> People don’t want to run their own servers, and never will.
Not really much related to web3/crypto topics, but I think this is an indictment of servers, not people. If managing a server were easy and secure, lots of people would do it - for blogs, a minecraft server for the kids, to back up their pictures, and yes, to store their bitcoins or other digital secrets - they just don't want to manage a unix or windows server.
It used to be hard to install a webcam, now it isn't. No reason server software can't do the same thing - all we need is for some gigantic corporation to sink 100k developer-hours into it (which sounds like a joke, until you remember that there are several gigantic corporations who have very profitable side-hustles hosting servers, and who would be creating a whole new class of customer if they did this).
What is the benefit to the average user of running their own server? Most people (maybe even on HN) just want things to work. We buy connectivity services for our phones and our homes. I certainly don't want to run my own Wireless ISP to connect up my neighbourhood even if it was marginally cheaper (until I account for my time).
We buy storage services (for lots of reasons) from Amazon, Google, <your favorite backup provider>, etc. I don't want to run a large NAS and keep it running and backed up.
We buy messaging services (voice, SMS, email, IM etc). I don't want to run my own Asterisk VOIP PBX, my own OpenBTS node, my own postfix instance, my own IRC server.
I buy power services (electricity and oil). I don't want to run my own oil well, refinery, nuclear power plant etc. I do actually run some solar panels, but the amount of cognitive load that they cost me is very small. It is probably under 3 hours per year of having to fiddle with them.
In short, the cost in terms of time and energy from me makes it far cheaper to outsource all of these services to someone else. This doesn't prevent you from running any/all of these services, but I would suggest that you are in a very small minority.
Having said all of that, if I lived on an island with no services, I might be tempted to run some of them myself.
> What is the benefit to the average user of running their own server?
All the server-side use cases you can't do with a client alone. I think you misunderstood my comment; I'm not saying that running your own email server is easy, nor that it's hard but still worth it; I'm saying that the fact that it's too hard to be worth doing is a statement about the software that exists today, not some sort of immutable feature of the universe.
Anyway, that's the wrong question. The right one is: what new software would we make if everyone had their own server? The answer is, I dunno, but the hardware is good enough to find out; a cheap virtual server costs about as much as a streaming service, and quite a bit less than a mobile plan. It's well within reach for everyone in America to have their own VPS running their own email server. They don't, because Gmail is way easier, but that would cease to be true if we had better software. And, once there were a few server-side apps that were actually good, we'd probably make more (just as the advent of smartphones led to a lot of new use cases that would've been difficult to imagine before they were commonplace).
> not some sort of immutable feature of the universe
Except it is. Running your own server will always be more work than letting someone else do it, so unless there is a strong incentive people will let someone else run their server.
This is basically the Law of Leaky Abstractions. At some point you will have to deal with a problem yourself because no abstraction is perfect.
Why is there a graphical installer for the Minecraft client and not for the Minecraft server? Because of some fancy Law with Capital Letters, or because more work went in to the former than the latter?
The value to the average user is the possibility of self-hosting under unusual circumstances. It's like insurance. In a walled garden, when you get canceled there's nothing you can do. In an open Internet when you get canceled you can self-host. 99% of people will never need it but the option is valuable.
People will run all sorts of things they don't directly interface with if the setup and functionality is low friction. People run routers for example. If you had to SSH into your router and troubleshoot it just to figure out why you're not getting connectivity people wouldn't do it. Unplug it for 5 seconds and plug it back in? Still frustrating, but the UX has low friction.
If you can buy a little square box that you plug into the wall and it Just Works™ people would do it. People used to leave their home PCs running all day to allow them to perform server type functions.
When I build a home server, I generally shoot for low maintenance, but I do the setup myself. If I can do it once, I can do it once for a million people. Sane defaults, low friction UX, just the needed functionality, everything starts on boot and resets on reboot to a working state is all it really takes.
I agree, except for the part about this being physical hardware rather than a cloud thing. I find it very difficult to imagine even 10% of America buying something like this; a likelier model is, imagine that your typical $50/mo wireless plan included a $5/mo virtual server, and an app to manage it that looks like an easier version of cpanel.
If even 10% of america bought something like this the product is a wild success, and it would be enough pressure to ensure silos are unable to totally wall themselves off. I'm sure apple would love to drop support for SMS, but they can't, because some large percentage of Americans don't use apple devices, and so those that do wouldn't tolerate being unable to message their friends without apple devices.
I know people that aren't tech savvy at all that would buy a box they just plug in, boot up, that for example synced their contacts, pictures, ran a social media server just for them (mastodon maybe) and an email server and IM server and all they had to do was run an app on their phone and enter a password. You could build something like that and offer it to people for under 100 bucks. People don't run those though, because it's not as simple as that. Most people would rather have a product than a service. But the product is less profitable than the service, so companies build services, and so people use services.
Yeah, I think the success of Synology's NASes speaks to this - they're largely used as little home servers. And it could be even easier if someone built a box that functioned as a router and a server with dynamic DNS as an easy part of the setup. The UI would have to be really, really polished, but I think it could be done.
Symmetric home ISP connections would make these more useful, too. Sadly, that's not the norm right now, but perhaps that's because most people don't demand it.
> If managing a server were easy and secure, lots of people would do it - for blogs, a minecraft server for the kids, to back up their pictures
Easy 1-click deploy exists right now. Lots of VPS providers offer service specific deploy for things like minecraft, seedboxes, plex, nextcloud, etc. Check out Scaleway's InstantApps section to get an idea. https://www.scaleway.com/en/imagehub/
Yes, this is the sort of thing I'm talking about, but I think it also demonstrates that there's a ways to go yet. I suspect that one part of the solution will be abandoning linux and windows. They were built for performance and versatility, which just aren't that important in a server I use to host my blog and email, and come with a lot of baggage that isn't needed in a virtual-only OS.
Don't want to bother with OS updates? CoreOS exists. And containers abstract away the fiddly parts of running services. Firecracker/Ignite exists. I feel like all the parts are there to build what you want. Except there's no market for it. Even if it were free and easy to run your own Nextcloud instance, most users would never switch away from Google/Apple/Microsoft.
Techie users can roll their own servers. Power users can buy a NAS with 1-click service installs. Normal users don't even want alternative services.
I don't think that's true, people bitch about those services all the time. Everyone that uses FB has a gripe about it. Windows users have been complaining about Windows non-stop for thirty years.
I'm not arguing that people will switch off of those to crappy self-hosted replacements out of sheer spite against megacorps, I'm arguing that they will switch when self-hosted replacements are better and easier to use. Building a self-hosted platform that does what Facebook does more easily and conveniently than Facebook is hard, but IMHO it's easier than building AWS or Salesforce, and it gets easier every year, due to bandwidth and cloud hardware getting cheaper and big tech getting more user-hostile.
while all true; its also true no one wants to run their own servers.
the problems are practical. power and heat. noise. theft or disaster => backups; 3-2-1. updates, botnets, firewalls, static/external ips. ssh, vpn, or port forwarding. vlans? scaling? trust?
each of these things is a rabbit hole of problems and issues to solve.
I'm sure some people would want to run one in their basement, but I was referring to virtual cloud servers, and assumed the last sentence would've made that clear.
I'd agree that running a server is easy.
I built one for my company once. It ran on a standard PC (early 2000's).
But for some reason it never worked from my house. (Rogers ISP in Canada)
It took me hours and hours to find out my ISP didn't allow me to run a server. BellCanada ISP did, so the server worked on my co-workers system perfectly.
To be allowed to run the server I would have had to pay an extra 200$ a month, and be classed a business address.
It's not just that the tech is hard, the ISP's don't want it. So they gate keep.
Not really much related to web3/crypto topics, but I think this is an indictment of servers, not people. If managing a server were easy and secure, lots of people would do it - for blogs, a minecraft server for the kids, to back up their pictures, and yes, to store their bitcoins or other digital secrets - they just don't want to manage a unix or windows server.
It used to be hard to install a webcam, now it isn't. No reason server software can't do the same thing - all we need is for some gigantic corporation to sink 100k developer-hours into it (which sounds like a joke, until you remember that there are several gigantic corporations who have very profitable side-hustles hosting servers, and who would be creating a whole new class of customer if they did this).