> As you say, we've had the fediverse for over a decade now. How is that going?
Convenience trumps everything. All the parts of the iPhone existed for a few years before it -- especially PDAs with touch pens -- but what made the iPhone succeed was putting everything into convenient and easier package.
The amount of time worked on thing X has almost zero correlation with its adoption, as I think all of us the techies know.
> Even if we could get everyone to use decentralized services today, the internet infrastructure is not ready for it. Most ISPs still offer asymmetrical connections, and residential networks simply aren't built for serving data.
While that is true, let's not forget half-solutions like TeamViewer's relay servers, Tailscale / ZeroTier coordinators, and many others. They are not a 100% solution but then again nothing is nowadays; we have to start somewhere. I agree that many ISPs would be very unhappy with a truly decentralized architecture but the market will make them fall in line. I have no sympathy for some local businessmen who figured they will run with tens of millions with $50K investment. Nope, they'll have to invest more or be left out.
So there would be a market reshuffling and I'm very OK with it.
---
But how do we start off the entire process? I'd beet on automated negotiation between nodes + making sure those nodes are installed on much more machines. I envision a Linux kernel module that transparently keeps connections to a small but important subset of this future decentralized network and the rest becomes just API calls that would be almost as simple as the current ones (barring some more retry logic because f.ex. "we couldn't find the peer in one full minute"). I believe many devs would be able to handle it.
Convenience trumps everything. All the parts of the iPhone existed for a few years before it -- especially PDAs with touch pens -- but what made the iPhone succeed was putting everything into convenient and easier package.
The amount of time worked on thing X has almost zero correlation with its adoption, as I think all of us the techies know.
> Even if we could get everyone to use decentralized services today, the internet infrastructure is not ready for it. Most ISPs still offer asymmetrical connections, and residential networks simply aren't built for serving data.
While that is true, let's not forget half-solutions like TeamViewer's relay servers, Tailscale / ZeroTier coordinators, and many others. They are not a 100% solution but then again nothing is nowadays; we have to start somewhere. I agree that many ISPs would be very unhappy with a truly decentralized architecture but the market will make them fall in line. I have no sympathy for some local businessmen who figured they will run with tens of millions with $50K investment. Nope, they'll have to invest more or be left out.
So there would be a market reshuffling and I'm very OK with it.
---
But how do we start off the entire process? I'd beet on automated negotiation between nodes + making sure those nodes are installed on much more machines. I envision a Linux kernel module that transparently keeps connections to a small but important subset of this future decentralized network and the rest becomes just API calls that would be almost as simple as the current ones (barring some more retry logic because f.ex. "we couldn't find the peer in one full minute"). I believe many devs would be able to handle it.