We have not solved decentralisation in an accessible and useful way yet, and the incentives won’t change until we do. If ever.
We have not solved decentralisation in an accessible and useful way yet, and the incentives won’t change until we do. If ever.
I, and many like me, would pay for centralised service or any other service if it meant that we own our data and can tune the algorithms to our own preferences. I wont pay for doom scrolling, but would gladly pay for algorithm to serve me content that would better my human experience.
Governments have given corporation to much power, people need to rise up agains that, if it remains the same in AI age, we humans, and our collective mind would erode to the point of no return.
Unequivocally, users water plants that deliver in demand fruit while being most convenient and cheapest.
So, why can't I have that?
During my standard install of my favourite distro, I would only need to enter my name, subdomain and email password for everything to be magically installed, so I have a standard web site, some file sharing and email out of the box.
However, it would take me a fortnight to get this setup and I wouldn't have a clue how the email actually worked, if it worked. This wouldn't be my first rodeo either, so I wouldn't be starting entirely from scratch. I am also sure that there are some that have setup umpteen virtual linux machines that they could get everything done by tea-time.
Whether two hours or two weeks, it is still not that much work in the bigger scheme of things, which makes me wonder, why haven't I got some all-singing and all-dancing bash script that automates the whole process? But why has nobody else done it either, to make it fully open source and as easy to obtain as it can be?
Also, why can't I buy a glorified router box that does all of this? It could take the mainboard and power circuitry from any laptop, and, out the box, provide a decent web server, mail server and whatever else.
There is a suspicious absence of products in this space.
Step 0 is to secure that box, as routers are obvious targets, even before they have self-hosted data. There are some products based on RPi, NAS and router form factors.
> suspicious absence of products in this space
Earlier efforts:
Apache Wave (federated)
Chandler
Diaspora
FreedomBox
Microsoft Groove (p2p)
Urbit.org
Sandstorm.io
Active OSS projects include Proxmox (https://community-scripts.github.io/ProxmoxVE/), Paperless-NGX (docs), Immich (photos), NextCloud and others, https://github.com/awesome-selfhosted/awesome-selfhostedAs for security, it is all a bit meh. If you have a box that only runs https: with no other ports open, you are half the way there. If you are just running static pages then you are done. If you run a NextCloud type of beast then you are opening things up, but my hunch is that it works just fine with nobody losing sleep on it.
There was no choice but to use someone else’s computers for moving around large files. Plus CGNAT and whatnot making people have to use dynamic DNS. If a turnkey solution could have existed 20 years ago, maybe a market for it would have developed before the big companies locked it down.
Even if I’m wildly in favor of user control over data, I’d venture to say that there still is no choice but to use someone else’s computers, and not just for performance reasons. If applications have to gather every individual user’s data that gets shown to another user from somewhere outside their servers every time, won’t reliability and consistency and UX likely become nonexistent, in addition to the unusable performance?
Decentralized does not need to be slow like that. And very limited upload does get to be a problem if you want more than a couple people/servers to be able to access your media posts at the same time.
The person you replied to is assuming a reasonable distributed system.
But self-hosting machines are susceptible to the "I can only upload pictures and videos at 5-10mbps" problem. That requires more difficult peer-to-peer systems.
The first problem only requires getting small bits of data onto the same machine. The second problem requires getting large amounts of data onto many machines. Or reasonably symmetrical upload speeds.
Even today, I doubt I could get anyone to just give me a smartphone.
So the next best thing is trying to operate in the constraints that apply, such as most people being unwilling to learn new things and going down the path of least resistance.
That some people don't want to spend time learning the thing that you happen to find interesting doesn't mean they're wasting their lives.