poison implies the existence of poidaughter and poienbykid.

Is there a website that ranks computer go engines (i.e. programs that play the board game go) by their playing strength, similar to the Computer Chess Rating List but for Go?

Search engines keep directing me to either rankings for chess engines, or human go players.

@fdroidorg@floss.social Update:
It appears to be gone from F-Droid ​:blobbee_party:​

"But why do I have to learn math?!?"

So that you can understand posts like the one I am quoting.

RE: https://mk.absturztau.be/notes/as22k3yodr0601dd

Awhile ago, some LLM users who were angry Flathub rejected their "apps" made their own flatpak remote called Flatpark, which hosts "apps" that are (almost?) all fully generated by LLMs.

I wonder if Flatpark will continue operating, or if they will just submit all their LLM "apps" to Flathub now?

Dumbing of Age Book 16 ends with Dorothy becoming a superhero.
https://www.dumbingofage.com/comic/hairdyediversion/

libwebp 1.6.0 (the latest stable version) can do a better job losslessly compressing my avatar than libjxl 0.12.0 (also latest stable version).

Using the settings -m 6 and -lossless, I can get my avatar down to a 78 byte webp file.

On the other hand, I just wrote a script to exhaustively try every encoder setting with cjxl. The best lossless size for my avatar is 79 bytes, achieved using cjxl with -e 10 (the "buffering" option of cjxl doesn't make a difference for such a small image).

Though I guess the resulting files are actually the same size on my disk, since avatar.webp is a longer file name than avatar.jxl by one byte.

Flathub has officially removed their anti-LLM policy​:blobbee_sob_loud:​
https://github.com/flathub-infra/documentation/pull/641/files

They now require LLM-usage to be "disclosed," but the new policy allows them to accept LLM-generated apps anyway, making the "disclosure" pointless. There's no indication that apps which disclose LLM usage in submission will even make that information visible to users who may want to avoid them.

The conservatives will never be able to make "fetal personhood" into a coherent legal framework because we don't even have legal personhood for actual living breathing children.

Google: Here's our new cloud gaming platform. To play, you must either subscribe or purchase games individually. You get a choice.

Game "journalists": Google Stadia is a scam! You need to both pay a subscription and buy games individually!

Microsoft: Here's our cloud gaming platform. You must pay a subscription to use it. There is no option for individual purchases.

Game "journalists": Wow, Xbox Cloud Gaming is such a great deal!

Microsoft: Hold on, this isn't profitable enough. Even if you pay a subscription, we'll only let you play 5 hours per month. But you can individually purchase more time.

Xbox Cloud has now become everything the anti-Stadia-FUD people said Stadia was.

(In fairness to the anti-Stadia-FUD people, if Stadia had gotten more users, Google would have eventually made it worse, possibly by requiring both a subscription and individual purchases.)

Pandora's Tale Wiki launched four years ago today!
Here's my annual update post.
https://pandorastale.wiki/Meta/Blog%2Egmi

#PandorasTaleWiki #Wikiverse #Webcomics #Geminispace

https://f-droid.org/packages/com.dobby.vpn
WTF is this?

EDIT: Apparently it's also LLMs.
https://github.com/DobbyVPN/DobbyVPN/tree/main/.agent/skills

Also, the project has only existed since 2024, so they can't say "we chose that name before JKR showed who she really is."
https://github.com/DobbyVPN/DobbyVPN/commit/04cb7bf1e654c20163d6c757d12e1d6ccf0a66bb

Also, it was added to F-Droid in December 2025.
https://gitlab.com/fdroid/fdroiddata/-/merge_requests/29474

@fdroidorg@floss.social why are you hosting this?

Holy fuck
https://thisweekinvideogames.com/news/saber-interactive-acknowledges-undisclosed-genai-use-in-rideshare-stimulator-denies-replacing-lead-writer-with-chatgpt/

So to be clear, the order of events is
a) Writer Stella Sacco claimed she was fired from working on a game and replaced with a LLM. She refused to name the game because it wasn't announced yet.

b) Two years later, Saber Interactive announced a game, and the Stella Sacco said it was the game she referred to previously. The Steam store does not disclose LLM usage, which if Sacco is telling the truth is a violation of Steam's policy.

c) When asked to comment, the CEO of Saber claimed he had never heard of Sacco and he had to ask an LLM who she was. Then he asserted that his company has never and will never replace any workers with LLMs, which is totally something you can believe from a CEO who asks an LLM who a former employee is.

d) Then in the same interview, the same CEO says he wishes he had fired Sacco and replaced her with an LLM. Ya know, the thing he just said he has never and will never do.

Apparently the JavaScript version of Simon Tathum's Puzzles works in Servo. Servo is now a gaming browser.

#SGTPuzzles #Servo

Idea for Simon Tatham's Puzzles:
@simontatham@hachyderm.io

A setting for Loopy to generate a random planar graph embedding.

The player would select a number of vertices, then the game would generate a planar graph with that number of vertices using the algorithm from Untangle. The game would automatically solve it and then use the resulting planar embedding to generate a loop and clues.

Now I just have to learn C++ to implement it!


An even better possibility would be a new game that somehow required the player to simultaneously solve Untangle and Loopy, like how Netslide requires simultaneously solving Net and Sixteen. Maybe the numbers from Loopy are fixed on the screen, and you have to Untangle a graph with two edge colors in a way that produces a correctly solved Loopy puzzle? Seems like it might be hard to implement the generator.

Fuck, I now need a replacement for rclone. I use it for making both local and remote backups. Any suggestions?

#rclone #Backups

"An actor who is starring in new TV show X"
(Fact check: The actor in question is playing a tertiary character in new TV show X.)

This software is licensed under the GNU GPT version 3 or later.

Intersex Demiromantic Gay Asexual Fem (IDGAF)

Flathub currently has a policy which prohibits almost all LLM usage in both app development and app submission.

Unfortunately, it seems they are considering weakening the policy.
https://github.com/flathub-infra/documentation/pull/641

Those arguing for a weaker LLM policy make one point that I think has merit. The current policy is a double standard which favors proprietary apps. A proprietary app developer who doesn't publish their source code and doesn't explicitly publicly say "I use LLMs" could slip by the current review process, even if their code is mostly LLM-generated.

While this double-standard does seem like a problem, I don't think weakening the policy is the way to go.

#flatpak #flathub