Friday January 30, 2026; 9:01 AM EST
Yesterday I wrote the idea "small pieces loosely joined" was central to what we mean by the web. #- Now I'd like to add another criteria. "All parts are replaceable." I think it's self-evident what it means. And of course there is no such thing, but the internet itself comes very close to this ideal. #
- Somewhere there has to be a naming authority that can turn a string of characters like "scripting.com" into a physical address that a machine can understand, like: "16.15.217.109." In all likelihood, the machine your browser gets the answer from is replaceable, and maybe even the machine it gets the information from, but at the end of the chain of machines that cache the result, is the authority for the .com TLD. That authority should do as little as it possibly can. For .com, the authority is Verisign, and actually that server doesn't return the address of scripting.com, it returns the address of the authority for that domain and for scripting.com that is hover.com, where I have registered the domain. #
- The point -- the internet was designed to be very decentralized, and it does as perfect a job at that as is technically possible, but even the internet isn't totally decentralized. And some of the systems that claim they are part of the web are far from minimally centralized. If you were to knock out x.com for example, that would also knock out every every account on x.com. But that's okay because they don't claim to be decentralized. But Bluesky does make that claim, and they have some ability to serve other functions for other apps, but for using Bluesky itself, it is no less centralized than x.com is. So we really do have to tighten up the discourse on these systems and stop repeating the claims that companies make about their systems. #
- This is something that imho should be discussed at the next fediverse conference. Do we really want decentralized, and if so we should be clear on what systems are and aren't decentralized. #