provisionalidea

my most curmudgeonly current opinion is that

people who think a machine learning algorithm trained on internet language is on the verge of doing hard intellectual work

have no idea what hard intellectual work is actually like

the slow, painful, migraine-inducing, comprehensively-researched, logically-contiguous labour of love that goes into every great work of art, science, and literature

the deep skill that takes years to hone like a fine blade

Because quality work isn’t just about sourcing information or assembling words. It’s about having something you desperately want to say and struggle to first understand, then communicate, and fight over and over and over again to clarify enough to get onto a page.

And if you’ve never experienced that beautiful, infuriating agony, it’s easy to think a machine can just generate ideas and solve hard problems.

Because you don’t see the work that goes in, and think it’s a simple matter or serendipity, “intelligence”, or pattern-matching.

The following needs to be stated clearly:

People in tech are not afraid of AI.

They are afraid of an incumbent monopoly instantly rolling out a net-new innovation to every enterprise business before startups can even play around with it.

Which is a terrifying new dynamic!

An entire community who believed in their heart of hearts that “startups innovate, incumbents catch up” just saw that narrative undeniably violated on a world-changing technology.

And that panic is more a byproduct of worldview collapse than anything else.

Hello friends! If you would like to read some engrossing longform about Israel/Palestine this weekend (maybe with a good coffee or tea?)

I strongly recommend my newly-published article on the causes of the Oct 7 crisis in @newlinesmag:

https://newlinesmag.com/argument/how-changes-in-the-israeli-military-led-to-the-failure-of-october-7/

computers have been automatically generating news stories -- particularly in the business press at large newswire outfits like Reuters -- since at least 2007.

the strangest thing about self-described radicals reviving "Luddite" as a term of endearment is that it aligns them firmly with the bourgeois guilds, against the nascent factory workers

lionizing professional monopolies smashing new machines because they felt threatened by the prospect of mass production (and mass accessibility of goods and services) just because you're critical of modern tech companies is not the argument you think it is.

over the past few days I have found my little corner of the bird app completely taken over by people who think an evil superintelligent AI is going to take over the world and destroy humanity

I do not know where all these people came from

it is fundamentally different from the fear of computers taking our jobs

and it is coming from inside the tech community

and challenging that narrative results in incredibly vitriolic responses

I know it’s dumb, but the biggest reason I was off Mastodon for a while was because my phone broke and my old phone didn’t have space for the app.

I think in dependency trees a lot. Not just in terms of engineering or resource planning, but structured innovation, too.

“What does success look like and what would we need to know to get there” is as valid a question as any about personnel, capital, time, or infrastructure.

there’s a flavour of tech thinking that starts from the assumption that everything is already known. There is no place for science or discovery in this worldview; anything unknown can merely be derived from sets of arbitrary assumptions called “first principles”.

This thinking results in problems being redefined to fit available solutions, instead of being solved by deeply studying a problem and solving the unknowns until a proper solution becomes possible.

Which makes it a worldview devoid of real innovation.

Types of Visual Programming Languages (VPLs)

"Even a fully automated society would not be rid of red tape. Safeguards would be extensive; methods of appeal from errors or abuses would have to be developed. Most of all, the machines would impose an unyielding set of obligations and prohibitions on their users."

Listening to a lecture from 8 years ago where the scholar (who studied newsrooms) suggests a future where people submit news, a computer agglomerates it, and it spits out a story

the new LLMs do summarization and reformatting well

pair that with a social scraper and she’s right

the best version of this future is one where journalists are free to do actual longform analysis and deep investigation because they’re freed from the incessant loop of scanning social media for BREAKING CONTENT

which Reuters has been experimenting with (“breaking views”)

HOWEVER

The scholar points out that companies like Narrative Science have been producing algorithmic news (including off tweets) for years — they were doing it when she did this lecture 8 years ago. And yet journalists have not yet been freed; they’ve instead faced tighter time pressure.

Francis Crick’s intro to Images and Understanding

the conflation of hype and optimism in tech is weird to me

as if adding the tiniest drop of critical thinking turns you from an optimist to the most pessimistic hater

instead of someone thinking pragmatically about how to get from ideal to reality with minimal bumps on the way

for example:

"have you thought about how this change affects the user?"

"WHY DO YOU HATE TECHNOLOGY?"

or:

"maybe you shouldn't go around telling people you want to make their jobs obsolete."

"WHY DO YOU HATE THE FUTURE"

or even:

"I don't get why they hate us. Why are they against progress?"

"Did you tell them there's no place for them in the future again?"

"I mean, yeah."

"Maybe...don't do that next time?"