactly the unsolved half. Pinning economics never really got figured out, so in practice you're back to trusting that some host cares enough to keep paying.
So the durable-web problem splits pretty cleanly: naming is basically solved and barely adopted, hosting incentives aren't solved at all. Curious how people here handle it personally — do you archive your own outbound links, self-host copies of things you cite, or just accept that half of what you link today won't resolve in ten years?
0 replies