Welcome to a new issue of This Week in Plasma!
This week there were many user interface and performance improvements — some quite consequential. So let’s get right into it!
Do you seriously accept AI-generated code into KDE software? That’s concerning.
Let’s even ignore ethical debates and code quality questions, models used for this are trained on a lot of proprietary code. We do not know whether this constitutes a derivative work. Proprietary projects like Windows obviously do not care about copyright (unless it’s theirs), but libre projects should hold themselves to a higher standard in my opinion, as we really do rely on the code being owned by the people who write patches.
When using the Systemd integration functionality (which is on by default if Systemd is present), programs will no longer fail to launch while there are any environment variables beginning with a digit, as this is something Systemd doesn’t support. (Méven Car, link)
Credit goes to @Christoph_Cullmann I only reviewed and merged 
I hate to be that person … but what does it take for a bug like this to get looked at?
I get that it’s not a priority, because it’s not a major breakage, but it’s annoying and seems like it should be an easy fix. Someone else has chimed in that it’s happening to them too. So do I just change it to Confirmed? Will that move it along?
And what a week it was.
Oops, sorry about that! Fixed.
There’s an interesting parallel with KDE’s “real name” policy. In the past we were very firm about requiring one for contributions.
But… how do we know the name someone picks is a real one? If it’s obviously fake like “uwu I’m a Kawaii Dragon” or “Max Mustermann”, then sure, we’ll know. But otherwise, it’s impossible. At a certain point we realized we were penalizing people for being honest about desiring anonymity and started accepting patches no matter the name.
It’s a similar situation with AI, I think. There’s no way of knowing unless the use is obvious and not hidden — and at that point rejecting it would be be penalizing people for honesty.
Windows can now be selectively excluded from screen recording! This can be invoked from the titlebar context menu, Task Manager context menu, and window rules. (Stanislav Aleksandrov, link)
Oh! No more accidentally showing OBS. Or… accidentally shlwing anything you don’t want to show on OBS. Which, as someone who follows some streamers, seems really useful for them.
That make sense - as a user, as long as the quality remains good, then it’s all good. But is the quantity of AI generated reports and submissions isn’t a concern? It’s one thing with AI users ignoring project policies, it’s another to submit it because you have “the blessings” of the policies.
Indeed our community depends on humans to review MR and LLM can scale patch producing of fluctuating quality, but not really reviewing. And we need human to learn our codebase and technologies to become reviewers in term.
Plus working with LLM simply removes the fun out of programing IMO.
In dolphin, we had a first MR like this, the code wasn’t good, the contributor not accountable and closed the MR bringing less effort than the reviewer. This can’t scale.
I quite like the fedora AI-assisted contributions policy.
For what it’s worth, I think allowing LLM-written controbutions is a huge mistake, personally if this actually goes through I’m going to try to find another DE. I like Plasma, it’s the least bad Linux DE IMO, but not if it jumps on the AI bubble too and starts doing unethical stuff like this
This is well tread ground, we can refer to what other projects have done both philosophically and practically here. I will strongly advocate that KDE (or some subset) should adopt an “AI” contributions policy like Servo’s. There’s a lot of tension that comes with this topic, especially if it lingers unresolved for a long time. I’ve written out a longer response here, if you can take a moment. https://korcenji.neocities.org/Writings/KDE-Take-A-Moment
Folks concerned here will get no strong support for LLMs in general from me. Over the past few years I and others in KDE have rejected probably hundreds of obviously garbage LLM “contributions” in the form of hallucinated bug reports, novel-length feature requests, massive un-actionable code dumps, and so on. It’s a pain in the butt to filter out this garbage. And of course that’s to say nothing of the copyright issues, environmental issues, downward pressure on quality from novices using it recklessly, etc. I would overall call myself an AI skeptic.
And frankly, this was the first example of an obviously and acknowledged LLM-generated code change that actually seemed correct to me. Even then, it was accompanied by a novel-length description that had to be hugely pared down.
It’s probably worth coming up with some kind of policy that makes sense, yeah. I haven’t looked at Fedora’s policy, but I’m sure it would be useful to. The Fedora folks are always reasonable.
Speaking personally I think would not be in favor of a policy that incentivizes people to behave deceptively and isn’t enforceable. This is what happened with the former “real name” policy, and despite no formal agreement to scrap it, people basically started deciding on their own to ignore it because it wasn’t workable.
My concern is that if we take the polemical position of “no LLM or LLM-assisted contribution allowed,” all we’re doing is encouraging people who will use them anyway to be deceptive about it. In cases where they succeed, we’ve incentivized people to behave immorally and gained nothing. In cases where they don’t, it will cause an argument about whether the contribution was really written by a human that isn’t provable and risks exhausting all involved.
Thanks for the write-up at Korcenji . I’ll read through the links there. Of course do note that any policy KDE comes up with here (if we do decide to do that) will be a collective decision, and not my personal viewpoint. And whatever we agree on, I’ll happily abide by it.
As backwards as this might sound, it honestly feels like unknowingly getting LLM-written contributions by being deceived by people so honestly terrible as to be willing to lie about it, is the more ethical position over knowingly and willingly participating. The end result is probably roughly the same from a technical perspective, but you can’t fault someone for being deceived or manipulated.
That’s why strict enforcement of such a policy is important, so that if a person gets caught breaking it they will immediately get punished, without exception. Which should deter most people from breaking the rules.
The AI authored code contribution was pure fun 
anyway
KWin’s zoom effect now saves its current zoom level shortly after you change it, rather than at logout. This prevents a situation where the system is inappropriately zoomed in (or not zoomed in) after a KWin crash or power loss.
This is a bit off topic, but I recently had to disable the zoom shortcuts on a PC I maintain for family because multiple people managed to accidentally zoom in (one was a toddler just pressing random keys while no one was looking), but then no one could figure out how to zoom back out, especially since it persists on reboot, and they just waited for me to visit so they could use their computer again.
I wonder if there is a way to make it so that zero-dot users don’t get stuck, that doesn’t create issues or inconveniences for the people who rely on this as an accessibility feature.
If I may, can someone look at merkuro calendar crashing when adding a task or event? Surely you don’t want me to use Google application or search engine.
Keep up the excellent work, guys!
Can you file a bug report? Anecdotally I have similar experience - seems like there’s something that’s easy to trigger by accident, and I think it might make sense to show a hint while in that state (while allowing people who know what it is to permanently dismiss it).
Made a major improvement to the smoothness of animations throughout Plasma and KWin for people using screens with a refresh rate higher than 60 Hz!
Will this get backported or do I have to wait until the 6.6.0 beta to get this.
A bit irrelevant, but since you brought it up:
I’m one such person that would really like to contribute to KDE, but I’m concerned about my privacy. If it comes down to legal issues, I have no problem divulging my legal name in secret if need be. But in general, I’d rather keep my real name private.
Would I be able to do this?