marcelorinesi.bsky.social - Profile | ThreadSky | a Reddit-style client for Bluesky

"Threat" is a bit of an understatement - if you have to literally litigate your Constitutional rights every time, for every person, then you don't have a Constitution, you have a legal strategy. A republic, if you could have kept it. At best you'll have to retake it. via @estebanjq.bsky.social

submitted 24 minutes ago • 0 comments

I'm happy (and/or in the clutches of confirmation bias) this matches from an slightly different angle my view of the labor impact of genAI. Note that this explains why companies keep trying to deploy LLMs on high-expertise rather than low-expertise... 1/ by @davidautor.bsky.social and Neil Thompson

submitted 14 hours ago • 1 comment

Points where deserved, that's very much the right response. via @brendannyhan.bsky.social @robshum.bsky.social

submitted 15 hours ago • 0 comments

My usual view of "flooding the zone" is in terms of degrading institutional/societal information prioritization channels (and secondarily, individual attentional resources); this paper is an interesting look at a different possible mechanism via (my phrasing) shallowing out coordination points.

submitted 22 hours ago • 1 comment

I'd like to remind @businessinsider.com 's headline writers that Salesforce sells AI - this is a marketing statement, not an audited SEC filling. Well, "remind" - you know this. Come on. via @edzitron.com

submitted 22 hours ago • 0 comments

People I trust do use AI more than I expected; the tech is quite good. OTOH, they haven't "gone exponential." My model of genAI is that it's only useful paired w/expertise & autonomy, so this won't generalize in the way companies would like. via and agreeing w/ @marypcbuk.bsky.social

submitted 22 hours ago • 1 comment

And, sadly, politics. via @jgkoomey.bsky.social @drjlhazelton.bsky.social

submitted 22 hours ago • 0 comments

A lifetime of reading SF dystopias and I never considered the possibility of a Cronenberg one. (I mean, Philip K. Dick - but his worlds were crazy orthogonally to this.) via @marypcbuk.bsky.social

submitted 23 hours ago • 0 comments

One of the funniest aspects of US culture is the assumption that Republicans are the Respect The Military guys. If I were a military professional I'd spend 1/3rd of my time planning operations against this guy and another 1/3rd reminding myself that I promised not to. via @alanallport.bsky.social

submitted 23 hours ago • 0 comments

Generalizing "cyber" beyond "IT," the definition of "cyber-warfare" is "make your enemy stupider." Dunno what those bombs did to Iranian capabilities, but in state capability terms the US keeps nuking itself over and over. via @vermontgmg.bsky.social

submitted 23 hours ago • 0 comments

1. All kudos to @avsecz.bsky.social et al. 2. Note to media: this improves existing methods (cool & hard!) but doesn't "solve biology." Please tone down the bio AI startup hype BS. 3. @avsecz.bsky.social I'm begging you on my knees to remember Hamming's dictum and do more mechanism discovery.

submitted 23 hours ago • 0 comments

@krishardies.bsky.social knows more about this than I do, but my first reaction is that it has some undesirable systemic features (although some of them can be mitigated via more complex rules inducing better information aggregation dynamics). But... 1/ via @dingdingpeng.the100.ci

submitted 1 day ago • 1 comment

I like this both as a definitional statement and as a diagnostic tool. You can always trace back allowed ethically and/or intellectually shitty behavior to structural conditions and incentives that are or will make it common. via @marypcbuk.bsky.social

submitted 1 day ago • 1 comment

A way to put it is that we were already in a race to the bottom in the volume-vs-quality tradeoff in every context without an ethos of intellectual/informational integrity, and genAI made the incentives exponentially worse. via @rusty.todayintabs.com @dynamicsymmetry.bsky.social

submitted 1 day ago • 0 comments

Risks, informational asymmetries, etc go w/o saying. We talk about FAFO re: Trump's policies but this is the FO phase of a longer trend of tech/data collection "but only for {context}/against {group}." Ultimately, democratic control of security forces is the sine qua non. via @marypcbuk.bsky.social

submitted 1 day ago • 0 comments

The way Wall Street has echochambereded itself into panicking about a guy who, in absolute terms, has no radical policy views is a disturbing reminder of the often flimsy (or to be fair: extremely specialized in a culturally over-appreciated way) collective cognitive infrastructure of "the market."

submitted 1 day ago • 0 comments

As a professional taylorist, I'd frame this [100% agreed-with view] as: * Your utility function is votes, not money. * Optimizing for that implies different mechanisms, cultures, in-house knowledge, etc.

submitted 1 day ago • 0 comments

This might need to happen over the still smoking rubble of the current Democratic party leadership & strategic culture. Judging by their performance against an authoritarian attack by the stupidest and cruelest administration in living memory, this counts as a bonus. via @drjlhazelton.bsky.social

submitted 1 day ago • 0 comments

Kudos where merited - more restrained than I'd have been, but that's probably good. via @chadbourn.bsky.social @drjlhazelton.bsky.social

submitted 1 day ago • 0 comments

Honestly the way wildly influential segments of US society -not just the billionaires- have a deranged view of the policy spectrum ["communist"???] is an under-studied factor in this ongoing debacle. They don't watch Fox News. Are they all on X? Is it the NYT/WP? 1/ via @drjlhazelton.bsky.social

submitted 1 day ago • 2 comments

One would hope it's entirely unnecessary to remind lifelong politicians that access to information is the basis of the job they swore to perform, and therefore they have a duty to fight for it for however many news cycles it takes. via @drjlhazelton.bsky.social

submitted 1 day ago • 0 comments

cc @dingdingpeng.the100.ci (what I think of as "lighting the DAG-signal")

submitted 1 day ago • 0 comments

This confirms (w/better data) the dependence of online gambling on high-volume players. This has/should have public health/strategic implications. FWIW I've seen similar patterns in other industries I've worked with. Addictivity leads to nonlinearity, specially in the low-friction regime.

submitted 1 day ago • 0 comments

This sort of thing should lead to high fives and a round of applause for @natashabertrand.bsky.social: the President getting so angry at your accurate coverage they call for you to get fired should count as a minor Pulitzer prize. Trump's... 1/ via @alexisconran.bsky.social @t0nyyates.bsky.social

submitted 1 day ago • 1 comment

That's my experience as well: you can't change a cognitive architecture w/o impacting politics, and you can't do that without impacting on [self-perceived status. Few people who get to have power will ever take that trade. (Now, LLMs are coded as social inferiors... 1/ via @grimalkina.bsky.social

submitted 2 days ago • 1 comment

Cognitive degradation in the powerful is inherently contagious to actors whose utility functions are more social (depending on favor) than pragmatic (depending on impact on the world). Trump works as a brutally clear classification test. via @t0nyyates.bsky.social @llyfrgellbabel.bsky.social

submitted 2 days ago • 0 comments

An understandable but deadly cognitive blind spot in financial culture/collective processing systems is the fungibility and scalar nature of risk: it's built to deal with in-regime risks but not risks to the regime. (True for climate change, also e.g. authoritarianism.) via @csissoko.bsky.social

submitted 2 days ago • 0 comments

Clarifying [because I <3 all this stuff]: "Singularity" is a feature of the map ---the equations we're using to model the physics--- not of the territory. It's less a hole in the world than a hole in our equations. The article says "some people... 1/ via @drjlhazelton.bsky.social

submitted 2 days ago • 1 comment

Competitive authoritarianism doesn't mean "tanks on the streets." It just needs 10x times the sort of thing GOP already does + ICE goons harassing people in selected polling places + DOJ assistance + a direct refusal to accept specific races lost. It's preventable... 1/ via @jmberger.com

submitted 2 days ago • 2 comments

Do read Leavitt's statement in the article: "flat-out wrong," "leaked by an anonymous, low-level loser in the intelligence community," etc. The damage (symbolic and very real) they are doing to the US intelligence system (for _every_ definition of it) is epochal. via @drjennings.bsky.social

submitted 2 days ago • 0 comments

My shorthand model is "unit cost of bullshit drops from lower than good work to ~0" The first order effect is "much more bullshit" but I suspect leading to a separating equilibrium consistent w/a better-than-ever quality frontier but not w/contemporary scalability assumptions. via @tedunderwood.me

submitted 2 days ago • 0 comments