The Challenges of Self-Hosting Adoption

As I’ve been responding to various discussions in this forum, I’ve been pondering the challenges associated with promoting self-hosted services. I thought it might be insightful to share experiences and hear about others’ struggles and successes in the realm of self-hosting.

I’ve been an advocate for self-hosting for quite some time, exploring various FOSS technologies. My journey has included everything from establishing shared storage spaces for collaboration among freedom-minded individuals, to implementing services like XMPP, Matrix Synapse, Jitsi, Retroshare, and deploying various tools on my VPS and FreedomBox. These technologies offer incredible potential for free and collaborative human interaction. Yet, the challenge I consistently face is mainstream user adoption. Despite my enthusiast group’s best efforts, the solutions we have set up often remain underutilized and eventually fall into disuse.

The resistance typically comes in forms like “It’s too complicated,” “I don’t like the interface,” or “I’m already settled with [a popular cloud service or messaging app].” While these may be valid concerns from an end-user perspective, I believe these technologies address many user needs. However, our initiatives to encourage adoption haven’t been as successful as hoped. Perhaps this is a marketing issue on my part. I even tried organizing weekly presentations to explain the benefits and usage of FOSS technologies, but interest waned quickly. It is clear not everyone shares an enthusiasm for computing, preferring the path of least resistance.

I personally view this as a missed opportunity. In an age where computers are an integral part of daily life, a basic understanding seems essential. But that’s a topic for another discussion.

I’m curious about your experiences and approaches. Have you found success in encouraging the adoption of self-hosted services? What strategies have worked for you?

I want to make an important distinction here: the act of self-hosting is not for everyone, and that’s perfectly okay. My focus in this discussion is not on encouraging everyone to become a self-hosting enthusiast or expert. Rather, I’m addressing the challenge of mainstream users adopting services that are self-hosted by others. It’s about bridging the gap between the technical world of self-hosting and the everyday user who may benefit from these services without needing to delve into the complexities of hosting them.


i hear you… lately, ive been telling my friends that i would support setting up and maintaining their fbx home servers if they wished; especially emphasizing nas, email and web hosting.

its barely caught attention… hardly anyone is interested in the privacy aspect (even though everyone complains about being spied upon). another one of the reasons being that email requires static ip and thats a paid service (comes to around 30usd p.a. in my country). then, the cost of setting up and maintaing a home server vs using paid third party services doesnt prove its worth to them, they just stick to what theyre used to, so that hasnt brought me anywhere either.

im at a point that struggling to rationalize self hosting isnt getting me anywhere. people are more used to following where others passions lie (like owning an apple product). foss has to popularize and become a passion if its going to spread (or there has to be a disaster where people are personally, directly and genuinely effected by the outcome). in todays consumerist world, people have to accept it as a concept or trend to follow it. right now, i dont have belief that that will ever happen. ive come to the point of using foss and self hosting just because i believe in it…


I am a member of

They are promoting free software and provide online services running with free software and with a focus on privacy. The services provided run on their own physical machines using 100% free software, including firmware and bios. They are actually recommending people not to use online services, but anyone using the services they provide is invited to learn about how it works and get directly in touch (XMPP, physical meetings) with other users and those who manage the services and the infrastructure, so that users are somehow in control. They help people using more free software, even if not 100% free. This is the best alternative to self-hosting I heard about, and it is good to educate people. They used to be member of the CHATONS collective but decided to leave as that collective decided not to require using exclusively free software.

About self-hosting: provides a VPN service with fixed IPv4 and IPv6 prefix and no filtering, for 6 euros/month. I don’t know whether they are allowed to provide services to someone who is not resident in France.


Those who use free software vastly underestimate the addiction to convenience of those who don’t. When I’ve tried to entice people into using some of the FreedomBox apps, they’ve usually just laughed. To my face. And these are my friends and family. So my short answer is no, I haven’t been very successful.

Until people come to recognize the threat to freedom posed by pervasive surveillance, adoption of self-hosted services may best be furthered by striving to provide timely, effective solutions to specific group-oriented use cases. A freedom and privacy advocate in the group may be able to influence the choice of a self-hosted home for the group rather than a “free” social media platform tied into the surveillance economy. However, this can work only if a credible and practical self-hosted solution exists.

The convenience of a consistent UI is very nice, but it’s not a feature of the current free software ecosystem, which is decentralized and largely depends on the volunteer contributions of individuals and small teams. In contrast, the surveillance monopolists, with their centralized control and deep pockets, have achieved a higher degree of convenience through a relentless effort to eliminate any friction that would impede user data from flowing to them to be monetized.

Convenience is great, it’s just not the highest value. As much as most people dislike the creepiness of targeted advertising, they don’t (yet) perceive pervasive surveillance as a threat to their freedom. Given the relative inconvenience of free software, which is less a technical issue than a cultural one, I don’t see mainstream users willingly adopting self-hosted services based on free software as long as convenience remains at the top of their hierarchy of values in selecting which services to use.

That’s not to say that FreedomBox and self-hosting in general are doomed to irrelevance. Although I’m not sure FreedomBox can achieve the original vision of a home appliance as easy to use as, say, a toaster, it might achieve parity with consumer-grade routers, whose admin functions, to the extent they’re ever used after initial setup by an ISP technician, are likely exercised by a tech-savvy friend or relative rather than by the actual owner. Consideration of who will be self-hosting and how is relevant because the adoption of self-hosted services by mainstream users depends on the availability of self-hosted services.

Aside from convenience, the principal barriers to adopting self-hosted services are inertia, which I would describe as a sort of natural laziness (i.e., “an object at rest remains at rest”), and switching costs, the price paid in effort, loss of data, and loss of contact with others, involved in moving from one service to another. The surveillance-based social media platforms maximize switching costs to keep their users captive.

Mainstream users might use self-hosted services to participate in groups where their motivation to participate exceeds their inertia and any switching costs involved. I’m thinking of book groups, neighborhood associations, schools, fan clubs, sports clubs, and similar groups. Particularly if there’s a member of the group who can advocate for a self-hosted solution, perhaps even providing the self-hosting themselves, then the group may avoid sliding by default into setting up on a major social media platform. If a self-hosted solution can establish a beachhead, users will, perhaps grudgingly, begin to use it. They’ll then be in a position to discover other self-hosted services.

The success of this approach depends in part on chance and opportunity. It was unfortunate there was no well-established self-hosted alternative to Zoom at the start of the Covid pandemic. The key is that a solution, or at least an easily followed recipe for a solution, exist when an opportunity to establish a self-hosted community presents itself.

That brings me back to concerns around supporting self-hosting, but with a focus on driving user adoption. In addition to continuing to improve the stability and reliability of FreedomBox as a platform for applications (e.g., backup/restore), attention would need to be given to use cases, testimonials, and tutorials. Use cases to identify and fill gaps in functionality, particularly gaps that affect multiple use cases (e.g., a Discourse-like app, a mailing list manager); testimonials because the example of someone who has successfully done what you’re considering doing is a powerful motivator; tutorials on both self-hosting setup and use of services, perhaps aligned to use cases, because services may need to be configured differently depending on the use case, and because the UI quirks of various services that might otherwise exhaust the patience of the convenience addicted can be mitigated by a demonstration of how to overcome, work around, or endure them. The attention to use cases could extend beyond written recipes to follow to include creation of templates and scripts that could be used to simplify setup for specific use cases.


At the beginning of the Covid pandemic, I was the one to setup Zoom for a group of 5-6 people to replace regular physical meetings as this is what I heard of at the time. Unrelated to this, with the extra time at home, I read the FSF and GNU websites for hours, became an associate member of the FSF and wanted to try free software solutions. I tried many too difficult things but I found out that Jitsi is easy to install and works well on a small VPS.

As the initial Zoom subscription was going to an end one year later, I told that we would use Jitsi from now on. It works with nearly any browser (excluding internet explorer) on desktop and mobile phone and even have very good functional mobile phone clients for android and iOS. Most difficulties were to allow the microphone and camera sometimes, then it happened that someone was more or less temporarily with audio only,

At least one complained that they liked Zoom better but we have continued using Jitsi, we never resumed physical meetings and I haven’t heard any complain for a while. I only realize it now but at last meeting, after we were suddenly all kicked out of the meeting, everyone reconnected very quickly and discussions resumed as if nothing had happened, so no friction at all.

I run Jitsi on Debian on a VPS. We had more issues when we were once 12 instead of 5, perhaps the load on the VPS was too much but I am not sure. I now use a VPS with Gandi, which has a nice option that you can change the resources on an hourly basis, so I increase CPU/memory resources before meeting start and reduce them to the minimum after. Not sure it is necessary though. My desktop is currently not working but when it was working, I also ran Jitsi on it (Trisquel installed) and we had a few meetings hosted on it without any problem (and to my surprise, CPU and memory usage were really low).

At the same time, this small group is focused on using Gmail and Google Drive and I had no success suggesting something else there, but from the beginning I was not the one handling this. So they don’t care about freedom or privacy, but if I can take care of something for them and make it work reasonably well, they can adopt it.


I aso think that backup/restore is a key feature. It already has very nice things but it is unfortunate that there is no option to backup the data shared with samba, that makes usage as a NAS a bit risky, unless you make backups via a computer that accesses the shared data.

Besides, sftp already works on freedombox, but it would be nice to have an option in plinth to allow only sftp to a user and to a particular location, e.g. to the samba shares it has write access to. That could be used for remote access and for remote backup of another freedombox.

After configuring this several times for a user on a Unix-like system (it is the same for any) and re-doing the same mistakes (permission mistakes), I wrote down my own recipe for it, but that doesn’t go via the user management system of the freedombox.

@Avron: It sounds like you were successful in doing what I hoped was possible. Congratulations!

I tried moving an in-person group to Jitsi at the start of the pandemic but failed. This wasn’t on a Jitsi instance I was self-hosting, but rather the instance available to associate members of FSF. A couple of people were having trouble with it and ran out of patience before we could work through their issues, so we ended up on Zoom.

I have no VPS experience, but given that ISPs often make it difficult or impossible for residential users to host mail services, a VPS may be a good choice when email is important. The cost of a domain and basic VPS with Gandi seems to be around $100/year (a bit less the first year), which is likely not out of reach for most groups, even if the freedom advocate in the group has to foot the cost initially to get set up. Other forum threads on VPS refer to a wiki page about FreedomBox in the public cloud, which notes that the cloud may be less secure and private than one’s own hardware, but unless I missed something I don’t see a path from the wiki homepage to the public cloud page. That page says “It is recommended to use a cloud instance for trial purposes only”, so maybe that’s intentional.

A different worry is that needed functionality will not be packaged for Debian or will drop away because upstream projects are tending to move to distributing containers or rolling releases without backported security patches. That challenge to the Debian model of security and stability and its implications for FreedomBox deserves its own, separate discussion.

1 Like

One thing that works: hosting someone else’s stuff on your server, then gradually teaching them to administer their bits of the server, intervening only when they need help. This acts as an on-ramp to selfhosting. Non-technical users gradually learn and take over more of the responsibility.

It also removes the major difference between an SaS web offering and selfhosting: the need to set up hardware.

Unfortunately, using Freedombox currently makes sharing servers much harder. Hosting multiple domains on one Freedombox can be done, but not in the GUI; it gets kludgy. Having multiple email domains can be done, but through quite extreme kludges. I’m not sure if anyone has tried running multiple Ikiwikis, say. If you are going to on-ramp others, it seems like it is currently easier not to use Freedombox.

Long version follows

I find, and some of my friends have found, that if you run a server, you get asked by friends, family, the local chess club, etc., to host something for them. So you wind up with several virtual hosts, on several different domains, supporting misc activities, ranging from a flat static website telling you when and where the chess club has met for the past century, to a private forum for a support group.

Unfortunately, setting up name-based virtual hosts on FreedomBox is a pain. It gets you messing with config files in a way which can break your entire install. Security certs for multiple domains also require manual messing about.

I know someone will suggest simply setting up multiple Freedomboxes, but that would mean a severalfold increase in power consumption and other upkeep. Ceasing to support less-tech-savvy others is also not really a good option.

Would it be possible to add virtual host support to the Plinth GUI? Just the ability to add multiple domains, and manage certs for them, would be really useful.

Setting users to belong to one domain or another would also be nice, and reduce the trust level needed. Giving users e-mail and XPMPP addresses according to their domain(s) would be great. A superb feature would be if the backup/restore was per-domain, and you could restore each domain onto the same or seperate servers. This would make it easy to fission Freedomboxes, or merge them, depending on the social needs of the users and admin(s), with no visible effect on the frontend.

There are quite a few posts on the discussion forums asking for support related to virtual hosts.

Getting an SSL cert for multiple virtual domains involves hacking around, as the GUI isn’t set up for it. Can I assume that it auto-updates okay?

Kludging multiple e-mail domains is possible but complex:

1 Like

One highly-motivated group of potential adopters: people running micro companies that started out selling on Etsy or Amazon, who now want to be independent and cut out the extremely expensive middleman. These people are organized enough to fundraise and co-ordinate with one another, but they don’t have much money, they have no IT staff, and they need very low overheads. They are often setting up their own servers anyway, and looking for the best way to do it. They are articulate and specific about what they want.

A basic federating self-hosted storefront package would be an excellent addition to Freedombox, and it seems more likely to earn recurring donations than many packages.

If anyone knows of a good option already out there, I know some people who want to know…