Tessera
Folgen

Ryan Randall

@www.ryanpatrickrandall.com

A better world is possible, together. Instructional design, humanities/Cultural Studies, pedagogy, open ed (OER/OA), library work, accessibility, ADHD. https://www.ryanpatrickrandall.com/ https://hcommons.social/@ryanrandall

5381 Posts2530 Follower3538 Folgt
wendy norris@wendynorris.bsky.social · 7h ago

Meanwhile, while we're all simultaneously horrified by Faine's must-read threads on abuse permission structures within the "AI Safety" world/sportsball scores/fascist capture of democracy/etc., Jonny's still cookin' on the surveillance capitalism disaster that is Meta Muse.

This is I think the most consequential thing I found in the muse VM, or at least the thing that would be most worth following up on. there are no other plain string matches I can see for what uses it, I'm looking in the binaries now, but the comment seems to indicate pretty clearly that a) meta uses anthropic's models internally, presumably to develop the thing (the code, comments, and skill text reek of claude), but also b) they use some kind of internal proxy that intercepts and proxies all requests to anthropic's API. the only reason I can think of for doing that would be model distillation.OH what's this??? /etc/hatch/env hmm a feature flag labeled as a "killswitch" for proxying requests to anthropic is certainly interesting to seeeeeeeeeeeee in meta's big AI bet. edit: sorry to be a bit cryptic here, the implication was definitely corporate espionage, more clearly said:The system prompts are indeed in the hatch binary and are the things that contain all the behavioral guidance. I cannot for the life of me figure out why every AI company decided to make system prompts be a clientside thing rather than something that gets injected serverside. Like maybe some you're supposed to be able to edit, but not things that influence malicious behavior?!? The entire security posture here, if there is one, is oriented around preventing the LLM from misbehaving, but having the entire context be user-editable and user-visible is a sort of mind-blowing neglect of a misbehaving user. even for the "fully transparent agent" they're trying to do, it makes very little sense to store the full message logs of the subagents including reasoning output. That basically gives me a full map to see what the LLM is consulting when it refuses commands. I would also bet money that the system prompts are strings in the hatch binary even though that makes no sense to have clientside, but haven't looked.Am working on responsible disclosure so will pause thread, but unrelated to that I do want to say that this experiment in continual, boundariless context window backed up by infinitely refreshing context summary in the form of accumulating markdown documents works terribly compared to environments where you can control the context. Like that is an absolutely fundamental part of using these things and muse demonstrates that the "tree of markdown summaries" really does not work.
Ryan Randall@www.ryanpatrickrandall.com · 7h ago

In retrospect, my "remain obstinately aghast at the 'Higher Ed & Concussion Ball Exploitation Matrix' for decades" experience probably should have better prepared me for higher ed's "Let Us Now Praise Famous Slopforce™ Platforms" turn. And yet.

Andi Zeisler@andizeisler.bsky.social · 9h ago

The connections being drawn between the Cornell 7 and Brock Turner bring up something crucial: The outrage at Turner’s insanely light sentence has ultimately made it impossible for him to blend back into society. Where legal redress failed, community stepped in to make sure no one would forget.

Yvonne Lam@yvonnezlam.bsky.social · yesterday

I have been reading @annaleen.bsky.social's book about cities (highly recommend, btw!) and learning about how archeologists use LIDAR to find sites of former cities, so I am charmed that scientists use it to weigh bears!

Deb Chachra@debcha.bsky.social · yesterday

Second-screening the Katmai Falls bearcam today reminds me: Q: How do you weigh bears for Fat Bear Week? A:. By using LIDAR — laser-scanning bears to quantify the chonk volumetrically, and then using estimates of ursine density to convert it to approximate mass. www.geoweeknews.com/articles/sup...

Ryan Randall@www.ryanpatrickrandall.com · yesterday

Today's tech weirdness (of the old-fashioned variety): - digging deep into Harper documentation (mostly to make spellcheck squiggles more visible in my new editor) - discomfort over Harper being an Automattic joint - discovering a useful config template via an *archived, not current* GitHub repo

AAUP@aaup.org · 2d ago

Join us on Monday, 9/28 to learn why billionaires like Marc Rowan want to destroy academic freedom & how faculty, campus workers, & students can fight back to defend the freedom to teach, learn, research & speak out without government censorship. RSVP 👇 us06web.zoom.us/webinar/regi...

THE BILLIONAIRE ATTACK ON UNIVERSITIES RSVP TO THE WEBINAR 9/28 @ 3 P.M. ET
Ryan Randall@www.ryanpatrickrandall.com · 3d ago

Using a fancy, fun new code editor is all fun & delight until you realize you haven't enabled spell check in a way that lets you perceive typos as you make changes to your website. (Also, huge THANK YOU to person who reached out about that! The internet can indeed still be kind sometimes!)

Mike Hoerger, PhD, MSCR, MBA@michael-hoerger.bsky.social · 3d ago

The top graph shows our county-level COVlD map using the CDC's Standard Benchmarks. The bottom map shows the downshifted levels under the CDC's new system. This really makes a difference for early warning in the middle categories. I've highlighted 5 areas.

Graphic of old vs new systems
NBC News@nbcnews.com · 4d ago

The CDC raised its benchmark for levels of the Covid virus in wastewater, adjusting a key metric for assessing the disease’s transmission. As a result, officials may now classify a level that once counted as high transmission as moderate instead.

Ryan Randall@www.ryanpatrickrandall.com · 4d ago

Tiniest smol thing I've learned in the last few weeks that keeps blowing my mind: the CommonMark flavor of Markdown lets you indicate a new line with a backslash (\, which is visible) as an alternative to the usual two spaces ( , which is my muscle memory, ergonomic, & often invisible by default).

Roopika Risam@roopikarisam.bsky.social · 4d ago

Wow: “To be a librarian is to dance along the cracks of the institution, engaged in a thousand small teaching acts.” Good thing this is OA or I’ll soon be in violation of copyright.

Abigail Goben @hedgielib.bsky.social · 4d ago

"Can you come and talk to them about AI usage and the risks?" -- it was a request from an *extremely* frustrated faculty colleague. And so I put together this framework. The ACToRS in AI Framework hedgehoglibrarian.com/2026/09/22/t...

The ACToRS in AI Framework

About a year ago I was invited to give a lecture designed to spark critical thinking in graduate students when considering using generative AI and other tools that had started applying the “a…

hedgehoglibrarian.com

Prisonculture@prisonculture.bsky.social · 5d ago

Every person running on the Dem side can say: 1. The GOP controls all branches of government. 2. Are you happy with how things are going? 3. Vote blue. 4. What do you have to lose? [use Trump's own words against him]. The end. lol.

Taniel@taniel.bsky.social · 5d ago

A post that’s so eye-opening because this is exactly what liberals will say is GOP distraction tactics, almost as a parody— and here it really is what Gaetz is saying unironically as his GOP message.