Brian Hartley

Macroeconomist. Econ Ph.D. @NSSRNews. Interests - Macro, Computational, Ecological. 🐠 🐕🏔️

[Brian Hartley @BrianJHartley]

@alz_zyd_ Which parts of statistics? Just curious

[Anthropic @AnthropicAI]

Introducing Project Glasswing: an urgent initiative to help secure the world’s most critical software.

It’s powered by our newest frontier model, Claude Mythos Preview, which can find software vulnerabilities better than all but the most skilled humans.
https://t.co/NQ7IfEtYk7

[Brian Hartley @BrianJHartley]

@GuthmannR I think the interesting question is when this argument doesn't apply - do we need markets to determine how valuable drinking water or aid in an emergency is in an individual's utility function? No. There is a part of the spectrum where the local information problem doesn't bite.

[Eldritch Brick - BE NOT AFRAID @FuckKoroks]

If wrestling is fake, explain this

[Brian Hartley @BrianJHartley]

@charlescwcooke Biden’s plan wrote down ~$400–500B in NPV — a one-time balance sheet adjustment. Trump unilaterally raised ~$180B per year in new taxes via IEEPA.
Depending how you discount the tariffs were nearly an order of magnitude larger in flow impact.

[Brian Hartley @BrianJHartley]

@BergsenBri66275 @noodlesli2016 @Snegglepuss85 @erikbryn Mostly true, but “only per-capita matters” assumes everything scales smoothly and privately. Public capital and local institutions often don’t - a lot of times there are thresholds or other possibilities that are opened by simply having a larger economy.

[Brian Hartley @BrianJHartley]

@RyanGirdusky This isn't a good counterfactual. Social Security isn’t a personal “account” with an S&P-like “yield.” It’s social insurance + a current transfer: benefits to current retirees are related to current output ultimately. You're ignoring one side of the ledger here.

[Brian Hartley @BrianJHartley]

@Zach_Carney16 @TheStalwart But you can even ask another AI session "what language or packages should be used? how should this project be organized? What should be the outputs/visualizations? How to validate results?" Having a clear plan at the start can be very helpful.

[Brian Hartley @BrianJHartley]

@Zach_Carney16 @TheStalwart Just grab a cup of coffee and come up with a quirky/simple idea that seems fun or interesting to you. If you have a claude sub, open up a claude dialogue box in one window and have it guide you through how to use the claude code CLI in baby steps. And then build a little model!

[Brian Hartley @BrianJHartley]

@TheStalwart That's great! As someone who does modeling and stats for a living, I've found the same to be true. Created many silly repos and models to nowhere, but my understanding has grown more in a year than the rest of my career! You can really explore and play so much faster.

[Brian Hartley @BrianJHartley]

@MattBruenig @mean_field_zane @TheStalwart I wouldn't wholly defend a deserts theory either. My point is definitional: requiring determinism is non-standard and seems to commit you to denying any ethical mapping from choice under uncertainty to outcomes, does it not? That collapses “just deserts” almost by definition.

[Brian Hartley @BrianJHartley]

@MattBruenig @mean_field_zane @TheStalwart So any departure from return determinism nullifies just-desserts? Isn't what matters in the first-order the EV behavior ex ante not the realization ex post? Seems too strong

[Brian Hartley @BrianJHartley]

@S_Surprenant There are also other ways you can make the point w/o the NK model (just enumerating the channels, VAR lit re: price puzzles, etc..).

[Brian Hartley @BrianJHartley]

@BobMurphyEcon (not taking a position here, I just think the discussion is overly simplistic)

[Brian Hartley @BrianJHartley]

@BobMurphyEcon But still, adding variables is not a free lunch guaranteed to resolve identification problems. The issue is still empirically debatable at least, so pretending it’s just stupid dilettantes who can think monetary policy can have inflationary effects is wrong-headed. 3/3

[Brian Hartley @BrianJHartley]

@GeorgeSelgin An obvious thing, but unfortunately, progress in this area is remarkably limited. Most alternatives to typical DSGE models, like PK-SFC models, are mostly or entirely demand-led. Their usefulness is primarily (only?) in a setting where resource constraints do not bind.

[Brian Hartley @BrianJHartley]

@GeorgeSelgin Taking this criticism to heart, the move in that community (or just non-neoclassicals, broadly) should be, in part, to build models that articulate the interaction of effective demand and realistic supply constraints. And empirically test those mechanisms.

[Brian Hartley @BrianJHartley]

@alz_zyd_ Hard to disagree with any of this. As an example, Neoclassical models can be useful, obviously so. What is objectionable is the notion that those foundations are *required* for usefulness.