Maxine 💃🏼

Representational Momentum Transcends Motion Dillon Plunkett & Jorge Morales (2025) Psychological Science

When we see something that's moving, our memories about it end up projected forward in time: We remember it further along than it was. In a new paper in 𝘗𝘴𝘺𝘤𝘩𝘰𝘭𝘰𝘨𝘪𝘤𝘢𝘭 𝘚𝘤𝘪𝘦𝘯𝘤𝘦, out today and led by @dillonplunkett.bsky.social, we demonstrate that this happens even when there is 𝙣𝙤 𝙢𝙤𝙩𝙞𝙤𝙣 𝙬𝙝𝙖𝙩𝙨𝙤𝙚𝙫𝙚𝙧.🧵

The drive to make versions of 39 and (the resulting state downstream of) 42 happen is a major reason I do what I do.

Actual Bluesky challenge: post your favorite work of art from fifty years before you were born.

YouTube video by TVUVEVO Nico - These Days (Visualizer)

traveling alone soundtrack youtu.be/52ZdwZ7Ig-8?...

the obvious answer is the creation of a centralized authority over all AI headed by one pereon

but we’re really worried about dictatorial control over AI in China tho

This was my first response when it came out.

Like—I mean I get it, standards are important, Schelling points and all that.

Also … this one? …

All of these were created by acts of Congress.

There probably was a time when someone would mention something about a Constitution, but alas that time seems to have long since passed.

is this AGI

The notion that there is any scalar quantity within evolution that is globally optimized over time is looney; evolution says “what makes more, there’s more”, and so any tradeoff in service of making more is a potential strategy! When a new affordance becomes realizable, all bets are off.

But, the reason neurons do it is because in distant parts of evolutionary history, it became necessary to synchronize motion of an appendage—and so, a new parameter entered the equation, transmission speed! It made total sense to trade off energy for speed, because it provided new affordances. +

Even phrased about energy expenditure, I find these arguments to be a bit ridiculous on face also. Your body computes Everywhere. So do organisms down to the scale of even a tiny stentor. Neurons have a Massive energy expenditure for the amount of compute you get! +

childhood is just embedded internal memetic evolution.

A hinged perspective on LLM use in a practical setting!

Yeah I would say my definition of positivism is actually a family taking as input your class of transformation rules.

I use that word because these blurry constructs have arisen because of a clear lineage in the western canon, an in practice are conflated with ways of knowing that are heavily institutionalized.

my definition is not unused.

Screenshot from Wikipedia: Positivism is a philosophical school that holds that all genuine knowledge is either true by definition or positive - meaning a posteriori facts derived by reason and logic from sensory experience. I2! Other ways of knowing, such as intuition, introspection, or religious faith, are rejected or considered meaningless.

Right of course—I was using a shortcut for microblogging.

I will point out:

Photograph of a blonde woman in a black hat, glasses, and a red top riding up an escalator in a large convention center. She is wearing a lanyard reading “Maxine Levesque, Forecast Bio @ Astera Institute”.

Hey Bluesky fam—

If you’re at NeurIPS and looking to work on something awesome in AI x bio, say hi to this lady (me!). We’re making real things people need with data nobody else has. And we’re recruiting! Bonus high-five for people from the atmosphere 😁

Thus, positivism is wrong (metaphysical truths can only be stated dialectically) and psychologically unhealthy (this internal fantasy serves only a dissociative role that stands in the way of genuine self-discovery in a world that is beyond self-actualized control).

Hence why I use it pejoratively.

Positivism is attractive in historical conditions where the world feels chaotic and out of control. Within the mind, the only way to cope is to produce the belief “No—through my intellect, I may know and control everything.” It is ego-defense in light of an outer world clearly out of one’s control.+

I use “positivism” as a pejorative because it has a core dissonance: it clearly prescribes a metaphysical position (“there is truth beyond us, and it is knowable only through science”) while also usually deployed in a way to counter other metaphysical positions (“there is no spiritual knowing”). +

Whether the paper is “positivist” is a question of interpretation: does one believe the “real” meaning & benefits of a concept like “impartiality” are actually discovered, and only discoverable, through the empirical process of science? Or are there other ways of truly knowing “impartiality”? +

This paper is neither positivist nor not-positivist intrinsically—they simply propose the descriptive power, and potential social benefit under a specific normative framework, of using a specific new set of terminology within discourse. +

Positivism is the belief that all genuine knowledge comes either from tautology or from a specific prescribed process of, essentially, crystallized and idealized enlightenment science—logical reasoning from data. +

You’re misunderstanding the sampling bias I’m referencing.

This is “1/3 of people surveyed about Coke while randomly out and about drinking the Coke they chose to buy were dissatisfied with Coke.”

Which I don’t deny is the case … *because*—my point—it’s an addictive and unhealthy product.

But ostensibly people using enterprise SaaS don’t actually have a choice: a certain tool is required or prescribed by a manager, etc., and people using it just use whatever the company does. Of course ratings are low for what isn’t necessarily your choice.

Screenshot from anthropic blog: Results Al's impact in the general workforce Overall, the members of our general sample of professionals described Al as a boost to their productivity. In the survey, 86% of professionals reported that AI saves them time and [highlighted in yellow] 65% said they were satisfied with the role Al plays in their work.

“65% said they were satisfied with the role Al plays in their work”

bro these people were interviewed **while they were using Claude**—that means 35% of people using your product were not satisfied with the product they were using while interviewed.

www.anthropic.com/news/anthrop...

One we get community agreement on schema lenses on-protocol, all of this will be moot: tooling will be able to codegen for all compatible / interconvertible types!