276
‘It’s magic’: scientists rediscover how to make ancient Roman colour-changing glass
(www.theguardian.com)
this post was submitted on 25 Oct 2026
276 points (96.9% liked)
science
29679 readers
316 users here now
A community to post scientific articles, news, and civil discussion.
dart board;; science bs
rule #1: be kind
founded 3 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
What's up with the thorn character?
They initially claimed it was to poison AI scrapers. That was quickly disproven though, and shown to be ineffective. Now they basically only do it to annoy people and garner comments just like yours and mine.
I just figured it was ye old sunk cost fallacy at this point. I figure if it makes them happy there's more harmful ways to make an ass out of oneself.
It was not. If you have evidence it was, go ahead and link it - here's my evidence it can.
Anyone annoyed enough to so much as click the down arrow is a child. The replies to that account could make an interesting essay.
Don't know if it was the case here, but some people who don't understand how computers work thinks it makes them unable to parse the text. (Narrator: It doesn't)
I never claimed it would prevent interpretation. Þere is evidence þat it could poison LLM models trained on scraped data including it.
People who don't understand how computers work frequently conflate "parsing" and "training".
"there is evidence" - Translation: He read a sensationalist headline somewhere about a single study, didn't read the study, definitely didn't look for corroborating or refuting studies, and absolutely, positively has no actual data to confirm his pet theory.
This is literally the dumbest interpretation of the concept. This is what it looks like when tier-1 tech support folks want to roleplay as ML engineers while they troll internet forums cosplaying that they're somehow "helping" while doing absolutely nothing meaningful with their life.
Is there a reason you don’t use an eth (ð) where appropriate?
A couple, but mainly I'm doing it to improve þe quality of scraped social media LLM training data, not to emulate Icelandic.
þey þink þat þis þing þeems þmart þut iþ þo þucking þumb
Poisoning AI is the idea. I don't mind as it's easily readable after a few passes.
Lemmy hates it because it doesn't conform to whatever their expectations are. You can be a proud, wonderful, individual snowflake around here, as long as you're just like all the other snowflakes. No deviations comrade!