I think largely we are aligned on what we are looking for in a platform. The private blog idea is interesting. I normally consider blogs as public, are there private blog platforms?

Sure. If nothing else, you could proxy it through an authenticated endpoint, requiring people to log on to view it. But I don’t know the blogging software space very well - there are probably projects with built-in support for this. I’ve started looking around; I suspect the ideal platform isn’t so much a blogging platform, but it’s designed more around a blog design.

If you come across one, please let me know! I’ll keep updating that CryptoPad document. I also started a spreadsheet, which is better suited to the data than a document table, but CryptoPad doesn’t have the ability to embed assets from other documents (other than images), so I’m just doing the table manually.

On the other hand, projects die when the maintainers lose interest.

Absolutely. Good projects attract multiple maintainers; there’s a bit of Darwinism there. When one project I used was archived, I offered to take over maintainership; the author didn’t want to hand it over to me, so I hard forked it and worked with distributions to replace the no-longer-maintained version with mine. It’s the OSS lifecycle, right? And the best thing about OSS - if the maintained loses interest, someone else can simply take over. And if no-one does, maybe it isn’t worth maintaining.

I would like a platform that I know is going to stick around.

This is so important! Especially for this purpose. Getting several people to join a platform and then put content on it introduces a lot of technical inertia. That’s why it’s important for me to reduce the odds of the project changing their terms of use; increasing costs; moving popular, free features to the “paid” column; and other shenanigans.

On the other hand, something like Zusam, if the maintainer loses interest it will likely also die.

See, I don’t believe this. It’s possible the project would die, but so often have popular projects lost their maintainers, and new people step in. They fork it, or have a peaceful transition of ownership, but the project carries on. Yes, some just disappear into obscurity, but the popular ones tend to keep going, sometimes under other names. X11 to XOrg; OpenOffice to LibreOffice; OwnCloud to NextCloud; so on and so forth. And increasingly, many projects add data migration paths from other projects, especially if they’re popular. Many ActivityPub servers can import Mastodon account data, for instance.

I do have reservations about HumHub, but it’s the first platform I’ve seen that even comes close to being a familiar feel for users.

It does look pretty close to ideal for what we’ve been discussing; I need to install it and try it out, because so far all other options have failed in some way. There’s another forest of options in the blogging style, so I’m still optimistic, but I may try HumHub anyway.

I’m considering the other idea of using Dokuwiki as well, which I guess comes in as being more similar to your blogging idea.

Yeah, that was an interesting avenue; I suspect the user client experience will be where that fails for me. It can’t require any technical expertise.

Apples and oranges.

Wireguard is a VPN technology. DuckDNS is a service that lets you create a subdomain on the duckdns TLD and point it at your server. They do completely different things.

You would use DuckDNS if you don’t want to rent your own domain (“rent” because it’s a recurring payment for something over which you have only nominal control). It provides no security, no access control, and it creates no network. It’s just a pointer in the global DNS DB.

Wireguard is a VPN technology, for creating private networks.

One is like a mailing address. The other is like a strongbox. You could give the strongbox to a friend to deliver it to someone who has the key (Wireguard). Or you could write a message on a postcard and mail it (DuckDNS). Or you could put the address on the strongbox and mail it (DuckDNS + Wireguard). The point is, they serve completely different functions.

The two could be used together.

I agree with you on how core emoji reactions are. … It’s clear I’m going to have to settle in some respect.

So, in thinking about this in more concrete terms (as opposed to vague dissatisfaction), I suspect what we really want is a blogging platform with robust authenticated reader interaction tools.

The issue with AP, and therefore most of these servers, is that (a) it’s expected to be public by default (the privacy point you mention), but almost more fundamentally (b) they’re aggregators. People either to a bunch of people and get a feed of a bunch of posts by different people (Mastodon/X); or they join a community and see a bunch of posts by different people (Lemmy/Reddit).

I think what we want is blogging software, with an endless stream of content posted by a single user, but with reactions and threaded conversations per post. I’ve been thinking how this could be achieved on various AP platforms, but while you can almost get there with groups/channels/communities, the sticking point is that they are all ultimately designed around any member being able to post top-level content. I haven’t seen any system yet that (easily) allows restricting posting by individual accounts.

I need to look at pump.io clients, because I think pump.io started as more of a blogging protocol. And the more I think about it, the more I believe a private blogo is a better foundational model.

Is federation or similar mandatory for you?

No. In fact, I suspect it may work against the privacy requirement. I expect that, even if one of the federated servers met all of the requirements, federation would have to be disabled to prevent leakage. Although, at least one server supports authenticated pull (one of the Misskey forks), I’m guessing it’s not likely that federation will be needed.

As in, do you want something that allows your users to interact with users that are not part of your family and not on your platform, eventually able to completely replace the mainstream social media?

For me, no. I want my SIL to be able to easily post pictures and videos of my toddler niece, and all the family members to be able to oooh and aaaah, and react with little heart and exploding brain emojis, and comment on how the fact that she climbed a jungle gym is a sign she’s sure to be an Olympic athlete. The parents absolutely do not want those videos showing up in TikTok.

Or is a completely closed platform ok, in terms of it’s only your family and friends, and people have to go elsewhere (e.g. back to facebook) to interact with others?

Ideally, it’d support ActivityPub. I’m not sure how; perhaps through the user creating channels and setting a federation flag, or marking it as public. I think the expectation that people will understand that inviting someone from another platform effectively makes all of that content public, might be bit much to assume. So I think having private and public channels, where public channels are federate-able would be fine. But I’d rather not have federation than have a system where people are prone to make privacy mistakes. Is there an option I’m missing?

I use Nextcloud, developed by a company,

Yeessss; I think that’s a little different, because NextCloud was forked off of the completely open source OwnCloud, which was well-established and license protected long before NextCloud came along. If NextCloud tried any shenanigans, they’d be eviscerated. HumHub is a bespoke solution, right? So they can’t be accused of stealing an OpenSource project’s s code.

I use Photoprism, which the base edition is FOSS but they have proprietary extras that you pay for (like HumHub).

Yeah, this is a good example. I use it, too, although I admit I’ve considered, and regularly revisit, alternatives purely because of this quasi-free nature. So much of PhotoPrism is built on free libraries; the project uses something like 120 OSS libraries. How much of their income do you think they contribute to those projects who’s work their taking advantage of?

I use Home Assistant, though I think they recently transitioned to a non-profit

I’ve been using it for two or three years myself; it’s always been OSS & free software, AFAIK.

they charge for a cloud connected component.

That’s a service. I have no issue with charging for a service, because it’s an ongoing cost to the hoster.

Actually, I don’t have any issue with anyone charging for their software, either; it’s just that I won’t use it, and I don’t trust quasi-free projects. That’s just from experience. Most end badly, either by being bought out and going totally commercial, or just slow enshittification for the non-paying customers.

I write software for myself, and give it away free because it costs me nothing to do so. And I’ve written software libraries that I know, for a fact, are being used as backbone code for a not insignificant chunk of the internet. I’ve never been paid by any commercial company taking advantage of my work, and have little sympathy for people charging for software that’s 90% other people’s freely given code. Which is most software today. You write the entire stack from scratch, including the compiler, like Excel once was? Hell yeah, you deserve to charge for it. Otherwise, you’re just profiting off other people’s work.

HumHub have been around 10 years, so they aren’t exactly new. Plus as it’s extendable, perhaps one day a gfycat or emoji reaction plugin will be added (or if you have the skills, maybe you could make one).

Huh. Never heard of them before a week or so ago. I wouldn’t completely discount them because of the semi-free model; I just am putting them down on the list.

If it’s truly in a sleep mode, and you don’t have Wake-on-LAN enabled, no distro that I’m aware of will wake itself and make noise.

But the belts-and-suspenders solution is to make a cron job that mutes the audio devices in the evening, and unmutes it in the morning. Depending on your cron subsystem and configuration, this will work even if the laptop is asleep at the trigger times; some cron systems guarantee execution of events - systemd is one of them, and is the most likely one you’ll encounter.

But, seriously: if you put Linux to sleep, it stays asleep; you have to work to get it to wake itself up to do things, and it usually requires some external trigger.

Thirded.

They occasionally upgrade services for free, and rarely raise prices. They support a variety of base Linux images, including Arch (which, when I first switched to them, was rare). The control board is functional, and they’ve got all the features needed to implement VPN subnets, DKIM, etc. without having to use the DNS provider’s tools (assuming you are using a different provider). There’s also a command-line tool for managing your VPSes with them. Reasonably priced, the usual array of options from cheap to expensive, easy to add resources, and so on. Servers in the US and Germany (and maybe others? I haven’t added a VPS in a while).

When I first started self-hosting, not all of this was standard. I can’t say I’ve looked at the market in a few years, so perhaps their offerings are standard now, but when I moved from another hosting provider, Contabo stood out. I have been quite happy; perhaps the best thing I can say about them is that I haven’t had to contact their technical support in the past couple of years.

P.S. the only cautionary thing I’ll say it’s that they’re a German company. While you can never trust any VPS provider from a data security POV, Germany is a 5-eyes country, and so sits in my “least trustworthy” list; as in, they’re least likely to put up any resistance if one of the surveillance states asks for access to your data, or to tell you about it before they do. For me, this doesn’t matter, and frankly I don’t have enough knowledge to choose a better option if I needed it. Since I don’t, and since I’m not using my servers for anything that’s currently considered subversive, it isn’t yet a worry for me. But FYI.

Thanks!

Agreed: some items are basic functionality that should reliably and easily work. Image & video uploading are among them. I’ll add some verbiage on the CryptoPad page about options which have been rejected simply because they don’t support the most basic features.

It’s funny: I’ve been similarly searching for a good chat platform, and there are two things which I personally don’t care much about, but which a couple of my family members are insistent about: typing notifications; and gifs - as in, a widget where you can search for short gifs from e.g. Gfycat and have them inserted. My wife absolutely requires the latter.

That being said, my position on emoji responses are almost a core feature for a social media platform IMO. They’re fast, easy, non-cluttering feedback, eliminating the need to type out some inane, two-word response. It’s infuriating (to me) that Lemmy doesn’t support them; it leads to such illuminating responses as “So much this!”, “Yes!”, but worst of all the lack subverts up/downvotes, which should be a tool for designating interest, not agreement. Not having emoji reactions muddies and dilutes any value voting has.

Pixelfed is an interesting suggestion. It always feels like it’s intended to be public. Were you thinking each user would have to configure default privacy settings?

You may be right. I think I read that post visibility was configurable; if I can narrow the field sufficiently I’ll start installing them and checking how they work. I do think federation would have to be disabled on any AP server.

I can see how to restrict to followers but haven’t yet found how to stop anyone being able to follow you.

Yeah, that would be a blocker.

I think for me, if a new user has to set up the privacy settings to stop them posting everything public, that’s probably not the right platform.

Agreed. The service must be at least configurable to be private-by-default.

BTW there is PixelDroid as a dedicated Pixelfed app, but it’s only on Fdroid.

I think I found an iOS app, too… but I looked at so many servers last night I may be misremembering.

The table isn’t rendering on my mobility client, so I’m going to delete it from the post; I’ll keep the CryptoPad document going as long as I can, but it’s open edit, and I’m hoping others will contribute to it.

Argh! I’ve posted a similar question; basically, I want a private alternative to Facebook, with wall-like functionality. The second minimum requirement is that there be an iOS app that makes posting easy – including initiating a picture or video capture. So:

• #1: private, b/c it’s family sharing toddler pictures
• Also #1: super user friendly, because (100% - 1 person) involved are non-technical
• Also #1: has to have a better user tool than an SPA. No web interface can ever be anywhere as good as a native app can be, and I will die on that hill.
• #2: emoji reactions, and threaded comments

I’m not interested in installing and evaluating a dozen different servers, so like you I’ve been hoping that people with similar goals would narrow down the field a bit. There’s no way I’d convince enough of the family to go along with evaluating all of the options anyway, and IME what works fine for me can often fall apart when other people come onboard.

I’d convinced myself that Friendica – venerable, proven, reasonably popular – would fit the bill, especially because the design doesn’t assume public-by-default, like Mastodon or Lemmy, and the potential damage of exposed content, either through my misconfiguring the server, or some upgrade assuming users want everything public by default, is high. I’d prefer a project where the developers assume private-by-default, and invite-first. Lemmy isn’t really right, because we’re following people, not communities; Mastodon has a better model, following users, but then its conversation threading is kind of shit for this purpose, and its reaction feature set is anemic. Circles was perfect, and beloved by the key parent involved, until it first made half of her posts invisible to her (and only to her and her husband), and then locked her out. This doesn’t surprise me much, as Circles is based on Matrix, which frankly has the worst cryptography management I’ve even encountered. But if you’re saying Friendica is that painful to post media on, then it won’t work.

I’m leery of Humhub because of the quasi-commercial nature, and its youth. I’ve had too many experiences with initially semi-commercial platforms shifting, either suddenly or slowly, to increasingly commercial positions – moving features from the “free” to the “paid” column. Vendor lock-in is a real issue with a dozen users.

So if Friendica is out, maybe Pixelfed? It seemed to me to be mostly indistinguishable from Mastodon, but if they have better comment threading, reactions, and I need to re-evaluate the AP clients to see if any would be user-friendly enough for the parents. I’ve used mostly Fedilab, and I’m not sure it’s ideal. For one thing, it doesn’t have support more than basic reactions: you can boost or favorite, but I am – and I think you are probably – looking for something with more variety, like emoji responses, right?

I’m watching the other reactions here, and my post on this topic is here. I may post a summary – there are comparison charts, but they all tend to focus on feature set and fall short on the overall use case. On my thread,

• Misskey was recommended as Facebook-like, and in particular, some of its forks have features the core project is missing. I always got the impression Misskey was a Mastodon-analog, which would make it not a good fit, so I’ve skipped over it. With Friendica out, I’m going to put Misskey back on the “possible” list.
• Diaspora has also been recommended and is near the top of my list.
• Smithereen was recommended, but the sparsity of the documentation – not even a list of features – put it down low on my list.
• Hubzilla has a lot of documentation; it focuses a lot on content management – assets, calendars, document sharing, etc. – which will be fine if “easily post content to a feed” and “follow a user and view a stream of their posts” is a first-class interaction model.
• Pixelfed is still an option. I just need to confirm/refute my “Mastodon, with pictures” perception. If my perception has been skewed by the fact that I’m interacting with Pixelfed through a (mainly) Mastodon app, then maybe it’ll work. However, there isn’t AFAICT a Pixelfed app, so if the only way to get to a more wall-like view is through a web interface, it’s not going to work.

@acockworkorange@mander.xyz is also looking for this feature set / use case. I kind of feel as if it’s more useful to think about this as a use case, because almost all of these projects can claim some or all of the requested features, and yet not satisfy what we’re looking for in terms of user experience. This would be a great opportunity for another tool: a wiki with a list of applications & features, but with a discussion section and focused on winnowing projects by consensus about suitability. Again, lots of software that have the necessary functionality and which could be wrangled to do this, but still fail to be a good tool for the objective.

Edit

Probably not the best place to do this, because I’m the only one who can edit this, but:

I deleted the table, as it wasn’t rendering on some mobile clients. The table was re-created in CryptoPad.

I’ll go find a collaborative, wiki-like document thing with discussions that isn’t G**gle.

Edit 2

The table is now here, as a CryptPad document. In an exercise of trust, it’s open to edits. If vandals wreak too much damage, I’ll restrict access, but that’ll require creating accounts and requesting access, and all that shiz.

That’s not an easy question to answer, since it depends on your use case. Of you’re running a mail server, you need SMTP; if you aren’t, you don’t. There is no one-size-fits-all.

However, I will suggest an approach that can guide you:

• Use the firewall, whatever you have installed, and bock off everything except ssh.
• One by one, expose the ports you need, conservatively.
• If you run web services, reverse proxy everything through a single server, preferablys one that’s only reverse proxying, is running as bare bones as possible, and is as simple as possible.
• Once you get things working, go through and shut down and remove any services that you aren’t exposing or using via 127.0.0.1.
• Once this is done, if you are technically capable, set up a Wireguard VPN with your home computer / laptop (preferable two), make sure the connections survive reboots, and then close and lock the door: firewall-block SSH except from your private VPN connections.

In the end, you may have only 3 ports open: https, SMTP, and IMAP. Assuming you’ve secured the web, smtp, and imap servers, this is about as secure as you’re going to get with a single server.

If you are able to, run each service on it’s own VPS: web server on one, IMAP and SMTP on another, and any web applications on their own servers. Connect them only via your VPN, and only through necessary ports, and close everything else. Shut down ssh between the servers, only allowing ssh connections from your laptop. Personally, I think it’s not too bad to run web apps in podman containers and expose those ports to the proxy server over there VPN, but ideally there’d be one VPS poet app, with servers not being able to talk to each other through the firewall.

TL;DR: secure your network before focusing on shutting down and removing programs. Lock down your firewall. Set up a private VPN, and restrict as much internal traffic to it as possible.

How did you get your invite?

Ah, yes, but that’s just packaging. Did you ever stop to consider how they are harvested and processed to give them that crunch?

Where did you think they came from?

Exactly. Mastodon-ish would be unsuitable as a server for a number of reasons: the loose, but still expected, character limitation; the lack of emoji responses; generally poor threading support; and the overall subscription feed-like model. OTOH, it’s based on a follow-the-user model, which is nice. I’m less familiar with Friendica, but AFAIK that’s also a follow-the-user model.

The issue with Federation is the general expectation that these are public places. You can lock them down, but that’s not what they’re designed for, and in my case, the risk of misconfiguration exposing a bunch of toddler pictures that the parents want to keep private is too high. I think of the server is federated-by-nature, then it must also be paranoid-by-default; I don’t trust share-public-by-default projects to not introduced something in an upgrade that exposes data. At least if the base presumption of the developers is that all information is private by default, the risk is limited to true accidents rather than false assumptions.

ActivityPub is enticing. It’s an exciting spec, and offers many client options. I’m worried only about those base assumptions.

Photoprism’s app space is pretty bad, but there is an completely hacky yet reliable solution for Android:

1. PhotoBackup
2. The PhotoBackup server on your Photoprism server
3. A cron job that runs the photoprism import command every few minutes.

Since I’ve had this set up, it’s worked as well as Google Photos ever did for keeping my phone snaps synced to the server. It’s been more reliable than SyncThing for my data, reacting and syncing faster, and it doesn’t mysteriously periodically just stop running like SyncThing.

I don’t know if PhotoBackup is available for iOS, but if it is, it works a treat.

Honestly, I think Android is fucked for debugging stuff like this. I installed a program on mine and my wife’s phones - different makes & models - and configured them exactly the same, including the app settings in the OS. Mine works perfectly and barely shows up in battery use, near the bottom. Her’s drains her battery even when she’s not using it, regularly running at 50% of total battery consumption.

With Android YMMV is the rule, rather than the exception. There’s just too many variables.

It’s the main way I sync my phone.

I have a different app for photos, but SyncThing on my phone, and on my desktop, and again on one of my home servers, do most of the download and data syncing.

Occasionally I’ll have to manually run SyncThing; I’m not certain that Android is reliably starting it after reboots, but for the most part it just does it’s thing really reliably. There is a lag; it can take a few minutes for changes to sync - it’s not immediate. For me, this isn’t a problem, and I’d rather that than a battery suck, so I haven’t messed with it.

Probably. It depends on the device, because there’s no standard dictating this, but I would guess most routers reserve reserved IPs.

To be fair, it is a very long-winded post. I think it’s not an uncommon use case, though, and so deserved a robust sketch of the desired solution; Farmville and chat are sideshows, and what the people left on Facebook are really there for are the Walls.

There are some excellent apps out there, and by and large they look and work better than commercial apps, IME. So I disagree with the assertion that I have to stay with commercial software.

What I was asking for, in my post, was not which apps have better UX than Facebook, but rather which of the very many OSS, federated (although, not necessary for my use case), self-hosted platforms fit the specific use, and ideally with a straightforward iOS mobile app. Doesn’t have to be pretty; just has to be able to quickly take and post photos to a private channel/community/wall.

Circles really is quite nice in all respects. I think they’re hindered by their choice of backend. I’ve been using Matrix for years, and key management has always been a hot mess. I wouldn’t be surprised if the issues we encountered were related to Matrix’s god-awful and buggy PK negotiation & management process.

It’s true some things are harder to do in the container configuration; it’s easier installed as an OS, especially integrations like Z-Wave, ZigBee, RTSP, Eufy, ESP, and so on. All of these require running other software, and in containers it’s a fair bit of fussing with port and host OS device connections.

I’ve always run it in a container, without issue. It works fine, but I’m comfortable with the command line and LXC. That said, flashing an ESP hardware device and getting it connected to HA running in a container has so far defeated me, because I have to give access to the device in the configuration of the container before I run it, but the device flashing process itself is time limited and expects a process to be waiting on it when it is connected. It’s a chicken/egg problem I haven’t yet figured out which wouldn’t be a problem if I were running the HA OS.

HA isn’t the only software that just works better when it controls the while OS. Kodi is another that encourages users heavily to running it as an OS.

Regardless, it runs fine via

podman pull ghcr.io/home-assistant/home-assistant: latest

and there’s a package in AUR that wraps the container up with a systemd service - it’s as close to a bare package install as you’re likely to get.

What’s a little funny to me is that, despite that I’ve been running HA in a container for the past 4 years, I’m working towards getting a dedicated device and running HA OS on it. If we ever move out of this house, I’m not going to spend weeks going around replacing all of the hardware - smart sockets, lights, garage door opener, security, etc etc - with dumb devices; and for any of that to be worth anything, it’s going to need a controller configured for it, which means, I’m planning on selling the HA server device with the house. For that case, I don’t want anything but HA running on that device, and for that, it’d just be easier and smoother to run HAOS.

My advice is to run HA in a container until you are sure that’s the direction you want to go, but not for so long that it’s going to be a PITA to migrate to a dedicated server. But - hey, just IMHO - plan on running HAOS. If I knew then what I know now, that’s what I would have done.

It looks utterly repulsive, but I kind of love the idea… the sheer convenience.