Marcelo Rinesi
marcelorinesi.bsky.social
Marcelo Rinesi
@marcelorinesi.bsky.social
Cognitive architecture designer and consultant.

On https://rinesi.com there are links to my blog and newsletters (one for articles, "what's new on arXiv," etc, the other for original short-short SF).
An expanded and horrifyingly useful Classically-inspired taxonomy of tasks.
Oh, shiny tomato!
the fact that we only have “herculean task” and “sisyphean task” feels so limiting. so here’s a few more tasks for your repertoire icarian task: when you have a task you know you’re going to fail at....
ghostlyarchaeologist.tumblr.com
November 27, 2025 at 2:39 PM
Great news! When you think about how aloof neutrinos are, the very idea of a "neutrino observatory" is awe-inspiring.

by @jeanna.bsky.social
China’s Giant Underground Neutrino Observatory Just Released Its First Results—And They’re Promising
Hidden beneath the hills of southern China, the JUNO observatory shows promise in solving neutrino mysteries
www.scientificamerican.com
November 27, 2025 at 2:37 PM
I fully agree from an intellectual and philosophical point of view. I'd also add the idea fails on its own terms: "the market" is a shorthand for "company demand" which isn't always aligned with the long-term interest of students. 1/

via @bildoperationen.bsky.social
In sum, calling for university curricula to “adapt to the market” is not neutral reform. It’s a normative position that privileges one cognitive interest over the others.
November 27, 2025 at 2:23 PM
"We all knew our lineage owned a star and a bacteria; we also had significant debt with a bank not much younger than our name."

#ShortFiction #ScienceFiction #AdversarialMetanoia
Collateral Dreams
We all knew our lineage owned a star and a bacteria; we also had significant debt with a bank not much younger than our name. But it was only after my father died and as the lineage chairperson I was…
blog.rinesi.com
November 27, 2025 at 1:15 AM
I distrust "national character" explanations but the United States political system has something of tradition of letting treason slide when it's done by rich white racist guys.

(Although looking at Washington et al you kind of get why.)

via @jonathancohn.bsky.social
November 26, 2025 at 7:34 PM
Strong candidate for If I Had to Pick *ONE Thing for Every Manager and Journalist to Read about Data Analysis."
November 26, 2025 at 7:10 PM
A classic, and details aside still relevant; wish more people building UIs and libraries had the time, resources, and training to think about these issues. via @icecolbeveridge.mathstodon.xyz.ap.brid.gy on aperiodical.com/2025/11/doub...
Notation as a Tool of Thought
www.jsoftware.com
November 26, 2025 at 7:08 PM
"most central banks are using AI mainly for basic work, such as summarising data or scanning markets" -- I get scanning markets (LLMs are the best we have at language) but unless they are using AI in better but non-contemporary ways, that's *not* as reassuring/conservative as they seem to think.
November 26, 2025 at 7:05 PM
In all seriousness, I believe a cultural expectation of reading your email once when you begin your day and sending replies and new emails near the end of it would satisfy most use cases, increase well-being, and nudge organizations towards more productive workflows.

via @drjlhazelton.bsky.social
I have arbitrarily decided that noon today is the cutoff for sending emails without being rude (not expecting responses).

Gotta go send some emails.
November 26, 2025 at 6:58 PM
That's a key observation: Generative AI lowers the cost of some forms of labor by [more-or-less/kind-of/if-you-squint] replacing it with a cheaper alternative, so order of magnitude its impact is on workers with no negotiation power or standard-setting institutions.

via @kevinriggle.bsky.social
It's also notable that this technique is largely used in professions that are already highly exploitative - voice acting, modelling, music. These are people who already struggle to find good work, and can't easily advocate for better deals. So there's not a lot of negotiation going on.
November 26, 2025 at 4:53 AM
I'm generally speaking skeptical of LLMs in mathematics (and fields other than, you know, language processing in its most literal sense) but interleaving them with formal checkers like Lean seems like an interesting approach; certainly don't do maths without one at the very least!
HERMES: Towards Efficient and Verifiable Mathematical Reasoning in LLMs
Informal mathematics has been central to modern large language model (LLM) reasoning, offering flexibility and enabling efficient construction of arguments. However, purely informal reasoning is prone...
arxiv.org
November 26, 2025 at 2:01 AM
Rubik's cube is a very interesting arena to study expert performance; I understand AI is where it's at these days, but until the still-mythical AGI, the development of human expertise is still a rate-limiting factor for a lot of things.
Universality in Collective Intelligence on the Rubik's Cube
Progress in understanding expert performance is limited by the scarcity of quantitative data on long-term knowledge acquisition and deployment. Here we use the Rubik's Cube as a cognitive model system...
arxiv.org
November 26, 2025 at 1:58 AM
I think the best framing for LLMs isn't technology but *information economics*: Welcome to a world where the marginal cost of the plausible, meaningless word is zero and therefore the information value of all world-writeable channels collapse.

via @puntofisso.bsky.social
When an AI-powered writing tool was rolled out on a job site, the length of proposals exploded. Signals employers used to identify good candidates — like quality of writing, and relevance of experience — became ubiquitous. That was OK for bad candidates, but terrible for good ones.
November 26, 2025 at 1:54 AM
This is incompatible with the rule of law: not in the sense of "this has to stop" but "the fact that is happening proves that the rule of law currently is, at the very least, conditional on how you look, and therefore broken." Individual wins and reprieves are just that.

via @t0nyyates.bsky.social
November 26, 2025 at 1:28 AM
"We all knew our lineage owned a star and a bacteria; we also had significant debt with a bank not much younger than our name."

#ShortFiction #ScienceFiction #AdversarialMetanoia
Collateral Dreams
We all knew our lineage owned a star and a bacteria; we also had significant debt with a bank not much younger than our name. But it was only after my father died and as the lineage chairperson I was…
blog.rinesi.com
November 25, 2025 at 5:17 PM
However far they get with this -this isn't an administration that excels at project planning and execution--the impact of this type of "AI-driven science acceleration" has consistently been negative[1], so I believe it'll further harm the already weakened US scientific system. 1/

via @jmberger.com
November 25, 2025 at 3:01 AM
As another example of the Trump-Milei[-Musk-etc] axis - the Argentine government has created a "National Agency of Migration" that explicitly references the CBP as one of its models.

FWIW, very few in Argentina -mostly the most online Mileists- are even... 1/
“Cambio de paradigma”: para reforzar las fronteras, el Gobierno crea la Policía Migratoria
El Ministerio de Seguridad presentará mañana la flamante Agencia Nacional de Migraciones (ANM)
www.lanacion.com.ar
November 25, 2025 at 12:29 AM
I wasn't at the right time and place to have one of these --my first computer, at age 5, was a Brazilian clone of a ZX Spectrum[1]-- but it's philosophically much closer to how I see computers than most things I see today[2].

[1] All... 1/
by @lproven.bsky.social
November 24, 2025 at 10:27 PM
If there's one area of AI --in its broadest and truest sense-- I'm optimistic over and fascinated by is mathematics; this sort of thing is very early days, but the combination of world-class mathematicians and increasingly powerful specialized systems will get us far.
November 24, 2025 at 10:22 PM
A not at all tongue-in-cheek observation is that the Chinese and North Koreans are on the cutting edge of state-level hacking, but nobody -*nobody*- has ever done it like the Russians when it comes to long-term asset development. If you know what I mean.
APT24's Pivot to Multi-Vector Attacks | Google Cloud Blog
PRC-nexus APT24 uses BADAUDIO malware in a persistent, multi-vector espionage campaign targeting Taiwan.
cloud.google.com
November 24, 2025 at 10:20 PM
Part of the planned changes in labor legislation in Argentina would include something like low base salaries + productivity bonuses; while Argentina's labor legislation and culture is (was?) quite... unique --not always in good ways-- the issues with "objective productivity metrics" are obvious.
Analizan impulsar convenios por empresas, cambios en indemnizaciones y la posibilidad de cobrar en pesos, dólares o euros
Se trabaja, además, en flexibilizar los vínculos y modalidades laborales, en crear un banco de horas y en una posible reducción de cargas sociales y condonación de multas, según un borrador al que acc...
www.lanacion.com.ar
November 24, 2025 at 10:15 PM
Wish I could take it. I also hope they look later into markets: LLMs are obviously hot these days in language, but I've always had a soft spot for markets as pseudo-minds[1].

[1] i.e. an Azatoth-egregore muttering to itself at high-frequency while the press mistranslates its nonhuman thoughts.
I'm teaching a new class at Berkeley next semester:

Linguistics 265: Biological and Artificial Language

From the syllabus:
This class is an introduction to a novel approach to language. Linguistics has predominantly focused on
human language.
November 24, 2025 at 10:08 PM
There's a fragile equilibrium here: the political system under-reacts to these threats because it assumes they won't act on them, and/so they escalate the threats. Question is what happens the first time an enthusiastic FBI Patel guy violently grabs a Senator off the street .

via @jmberger.com
“Within the White House, Trump and his lieutenants aren’t planning on letting up…Trump has told some advisers that he wants RICO prosecutions and conspiracy charges leveled against the six Democrats, a source with direct knowledge of the matter tells Zeteo…
zeteo.com/p/how-we-gre...
November 24, 2025 at 10:00 PM
As @edburmila.bsky.social notes, Venezuela's oil reserves aren't really worth it. Trump's US is good at individual-level grift but, as openly rogue as it has gone, it's really incompetent at imperialism.

Mind you, this doesn't make the US any safer for everybody else (or itself).

via @ejfagan.com
Rep. Salazar on Venezuela: "We're about to go in ... we need to go in ... Venezuela for the American oil companies will be a field day"
November 24, 2025 at 9:48 PM