Like most open-source projects it is in perpetual development. Yes, some parts are still not done and rough around the edges, but it is already a playable game. If you wait for it to be “finished” you might be in for a disappointment 😅 0 A.D. for example is still in “alpha” 20 years later…

KDE is surprisingly lightweight these days. Maybe try the Fedora KDE spin. Or if you need it even more light weight: the LXQt spin is also not bad.

On Fedora it got super easy as well. Debian is just by design not very friendly to closed-source drivers.

The economic model just doesn’t work in a winner takes all market. Maybe try https://veloren.net as a open-source multiplayer RPG (not really massive, but there are typically 100+ players on the main server at any given time).

Games like that which prey on vulnerable people with gambling addictions should be highly regulated and probably not exist.

No, following users is not supported by Lemmy. You could try kbin though which supports that.

Basically it just works except if you use a bleeding edge distro like Arch where the newest Linux kernels are sometimes still incompatible with the binary only Nvidia driver. Overall the experience is a bit better with an AMD GPU and the open-source drivers though.

Small update: Registrations are currently disabled and I am working on migrating this Lemmy instance to my server. So far it looks good, but I need to rewrite all the container deploy scripts to work with Podman, so it will take a few more days.

I am also still interested in kbin, but the project is undergoing massive rework right now (with a NLnet grant) and will probably not really be ready for easy self-hosting until later this year.

Uhm, yeah better stick with x86 then. There are more power efficient x86 chips though as a Ryzen is probably overkill for that.

Due to technical limitations only the very latest generation of ARM SBCs have more than 4GB ram.

For most server use 4GB is actually good enough as you don’t need to run a heavy desktop and browser etc.

IMHO much more important is good software / main-line Linux kernel support, otherwise ARM SBCs are just too frustrating.

SATA ports are indeed rare with ARM SBCs, but those with PCI-e ports and m2 slots exists. Probably the best supported board with PCI-e for now is: https://www.pine64.org/rockpro64/

This sounds quite nice, but I suspect it will initially be really expensive like similar ARM work-stations.

Yeah, no thanks. I’ll continue not buying their showelware then (or only at deep discounts years later).

Link?

Get an old refurbish hp thin-client or a thinkpad then. Something sub 200$ with at least 4 threads and upgradable to at least 8gb ram and sata ssd storage. Laptop has the advantage of built in battery (UPS) and screen/keyboard in case you mess something up.

An old laptop.

Synapse will struggle on an ARM SBC and forget about hosting email on a residential internet connection. It will be either blocked by the ISP or filtered directly to spam by most email providers.

A Mindustry Let’s Play would be also nice: https://github.com/Anuken/Mindustry

Hmm. It used to work at least. Maybe this was changed?

If you mean that a thread claims to have more comments than visible… yeah that seems to be a common issue. Ever since 0.17.x there have been a lot of UI bugs. I think if there are edits to federated comments those edits are counted as new comments in the over-view, but so far this bug hasn’t been fixed @nutomic@lemmy.ml

https://riscv.org/blog/2023/03/top-ten-fallacies-about-risc-v/

From what I heard they did remove some especially bad accounts, so I don’t think this is quite true. But yes their plan is to not moderate much, but rather filter what people can see.

A spambot is not going to use the official iOS app.

Reads a bit like a Blue Sky ad 😅

What I think this article is overlooking and in general what the idea behind content filtering in Blue Sky seems to ignore is that while there is certainly a subjective gradient in objectionable social media messages, its not something you randomly happen to come across.

Much rather it is either pushed into your face by some engagement optimizing algorithm, or intentionally spread by persons with certain agendas.

If it was just communities talking among themselves and having different opinions on stuff you would never even get to see them unless you intentionally join into a conversation of a community with diverging opinions.

Or to put it differently: it is perfectly feasible to run a fediverse instance with no block-list at all, as what you follow determines what your instance sees. A block-list only becomes necessary because people purposefully invade other communities to spread their hate and harass people.

And this problem is not really addressed by the Blue Sky content filtering at all. Much rather it seems to be just a way for the user to express some preference on how much random abuse the algorithm is going to shovel into their face on any given day.

Probably a conspiracy theory, but given the push-back to this, I wouldn’t be too surprised if the spam attack is being done by a disgruntled person that wants to force other instances to defederate from mastodon.social.

Just curious, when you say accounts would need to be recreated does that mean existing accounts will be deleted or just cleared and re-set ?

I would not attempt to convert the user-database at all. Most accounts are probably dormant (or spam accounts from before), and for GDPR reasons I would rather not like to take over a lot of user data like email addresses and passwords. Better to delete those and start fresh.

I wonder though what happens if someone tries to address an previously existing account or community on the same url via federation…

We probably also need to have a transition period during which previous users can reserve their user-names so that no impersonation can take place.

Online IDEs with tight CI/CD integration are definitely nice to lower the entry-barrier and get people contributing faster. The FOSS story there is a bit spotty, but interesting projects do exist.

As for “social coding”… that idea has been around for a long time, and well working FOSS solutions also exist, but all experience so far has shown that this is not actually very helpful or productive. I think these solutions in combination with an audio chat can be helpful for introducing new remote colleagues to a complex code-base, but otherwise they offer little value.

IMHO you can’t fund a service with meagre normal banner advertisement revenue anymore.

Someone wishing to fund a Fediverse service would have to write a deep data-mining system that displays personalized and targeted advertisement to their users and get sufficient investment to survive until they have a large enough user-base and scale for their data-mining to turn a profit.

Not impossible, I guess, but given the invasive nature of said data-mining they would probably be defederated quite quickly (if found out) as in a federated network you can’t cleanly separate whom’s data gets mined.

This is such a one sided telling of the story.

The dataset in question AFAIK does not including any of the artists work, but only web-links to publicly accessible works and they warned the artist that if he was to proceed with this via legal means that costs for lawyers would occur.

Maybe in this specific case it was just a badly informed person that thought they were doing the right thing, but in general such copyright trolling is a real problem and IMHO effected parties are completely correct in asking for legal fees to be covered by the person making fraudulent copyright claims.

I wonder how it compares to https://akkoma.dev/FoundKeyGang/FoundKey

I think the hype is driven by people that just want Twitter without Elon and realized the Fediverse is not that. I know that by saying so I somewhat sound like the people that the article is criticizing, but I think people that want Twitter without Elon are missing a big part of the picture, i.e that Twitter was and is bleeding money fast, so “their” Twitter was going to die one way or the other.

To build a sustainable platform you need to invest in it. People in the Fediverse have done so, but are painfully aware that it is a careful balance and that it can’t work with millions of Twitter users switching over expecting a gratis platform with no strings attached.

And this failure to understand these basic dynamics will probably drive them into the hands of yet another venture capital funded fly-trap and the circle will begin anew.

What about that is “anarchistic”? Its just plain old dodging taxes and wage-theft by what you describe.

Basically any IRC client supports connecting to multiple servers simultaneously, so joining channels on multiple servers was never an issue. Also originally the “network” in IRC implied open federation just like you are describing, but over spam and moderation issues it evolved into a allow-list federation and ultimately incompatible s2s protocols. I sometimes wish people on the Fediverse would learn a bit more about the history of federated systems like IRC to avoid falling into the same traps 😅

As for your hidden comment number: there is currently a bug in Lemmy that shows message edits as new comments in the UI.

You paint a very rosy picture of the Freenode situation. As a result, many people moved to Discord (and to a lesser extend Matrix) and the significantly smaller libera.chat is still waaay to centralized as if people didn’t learn anything from this disaster.

Also in the case of Freenode/libera.chat basically all the admins also switched, meaning little institutional knowledge was lost. This is mostly because the person who took Freenode over was indeed such a nut-case. In a typical corporate takeover the staff is (at least for a while) retained, meaning they can’t just set up shop in a different place easily.

On the one hand this looks really well done, but on the other hand: who wants to play computer games in what looks like a paintball stage? Too much realism starts to look boring and mundane at some point ;)

Not a bad article, but it glances over the fact that there are different types of colonialism, and “settler colonialism” is clearly something that happened in eastern Europe, although mostly pre-soviet times. During soviet times this mostly shifted to central asia and siberia and was at least initially driven by forced relocations of eastern european minorities.

This is untrue. In the early days most open-source software was written by hobbyists. The Linux Kernel was literally started by then student Linus Torwalds as a hobby.

I would even say that to this day most of the relevant FOSS software is either written by hobbyists or as a side project by some people employed in larger corps. Notable exception being Red Hat developed stuff. Sure there are also a lot of other pretend to be open-source software written by corps, but when you try to actually run it, it becomes quickly apparent that their intent is not to be actually used by anyone other than the corp itself and paying customers.

Some of them can be given a second life: https://slrpnk.net/post/280323

And so the enshitification continues in lock-step. surprised pikachu face

Obviously there are worse examples, but how is this anything but “more of the same”?

Very likely they are using the text and voice data to train a AI or offer that data as training material for other companies working on AI. Their recent change in privacy policy was pretty clear that this is happening and they extended it to voice data now.