35 comments

[ 3.0 ms ] story [ 89.9 ms ] thread
> However, let’s be real: the biggest thing holding self-hosted apps back is they’re still too difficult to install.

And run reliably. And secure. And admin/maintain.

sandstorm https://sandstorm.io/ 's dream is to make this as easy as installing a phone app. It's still rough around the edges but I think the model is completely right and the security it offers is excellent.
Nextcloud AIO has been great so far in my experience. Master container manages of sub containers. Borg backups included. Easy certs. Impressed with what they’ve done on that offering.
> A common critique is that web apps yield clunky user experiences. This will be debated endlessly in developer circles, but native feeling UI is not necessary for mass adoption; the bar is “good enough”.

Classic - love when engineers settle for "good enough." A good reminder that the average person has bad taste and takes very little pride in the end-user-UX.

The frontend is arguably the one part of selfhosting that is solved. Most people can use web interfaces. Everything else, not so much. Stray even an inch off the technical happy path and it's rabbit holes and foot guns as far as the eye can see.

CasaOS seems to be getting closer to concealing some of it (essentially an OS that is a giant app store for all the usual selfhosting suspects).

YunoHost is another solution to that. I've been using it for over a year now and it's really nice
This sort of hand-waves the core issue we've always had, which is that it's too error-prone and annoying to set up self-hosting. That, and that alone, is why it's still only done by enthusiasts, no matter how much technologists would like it to be otherwise.

Until it's as easy (and cheap!) to self-host as it is to sign up for an account, self-hosting will remain a niche.

For the last 30 years, there've consistently been companies looking to break into home computing appliances. With the maturity of container deployments, the ubiquity of low-power/high-performance SoC ARM platforms, the increasing consumer concern over data privacy, and the ongoing spread of fiber, the cake is just about fully baked.

Whether it's a big existing company or some startup, somebody will break open the market with a little puck that you plug into the wall and that comes preconfigured with something NextCloud-like and self-maintaining. Then the competition starts, a few generations of innovation happen, and voila.

That puck still costs money, where many existing services are free

It won't be directly accessible over the internet without either router configuration changes, or some kind of cloud service to phone home to

If it were addressable over the internet- would the customer have to set up a custom domain? Or memorize an IP address? Or rely on a cloud service to redirect requests through?

And then on top of all that, it'll go offline if the customer's home power or internet goes out

There's some more gap to be closed, sure, and you already don't have to be a programmer to set that kind of thing up. But I don't see this kind of thing ever traversing beyond the realm of enthusiasts

Services that are free invade your privacy, manipulate you with targeted ads, or use addiction to get you to spend money elsewhere.
Yep, and even with today's increased awareness, most people are fine with that tradeoff
I’d argue most are ignorant of it, or don’t understand, or are simply too busy with other things to care about knowing details of their apps.
When even on HN people think clearing your cookies makes your tracking consent meaningless, then I'd even say we are far, far away from people understanding what they give away.
You are talking about one month's cellular or cable bill.

It will take effort to set up and run one of these - it's always easier to get corporations to do everything for you, but is that what you want out of life?

Anyone could set up a small business supplying these with domain included, and even provide remote admin for a fee. Some people would like that, some would not. Hover will sell you a domain in five minutes.

You put the puck and your router on an UPS, like any security system. If you have a power hit longer than an hour, does it really matter if your website is offline? How often does that happen? Maybe solar power it, just for giggles.

Everything you're saying is still from the perspective of an enthusiast/idealist; try to step out of the bubble. Most people don't think about their technology that way.

They don't care that it follows abstract ideals, or that it sticks it to Big Tech. They probably don't want to spend an extra five dollars, or an extra five minutes, on those things. They want technology that makes their life tangibly better, that lets them do things they couldn't otherwise, and that just works. And most people won't understand the subtle difference between "self-hosted with remote-admin" and "not self-hosted" anyway.

I'm not judging them. I want simple most of the time too. I make some different choices here and there, but I often pick the easy solution when it comes to consumer tech, because I have other things in my life I'd rather spend my energy on.

I'm not sure self-hosting can ever get there, but if it does, it's still got a steep road to climb first.

Yeah, it’s like growing your own food or fixing your own car. Most people can’t, and don’t care.

Remember Heinlein:

A human being should be able to change a diaper, plan an invasion, butcher a hog, conn a ship, design a building, write a sonnet, balance accounts, build a wall, set a bone, comfort the dying, take orders, give orders, cooperate, act alone, solve equations, analyze a new problem, pitch manure, program a computer, cook a tasty meal, fight efficiently, die gallantly.

Specialization is for insects.

Most people could, but don't want to

(and again, I don't think this is inherently a bad thing; we have finite energy to spend on finite things, and we must prioritize)

Eben Moglen and the FreedomBox Foundation have been pursuing this since 2010.[0][1]

Moglen recognized very early the loss of freedom and agency involved in the stampede to the cloud. FreedomBox is entirely free and open source software.[3]

You can buy the puck[2] or make one yourself.[4] The erratic changes in residential IP addresses are handled by dynamic DNS. Like all private email servers, running a residential email server is difficult - the best approach is to use a service like Fastmail to relay for you.

Running your own Mastodon server or web site is straightforward.

[0] https://freedomboxfoundation.org/

[1] https://softwarefreedom.org/events/2010/isoc-ny/FreedomInThe...

[2] https://freedombox.org/

[3] https://freedombox.org/about/

[4] https://freedombox.org/download/

This already exists in the home NAS server market, and the killer app for that still seems to be Plex.
I had an managed webspace for half my life. It was a hassle. Going to self-hosted was the best thing.

- No more expiring certs (and I don't have to pay a premium for them)

- No more Apache

- I can decide which versions of PHP, Python or ehatever I run

- I can write simple bash scripts that email me in certain situations

- I can install tools that I need and debug problems so much more easily

This was it for me, I accept that for other people that equation might look fundamentally different.

I'd say its far easier now than say 20yrs ago and also way cheaper tbh, I'm excited for the next decade because I think there's still innovating cheaper energy, cheaper storage, easier integrations, etc.
While I agree that self-hosted PWA will have _some_ success (especially for easier use cases) and that cloud services are worse for the user than a local app in use cases suitable for the latter… it seems like a big load of wishful thinking to believe that PWA would replace cloud services in any significant way:

The truth is that we already had and have local apps that worked fine for the user (that is: native local apps, including cross-platform) and that despite that, cloud services gained terrain on them enormously. The main reason being that cloud services offer enormous benefits to the selling/operating company: the revenue is more calculable and less prone to the piracy issues of local software. The proportion of users who have enough awareness to recognize the problems of cloud services are a minority.

And the only thing that self-hosted PWAs really add compared to the classic native local apps, compared to which they have a lot of disadvantages, is that their development has a somewhat lower learning curve compared to cross-platform native software development (though the latter isn't that hard either and the reasons why many native apps were not cross-platform have nothing to do with technical problems in most cases).

As for the way the article pits "communities" against "big tech": I see that more as a matter of open source and open data communities… for which local PWAs really aren't that much of a big game changer as the article imagines.

I run some of these but I don’t think it will take off in the mainstream unless we can get the process of setup and administration out of the 1970s.

It needs to be like installing apps on a phone with seamless auto update. Running Proxmox or Docker or even one of those self hosting oriented distributions is too complex for most people to bother with. As soon as you say “install this distribution on your server” you have lost most people.

It’s not just lack of technical skill. Even most people who are perfectly able just don’t have time to hassle with it.

Yep. Shouldn't be any more difficult or less secure than running an app on your phone. I'm working on this with my main side project. Curious to see if there's actually a market for it, providing software to make it seamless and selling tunneling bandwidth.
You've probably already seen it, but if not, checkout https://sandstorm.io/ It didn't do too well commercially, but might have been ahead of it's time
I looked at it and it mandated login with a big cloud provider like Google, which almost completely defeats the purpose by giving them access.
It sounds like you looked at a very early prototype circa 2014? Sandstorm has supported login via email, LDAP, and SAML for a long time.
Thank you so much for sandstorm btw! It's a really incredible piece of software
Sandstorm is awesome, but it's nowhere near as simple as installing an app on your phone to use it. Selfhosting something like NextCloud shouldn't be more complicated than selecting it from an app store and going through a quick OAuth flow to tunnel traffic from a domain name.
Completely agreed - setting up Sandstorm itself is a bit of a hassle. Once it's installed, additional apps on the platform are extremely easy though.

Thanks for pointing out nextcloud! Hadn't seen that one before.

What's your project btw?

The real killer for Sandstorm is that each app has to be modified to work with it, and they tend to get out of date. It just never gained critical mass as a platform.

But it had many advantages in terms of security and usability. I think if we can ever truly get selfhosting for the masses off the ground, something like Sandstorm will eventually thrive.

> What's your project btw?

It's not public yet. I'm working on the client software side and partnering with PrivateRouter.com to build out the network. We're shooting for a beta release in the next couple months.

Selling tunneling bandwidth won’t be a good model. People will just use things like ZeroTier and Tailscale and go P2P.
The market I'm targeting (which may or may not exist) is non-technical. They can be expected to conceptually understand things like data ownership, e2e encryption, privacy, etc, but setting up ZeroTier would be beyond them. Again, the phone app model. Go to the app store, select NextCloud, install. It's up and running and secure by default. No CLI, no port forwarding, no DNS management, no IP addresses. Just install and use it like any other app, except it's running on your hardware and (preferably) a domain you control.
> It needs to be like installing apps on a phone with seamless auto update.

The problem is, even if that might be ideal for the masses, it is in direct opposition to the values of many self-hosters today (like me).

Ask HN about Ubuntu's snap packages, and you'll know what I mean.

pikapods is the closest thing ive seen to that so far. the server is managed by someone else so you just have to sign up and choose what apps you want, unlike yunohost or cloudron where you need to setup up your own server first

https://www.pikapods.com