Experts are calling it the biggest breakthrough in pancreatic cancer treatment for decades

- a daily pill that appears to shrink and possibly even eliminate tumours in some patients
with one of the deadliest and hardest-to-treat cancers.

Cancer became undetectable on scans in one in 12 of the patients given #daraxonrasib along with chemotherapy in early medical trials.

Tumours stopped growing in almost all of the 61 participants who had recently been diagnosed with advanced pancreatic cancer that had spread to other parts of the body.

But doctors stress it is a long way from knowing how long it might keep the disease at bay and should not be seen as a cure.

bbc.com/news/articles/cqrmgjnz

"The Workers’ Struggle Behind GTA 6" by Novara Media and People Make Games
youtube.com/watch?v=cpI7wkzzQ8

I've always enjoyed movies where the concept is essentially impending doom and people go down into a bunker and then finally coming out decades later to find that the world didn't end and everything continued as usual, so I was thinking about a modern day version where people get fed up with AI and head to the mountains, but then they realize they HAVE to go into town for something after decades of living off grid and they get to town and the shop keeper is like "Oh, yeah, the day after you left, the world collectively realized that the multitude of negative aspects of AI weren't worth the handful of benefits, so it was banned entirely and I haven't even thought about AI in about 20 years since the most annoying thing we have left is just Bing" and the mountain people are like "There's STILL Bing?!?" and then they all flee back to the mountains.

We have written up a post-mortem for the folks who have been wondering why our domain was suspended for a while on November 18.

The tl;dr: A person reported a malware-containing Codeberg repository directly to our domain registrar, which proceeded to suspend the domain without coordinating with our team.

We outline some planned steps to avoid this in the future. Read the full bit here:

codeberg.org/Codeberg/changelo

Show thread
rewrite it in rust so we can replace hard to exploit memory safety issues with easy to exploit supply chain attacks!

what would it mean for a service to have an uptime greater than 100%

@johncarlosbaez

Strave did random changes to my CUP paper with Alex Simpson, starting from the title, which is

"Euclidean interval objects in categories with finite products"

They changed "interval" to "integral".

I thought it was the copy editor, but now I see it was an AI...

If you haven't boycotted Springer publications yet - they charge high prices and demand complete rights over your papers even if you pay to publish open access - you may be convinced to do so by their new AI-based production system.

It's called Straive. Harry R. Lloyd, an assistant professor of philosophy, writes:

"Of course, one expects some common-or-garden incompetence from production companies, but unfortunately Straive takes this to a whole new level, introducing dozens of bizarre errors. For instance, in my articles they have: removed entire paragraphs; replaced several instances of the pronoun “one” with the numeral “1”; deleted countless possessive apostrophes; and removed numerous grammatically necessary connectives. (Oh, and they are also completely incapable of typesetting even basic mathematics.) One of my articles with Synthese had to go through six rounds of proofs, and even then Straive left errors uncorrected in the published version."

dailynous.com/2025/12/18/probl

I was paid a great compliment today at work. For the past 3-4 weeks I've been working on a high-priority study to explain where we might refine our analysis to remove excessive conservatism. This was an unusual request on a short timeframe and it resulted in a report just over a hundred pages. It went to the technical reviewer yesterday, someone familiar with the wotk and who has reviewed my work before (and vice versa).

She had no material comments, just a few (discretionary) editorial.

This is like typing in a program and having it run correctly the first time. It's theoretically possible but in practice it never happens. It puts the reviewer in a quandary - how do you demonstrate you thoroughly reviewed a document if you don't find anything that needs correcting?

One way of resolving this is to write up some basically meaningless issues; it's busy work for the preparer but it's easy to deal. I don't like making busy work for people so I've gotten into the habit of logging what I've reviewed and summarizing it. It demonstrates I read and understood the material and found it acceptable and it requires no further work on the author's part. It's no less busy work than making the author address some pointless BS, it just puts the burden on me not the author. I'm good with that.

Engineering has much less of a Reviewer #2 culture than academia; the purpose of the documents is different and reviews are not anonymous. There's usually no incentive towards Reviewer #2 behavior. The last report I sent to this reviewer was sort of troublesome - there were a crapload of citations into books that aren't easily available or common. She quite properly asked for Citation Please and I spent a day or two tracking down electronic copies or taking photos of the appropriate pages. That was kind of miserable for both of us but it was proper diligence. As a result of that review I became much more diligent at adding specific section or page numbers to citations - very much citing chapter and verse. It helps the reviewer find things and means I don't have to guess where I took material from. It adds a little time to document prep but it simplifies review and helps me remember where I got things from. "Here - it's in this 800 page reference book somewhere - good luck!" Not a good look and you don't want to build a culture where that's common. It's inefficient and hard on the reviewer; everyone's busy so have some courtesy - it won't kill you to type another 10 characters.

I know how pedantic of a reviewer she can be and I appreciate it because the work improves because of the review. I was about as shocked as she was that she didn't find anything.

She went hunting for typos and grammar problems and came up empty. She said it was like picking up a piece of bubble wrap that had already been popped.

In decades of doing this kind of work I've never had this happen. It's memorable. And what I love about the whole thing is that she was instrumental in getting me to up my game by being a dogged reviewer.

youtube comment:

> Massive amounts of AI coded patches for security vulnerabilities supposedly found by AI, patching mountains of AI coded features that didn't work well before, and work even less-well now. Digital Kessler Syndrome.

god yes. digital Kessler syndrome.

Ask your doctor if hiding under a pile of blankets until spring is right for you

As one example: Etsy just docked my account for multiple hours because I passed its captcha crap but something in the melange of Stuff quietly tried to fire an async fetch to a "domain" that was a string of line noise with no TLD and noscript went "WTF no???" and that got vee whacked for "suspicious activity"

That looks like dead-ass malware behavior.

Show thread

I sometimes complain about Google Maps being especially bad at giving directions for mass transit, but within the past month, it's gone from "That's kind of a silly way to get there" to "This is like asking somebody with dementia for directions". The routes are completely disjointed.

I asked it for a route to Staten Island and it suggested taking an expensive tourist boat to see the Statue of Liberty, and then walking across the harbor water to the Staten Island ferry that passes by the Statue. The Staten Island ferry doesn't go to Liberty Island and never has.

I asked it how to get to a bar in Brooklyn, and it suggested taking an uptown bus westward to Harlem, exiting at 133rd street, then walking east towards Alphabet City several miles downtown near 14th street and getting on a train that never gets anywhere near 133rd.

Show older
Computer Fairies

Computer Fairies is a Mastodon instance that aims to be as queer, friendly and furry as possible. We welcome all kinds of computer fairies!