Tony Curzon Price

Join our beta - firstdataunion.org/free-trial.h...

Google has no “right to innovate” - and what lessons regulators should learn from Götterdämmerung "Google cannot be permitted to leverage its dominance in general search to the GenAI product space" (Judge Mehta). Yet AIO and other copy-cat products show that's exactly what it's doing.

Another weekend substack - tonycurzonprice.substack.com/p/google-has...

Can the tired old incumbent cut off the (good) possible future that GenAI opens up? Perhaps. Antitrust authorities should do their bit to avert the danger.

"Immortal intelligence” has a big advantage over us mortals, which implies market power for just a few LLM labs. Here’s how to stop it. Geoff Hinton's Royal Institution lecture suggests the forces pulling to great concentrations of power in LLM labs are strong. But there are ways to resist.

My substack this week was inspired by Geoffrey Hinton's masterful, funny and worrying lecture at the Royal Institution in July. The lecture suggests the forces pulling to great concentrations of power in LLM labs are strong. But there are ways to resist.
tonycurzonprice.substack.com/p/immortal-i...

Descartes, Musk and the bad apple of an engineer. Why we need to own our own AI There's a big lesson in the world's best thinking model being explicitly coded to search for Musk's opinions on contentious topics. Data unions are a big part of the answer (as you'd expect me to say)

I'm back on the Substack after a busy month of June...
open.substack.com/pub/tonycurz...

My new post about a week in data/tech policy - this one about why we should be worried at the LLMs getting personal, and what we can do about it.https://tonycurzonprice.substack.com/p/charlie-chaplin-simone-weil-black

I've been wanting to write about Trump's view of the world as being shaped by how you run luxury hotels. It turns out to be closely linked to how you run platforms. Hence a link between his tariff mindset and the Vance/Silicon Valley "hands off my platform" mindset.
substack.com/inbox/post/1...

Merging Facebook data with administrative records is amazing for social self-understanding ... we need more of this sort of work, while avoiding creepy surveillance My colleagues' work provides a glimpse of what it could be for society to understand itself better … but to really unlock this power for good without creepy surveillance requires data unions

tonycurzonprice.substack.com/p/merging-fa...

Don't often find myself agreeing with this guy, but here I do. People donated time, data, skills, $s in the belief it would remain a non-profit. I have no idea of the law, but I hope he wins. A profit-seeking shareholder controlled LLM is a very different beast from a purpose-driven one.

Very pleased to release - slowly - into the wild version 1.0 of The Graeme Game - graeme.metakarma.org .

Would love to know how you get on.

(A game can be over in minutes; you need one other human to play with; you are playing against the machine to prove our common humanity).

Yes, Solid on its own puts too great a burden on the individual, when most of the value that comes from data - and the very reason that platforms have built both great power and compelling products - is that it is collective. Hence the "data union" metaphor.

as a pragmatist, I would first lobby for the minimalist version, then nudge and hope that cyberspace evolves in the maximalist direction.

the "minimalist" would impose on current platforms a duty to provide an API on fair/reasonable/non-discriminatory terms, so that the union would not be shuffling data around so much as managing access permissions [...]

I agree, @sigridellesmere.bsky.social, security and privacy need to be at least as good as they are in our platformised world. imo, There are 2 broad approaches: the "maximalist" would follow the Tim Berners Lee/Solid vision of separating apps and data and having specialist "vault providers [...]

To provide credentialised data to individuals that the public sector holds about them

I had some great conversations over the weekend at the Cosmos Institute symposium, especially, on the topic below, with @Gigi Danziger & @albertwenger.bsky.social . I had to write up the learnings for me... "How does enshittification come to LLMs and how to stop it?"
metakarma.org/blog/LLM-ens...

First post in the new place.

Seems appropriate to make it a screenshot from this morning ... adding
@pluralistic.net.web.brid.gy 's vocab to my personal dictionary in google Docs seems like a moment to memorialise.