Sebastian Benthall

I wonder what @dkthomp.bsky.social thinks about Georgism

This bitter exchange caught my attention. "two different [homebuilding] groups. (1) financial middlemen who ... control the industry, buying cheap and hoarding land. (2) actual developers ... are boxed out because they can’t borrow money." Builders= price takers, profit on low credit costs?

Thank you for hitting the nail on the head. I remember the last "post truth" moment -- 2016? -- but somehow this one feels more severe.

How paranoid! It's a capitalist controlled superintelligence from the present with forward causal force of course.

Traveling in the UK. It's comparatively quite nice over here.

Zohran didn't win by relying on standard turnout patterns. We had to fundamentally transform the Democratic electorate.

I was part of a group of NYC-DSA members that applied lessons from our previous campaigns to design the “universe” of voters to talk to. Here's how we won 🧵

I think it's fair to say that the US has two national visions, with different ideas, different tribes, different soils, coalesced in two parties, competing for central control. What's missing from that description is a characterization of the role of elites.

Congratulations!!!

Josh Tenenbaum's inspiring keynote at #cogsci2025 on growing vs scaling AI, the big questions of cognitive science, and the many open questions for the field.

Why does he put his own name in scare quotes?

I have been thinking about "National Brotherhood Week" for a couple of months now, so today's news hits hard.

Photo of Tom Lehrer

My last living musical hero is still my hero but unfortunately no longer living. RIP to the great, great Mr. Tom Lehrer.

Unless you're allowing extended cognition to staffers?

When do you think this present stratification event started?

Ah. So you are saying that use of an 'external' tool, specifically GenAI, result in an increase in 'cognitive' power? So, stratification in an extended-cognition sense, base on access and use. Hmm.

Huh?

I believe we were both at Berkeley when they redid the whole undergrad curriculum around "data science", which is no longer trendy. I wonder how they feel about that decision now.

Chasing an industrial hype cycle seems like a bad move for universities. Why can't they stick to fundamentals?

Data is that literally not of the same 'type' as the training distribution is indeed 'nullity'. The 'exception' in that case is indeed a TypeError, which would be due to a programming error. The ADS crashes, and the then company from which the system is procured does damage control.

Extrapolation is an option for the 'untrue' exceptions. This is one reason that the AI safety/responsibility are always talking about 'distributional shift'. If there were such a thing as a free lunch, there models wouldn't need to be retrained.

"ontologically outside" is perhaps different.

The 99-1 flop of the restriction on state AI laws in the BBB shows that the industry is not as (legislatively) powerful as they seemed at the start of this year.

In deep learnt automatrocracies, the exception is decided by a smooth interpolation between unexceptionable cases.

A model trained with a frequentist objective using i.i.d. observational train and validation data is almost surely converging to the non-causal model of the underlying data generating process. That is because using short cuts/spurious correlations leads to a lower loss than not, i.e. you are actually specifically incentivizing your model not to be causal. "Emergence" believers' core fallacy is the assumption the causal model would have the lowest error, and therefore better optimization would eventually converge to the causal model. The opposite is true - the best scoring model (per frequentist objective) is actually the one that is best at exploiting spurious correlations.

Beautifully succinct statement of the core problem with the blind application of statistical learning in AI!!

I just heard that ag school and vocational programs are going way up.

Oh no

There are many federal regulations about advertising mortgages (and some other financial products) that are not illegal.

Ok, GenAI is useful for coding. CoPilot's pull request reviews on GitHub are surprisingly good. Asking specific questions about solutions to problems, where some conceptual idea needs an explanation, is useful. There's a lot of ways to use it that aren't "vibe coding" at all.

That's a sensible theory. It has weird implications in equilibrium.

What, if anything, can lead to a successful populist electoral revolt against elites? Some have thought that class consciousness can do it. But that rarely happens in a consumerist society. "The people" are a crowd beast not a rational actor. What might do it is tabloid-level viral scandal content.

Tay (chatbot) - Wikipedia

I dunno. Microsoft Tay happened in 2016 and wasn't a huge surprise. Tay (chatbot) - Wikipedia share.google/F7kDUuSJCwte... "Internet company doing stuff with Internet content is racist, does porn, and gets DoD contract" feels standard. Is the surprise that there's no repurcussions for nazislop?

Tell that to the new factions that think that the best way to interpret the Bible is to consult a GenAI SaaS product about it.