Resentment-based politics is a trap. Even when the resentment is justified by the actions of some of the people you detest.
Individuals, not classes of people, are responsible for their actions.
Like sorry, but "a relative majority of rural voters are discriminatory assholes who periodically flirt with fascism" isn't a valid reason to remove subsidies from rural and working class areas. We live in a society.
Hot take but the answer is "because you live in a large, diverse society and that means fiscal support goes to lower income areas, even when you don't like them."
Doesn't mean it can't be conditioned on non-oppression, of course.
Although there is a reasonable argument that the fiscal transfer needs some amount of conditioning on certain social-cultural aspects. Hence diversity policies, equal rights protections, etc.
I think its important to separate the social-cultural-educational aspects from the net fiscal transfer.
Not really what you are going for, but its kind of a fact of life of a large fiscal union. Professional urban centers always subsidize rural and working class urban areas.
The IPE version of not doing so is the Eurozone crisis and Greece.
4. It can be a little uncomfortable to about the Tanakh, the Hebrew Bible, in a space where Christians (and culturally Christian atheists) also are. Even if they're well-intentioned, they have a very different set of assumptions and exegeses. It makes talking about certain things awkward.
nature occasionally red in tooth and claw but quite frequently just wants scritches
🔥🔥🔥
Completely off topic but this is why when we play "Normal Men FMK" everyone wants to marry propter.
Yeah, I think that's a good summary of many of my concerns.
For those interested in a more optimistic techno-futurism take, see 🧵... There are substantial elements of this that make sense, and some that I hope will be true once scaling occurs.
There are also a lot of interesting political questions re: labor, medical assistance, security practice, etc.
Good take. Everyone shitting on the WH Correspondent's Dinner needs to go rewatch the 2006 Colbert Roast. It doesn't happen often, but there tradition is important.
(Oh lord those hyphens. Oops.)
I don't have a counter to this one. Perhaps my gut underindexes on path-dependence and over-indexes on "this is a real problem that would generate political pressure."
Student loans are still a major issue, after all.
More seriously, it might take a real flashpoint effect/natsec push. That's possible.
Like, my gut here says y'all are overindexing on teething problems.
Isn't that also a regulation problem? It's PE all the way down.
To use the medical data discourse from a while back, just pass HIPAA 2.0, cybersecurity boogaloo.
See here for the kind of scope question I'm concerned with. I'm thinking 5-10 years out equilibrium, guy in basement not Chinese govt.
My concern is more "unregulatable guy in basement"
Obv there are real national security concerns here. No arguments on that scale.
I am having trouble figuring out how this doesn't just become "AI Windows Defender" in the medium-term, given the capex hurdles to be on the frontier.
Getting pushback from Ed, who knows more than I do here, so... A question - suppose OpenAI becomes the Oracle of institutional cybersecurity protection in ~5-10 years. The main SaS provider.
What hardware is required to develop & train models, regularly, that could defeat those protections?
Very specifically thinking about development & training requirements. I should've been more precise about that - I said "hosting" but that wasn't quite correct.
Can you expand a bit? Like, developing the new ones? Not my field and I'm quite interested.
Serious question: Can I develop and host Mythos on a gaming desktop? Because if not, the relevant # of actors to regulate and monitor remains small.
My econ-brain says the relevant part of your equation is the computing power, not the math.
Until I can host a Mythos-level agentic model on a standard gaming desktop, the number of actors who need to be monitored remains quite small.
To use @progressiveknife.com "morphine vending machine" example, this isn't Breaking Bad meth labs in the backyard, it's industrial-level production.
This looks like a good take on the surface, but if you put on a political economy hat, it's less so. The relevant part of the math + computing power equation isn't the math, it's the computing power! (1/2)
Basically, there's an ease of scaling potential there, if you can make the specialization the robotics equivalent of SaS-based rather than locked into differing forms and production lines.
Historically, that's a big ask, even for actual SaS. It usually comes much later, when the segment matures.
