@mouthfoot.bsky.social
Forgot the sauce for the chart: cepr.org/voxeu/column...
The economic consequences of war
Much of the literature on the economic effects of war has focused on specific countries or regions, relied on small samples, or spanned only brief periods. This column investigates the economic effect...
cepr.org
October 5, 2026 at 12:31 PM
Forgot the sauce for the chart: cepr.org/voxeu/column...
At risk of being the Bluesky well ackhsually meme dude, here's a trivial counterexample: imperialist/interventionist wars are good for GDP.
Ceteris paribus all good things do indeed correlate with economic growth but that ceteris paribus is load bearing. You definitely *can* have "bad GDP growth".
Ceteris paribus all good things do indeed correlate with economic growth but that ceteris paribus is load bearing. You definitely *can* have "bad GDP growth".
October 5, 2026 at 12:30 PM
At risk of being the Bluesky well ackhsually meme dude, here's a trivial counterexample: imperialist/interventionist wars are good for GDP.
Ceteris paribus all good things do indeed correlate with economic growth but that ceteris paribus is load bearing. You definitely *can* have "bad GDP growth".
Ceteris paribus all good things do indeed correlate with economic growth but that ceteris paribus is load bearing. You definitely *can* have "bad GDP growth".
... Pretty sure these people would change their minds pretty quickly if you could get through their skulls that _they are_ the subhumans that RW politicians are talking about. It's whats happening in the US to some extent.
October 5, 2026 at 10:49 AM
... Pretty sure these people would change their minds pretty quickly if you could get through their skulls that _they are_ the subhumans that RW politicians are talking about. It's whats happening in the US to some extent.
... The "deception" part however happens when these very same people get tricked into voting for people/parties who expouse majoritarian views but actually want to enslave 60% of the population and make glue out of another 30% (conservatively) ... [2/3]
October 5, 2026 at 10:47 AM
... The "deception" part however happens when these very same people get tricked into voting for people/parties who expouse majoritarian views but actually want to enslave 60% of the population and make glue out of another 30% (conservatively) ... [2/3]
I'm pretty sure that most people hold an understanding of "democracy" that is indistinguishable from the majoritarianism that had been already thorougly refuted by Tocqueville's time. It's a bad view, but it can be understood how you land on it if you didn't think particularly hard about it... [1/2]
October 5, 2026 at 10:46 AM
I'm pretty sure that most people hold an understanding of "democracy" that is indistinguishable from the majoritarianism that had been already thorougly refuted by Tocqueville's time. It's a bad view, but it can be understood how you land on it if you didn't think particularly hard about it... [1/2]
I dont think people (or even trump, as much of an idiot might be) think about existing mortgages when referring to the impact of yields on mortgages. Then again I guess xkcd caveat applies.
October 1, 2026 at 3:24 PM
I dont think people (or even trump, as much of an idiot might be) think about existing mortgages when referring to the impact of yields on mortgages. Then again I guess xkcd caveat applies.
... In the end this is the reward hacking mode as AlphaGo going bananas or the infamous CoastRunners experiment.
But the end result of repeated reward hacking behaviour is not "paperclip apocalypse" it's just "ok so using this is actually non-viable economically because of liability".
But the end result of repeated reward hacking behaviour is not "paperclip apocalypse" it's just "ok so using this is actually non-viable economically because of liability".
September 24, 2026 at 6:35 AM
... In the end this is the reward hacking mode as AlphaGo going bananas or the infamous CoastRunners experiment.
But the end result of repeated reward hacking behaviour is not "paperclip apocalypse" it's just "ok so using this is actually non-viable economically because of liability".
But the end result of repeated reward hacking behaviour is not "paperclip apocalypse" it's just "ok so using this is actually non-viable economically because of liability".
... Because this takes almost ~time than actually pulling the data yourself ...
This becomes even more fraught when you consider how pervasive "gray area" scraping is on the web and how badly agents navigate the gray areas.
... [3/n]
This becomes even more fraught when you consider how pervasive "gray area" scraping is on the web and how badly agents navigate the gray areas.
... [3/n]
September 24, 2026 at 6:29 AM
... Because this takes almost ~time than actually pulling the data yourself ...
This becomes even more fraught when you consider how pervasive "gray area" scraping is on the web and how badly agents navigate the gray areas.
... [3/n]
This becomes even more fraught when you consider how pervasive "gray area" scraping is on the web and how badly agents navigate the gray areas.
... [3/n]
You want to be able to ask "make a report on yearly Australian Medicare costs" without risking incurring legal liability...
... But you don't want to have to read the website/API and write a firewall limiting the web access 3/4 paths, which is what proper strict security would look like... [2/n]
... But you don't want to have to read the website/API and write a firewall limiting the web access 3/4 paths, which is what proper strict security would look like... [2/n]
September 24, 2026 at 6:19 AM
You want to be able to ask "make a report on yearly Australian Medicare costs" without risking incurring legal liability...
... But you don't want to have to read the website/API and write a firewall limiting the web access 3/4 paths, which is what proper strict security would look like... [2/n]
... But you don't want to have to read the website/API and write a firewall limiting the web access 3/4 paths, which is what proper strict security would look like... [2/n]
These situation are classic RL failure modes, just OpenAI is desperately trying to make it sexy.
You can't really firewall it properly *without making the agent a lot less useful*.
You want to give it normal, browser-like web access, and something resembling cURL, to scrape and read APIs [1/n]
You can't really firewall it properly *without making the agent a lot less useful*.
You want to give it normal, browser-like web access, and something resembling cURL, to scrape and read APIs [1/n]
September 24, 2026 at 6:14 AM
These situation are classic RL failure modes, just OpenAI is desperately trying to make it sexy.
You can't really firewall it properly *without making the agent a lot less useful*.
You want to give it normal, browser-like web access, and something resembling cURL, to scrape and read APIs [1/n]
You can't really firewall it properly *without making the agent a lot less useful*.
You want to give it normal, browser-like web access, and something resembling cURL, to scrape and read APIs [1/n]
A novel recently which has AI footprints all over it "c'était ca ou mourir" and indeed it was a massive hit with public and critics alike... I figure because the theme of the novel dovetail with this style almost perfectly... However if you read french LinkedIn at this point it sounds LinkedInese.
September 23, 2026 at 4:50 AM
A novel recently which has AI footprints all over it "c'était ca ou mourir" and indeed it was a massive hit with public and critics alike... I figure because the theme of the novel dovetail with this style almost perfectly... However if you read french LinkedIn at this point it sounds LinkedInese.
Just check out the new multi-awarded "big summer book" in french that has been denounced as AI "c'était ça ou mourir" iirc... kinda reads like default claude indeed. curiously works very well with the theme of the book, but "normal" written French sentence structure is utterly different...!
September 23, 2026 at 4:46 AM
Just check out the new multi-awarded "big summer book" in french that has been denounced as AI "c'était ça ou mourir" iirc... kinda reads like default claude indeed. curiously works very well with the theme of the book, but "normal" written French sentence structure is utterly different...!
Ternaries are a thing in Italian/French/English AFAIK.
In Italian it sounds like someone who grew up mostly reading videogame reviews.
In French it writes in a rythmic quick firing prose that does not correspond to any actual French lit I'm aware of, sound like translated beatnik English.
In Italian it sounds like someone who grew up mostly reading videogame reviews.
In French it writes in a rythmic quick firing prose that does not correspond to any actual French lit I'm aware of, sound like translated beatnik English.
September 22, 2026 at 9:53 PM
Ternaries are a thing in Italian/French/English AFAIK.
In Italian it sounds like someone who grew up mostly reading videogame reviews.
In French it writes in a rythmic quick firing prose that does not correspond to any actual French lit I'm aware of, sound like translated beatnik English.
In Italian it sounds like someone who grew up mostly reading videogame reviews.
In French it writes in a rythmic quick firing prose that does not correspond to any actual French lit I'm aware of, sound like translated beatnik English.
Dunno if this applies to your specific case but after trying to grow thyme unsuccessfully for years in med country, the one weird trick I found is that after transplanting it and for the entire first year it *hates* hot strong sun and you have to shadow it. After that it becomes nigh indestructible.
September 22, 2026 at 4:35 PM
Dunno if this applies to your specific case but after trying to grow thyme unsuccessfully for years in med country, the one weird trick I found is that after transplanting it and for the entire first year it *hates* hot strong sun and you have to shadow it. After that it becomes nigh indestructible.
Niche delicious Mediterranean recipe: dip the leaves in flour batter and deep fry. Amazing stuff.
September 22, 2026 at 4:31 PM
Niche delicious Mediterranean recipe: dip the leaves in flour batter and deep fry. Amazing stuff.
Basically the caveat is "you need to get the tumor first to vaccinate against it and the vaccine only works for you" which is... a pretty humongous caveat in so many ways.
September 20, 2026 at 5:27 AM
Basically the caveat is "you need to get the tumor first to vaccinate against it and the vaccine only works for you" which is... a pretty humongous caveat in so many ways.
Going to wear my annoying wonk hat for a second and say that given it only works on resected tumors and with 50pct effectiveness at that, and it's personalised medicine to boot, this is still very low effectiveness compared to "just buy more fucking mri machines" in healthy lifespan / $.
September 20, 2026 at 5:26 AM
Going to wear my annoying wonk hat for a second and say that given it only works on resected tumors and with 50pct effectiveness at that, and it's personalised medicine to boot, this is still very low effectiveness compared to "just buy more fucking mri machines" in healthy lifespan / $.
I think the ooze is "meld with last thing you touched" ooze
September 19, 2026 at 9:13 PM
I think the ooze is "meld with last thing you touched" ooze
Back in the day guys doing cartoon adaptations in Italy were basically constantly huffing paint if you have to judge by the faithfulness of their output (dont get me started on anime!), but Wiki says Splinter was a dude even in the 1984 US cartoon series.
September 19, 2026 at 9:13 PM
Back in the day guys doing cartoon adaptations in Italy were basically constantly huffing paint if you have to judge by the faithfulness of their output (dont get me started on anime!), but Wiki says Splinter was a dude even in the 1984 US cartoon series.
In the end what is interesting is mostly the mathematical chops/methodology, if novel they might be used for other stuff... The fact that they're playing with boussinesq equation that everyone in the field knows to be ass tells me that actual practical fluid dynamics is not the main concern.
September 8, 2026 at 7:16 PM
In the end what is interesting is mostly the mathematical chops/methodology, if novel they might be used for other stuff... The fact that they're playing with boussinesq equation that everyone in the field knows to be ass tells me that actual practical fluid dynamics is not the main concern.
Basically the big ass finding would have been proving (A) or (B) of the Millenium Problem, ie NS is good and stable and smooth. Instead they proven (C) (D), which are exclusive with (A)/(B), and it mostly boils down to engineers were right and Navier Stokes is ass.
September 8, 2026 at 7:12 PM
Basically the big ass finding would have been proving (A) or (B) of the Millenium Problem, ie NS is good and stable and smooth. Instead they proven (C) (D), which are exclusive with (A)/(B), and it mostly boils down to engineers were right and Navier Stokes is ass.
Navier-Stokes has a singularity (infinite local speed) under finite boundary conditions and time, this likely means just NS is a bit wrong in turbulence, from my CFD memories practitioners dont use/believe in NS in turbulence anyway and use rules of thumb and calibrated specialised models.
September 8, 2026 at 7:08 PM
Navier-Stokes has a singularity (infinite local speed) under finite boundary conditions and time, this likely means just NS is a bit wrong in turbulence, from my CFD memories practitioners dont use/believe in NS in turbulence anyway and use rules of thumb and calibrated specialised models.
Its just a disproof, ie they found a singulairty under finite boundary conditions, so by itself its kinda useless and sorta bad news... buuuut when the methodology gets published they might have something interesting in the methodology (much like the erdos unit distance problem)
September 8, 2026 at 6:52 PM
Its just a disproof, ie they found a singulairty under finite boundary conditions, so by itself its kinda useless and sorta bad news... buuuut when the methodology gets published they might have something interesting in the methodology (much like the erdos unit distance problem)