IFP

New Factory Settings piece today, also cross-posted in @Briefing_Book:

Take the Dan pill

Looking for an even more considered take on the state of modern systems neuroscience? Read my piece in Macroscience, IFP’s fantastic (meta)science newsletter! I explain why systems neuroscience is like the Matrix, delve more into state-of-the-art neuroscience tools, and explain why sustained team science efforts are essential if we someday hope to upload our brains to the cloud. macroscience.org/p/how-close-ar…

Very useful measure of pre-NEPA delays. Federal agencies measure NEPA timelines from Notice of Initiation (NOI) to Record of Decision (ROD). But sponsors and agencies do years of work *before the NOI. This is widely known but rarely measured. x.com/ThomasHochman/…

Leave it to Aidan to immediately find one of my favorite graphics from the report. Indeed, NEPA timelines are a great deal longer than many realize. We usually count from the “notice of intent”… but agencies increasingly spend years on permitting *before* issuing that notice!

@owl_posting is right that biosecurity spending too often anchors on countermeasures for the last threat, rather than preparing us for the next one. And it's a false choice! Instead, pathogen-agnostic and platform approaches can help us respond to known knowns, known unknowns, unknown unknowns — take your pick.

sometimes you meet someone and you're like "this is going to be amazing" - had deeply immense joy editing with @JassiPannuMD this past week and cannot wait for y'all to see what is cooking

Big ideas coming out of the @IFP Economics of Ideas, Science and Innovation PhD course ifp.org/economics-of-i…

Extremely under told story: NEPA defenders call the law the "Magna Carta of Environmental Law". But Nixon and the 1969 Congress had virtually no idea what they were creating when they passed NEPA. The law was never meant to be a vehicle for litigation or require 1,000+ page reviews. The framers of NEPA wanted to make major changes to protect the environment but they had little to no grasp on the lawsuits or the project-by-project review they were creating. Congress focused their debate on the creation of the Council on Environmental Quality (CEQ) but they barely discussed section 102(2)(C) of NEPA that created the requirement for Environmental Impact Statements (EIS) that now require agencies to produce 1,000+ page reviews. Congress seems to have had no clue that the bill they were passing would become a vehicle for citizen-suits under the APA. NEPA itself has no judicial review section and senior negotiators do not appear to have debated the potential for the law to allow lawsuits against every reviewed project. The bill passed the Senate in July 1969 by unanimous consent — it appears no senator was aware of the bill's far reaching consequences. In December 1969 the bill was rushed through conference so members could go home for Christmas, passing the amended version through Senate without a roll call vote. In a retrospective Professor Lynton Caldwell said: "The disruptive effects of the Act on the business-as-usual economy do not appear to have been foreseen by the Congress or by those interests most likely to have been affected." Nixon's White House seems to have completely missed the costs of the EIS requirement. No WH agencies flagged the costs of the EIS requirement when reviewing the bill. NYTimes coverage of NEPA's passage barely discussed the EIS provision, focusing on an "anti pollution" headline that was inaccurate to the focus of NEPA. The EIS requirement was only mentioned in the second to last paragraph of a companion article. Judicial review was not discussed at all. And nobody foresaw or debated the possibility that CEQ would claim authority to issue binding regulations and expand NEPA far beyond the scope of NEPA's authorization. Categorical exclusions, environmental assessments, rules for community engagement, cumulative impacts, the meaning of major federal action were all invented by CEQ despite Congress never giving them formal authority to issue binding NEPA regs. Far from being the Magna Carta of Environmental Law NEPA is a poorly drafted law that Congress didn't understand before it was passed and has mostly been built out by regulation. fhwa.dot.gov/highwayhistory…

A shame that this will destroy the robust US shipbuilding industry

manufacturing requires inputs that are often imported?

It is a special kind of insanity for a city to make it illegal to build buildings that are the same character as the existing neighborhood, and insist that anything new be smaller than what came before

I've studied past IRB reform efforts. I am convinced nothing but allowing IRB choice for investigators will work. This would introduce market discipline among IRBs. The great news is I & @matthewesche wrote how to implement this in detail for @IFP.

So appreciate @RepJoshHarder's support for our X-Labs framework to seed new institutions of science and glad to hear @NIHDirector_Jay sounds supportive 👀

Bernie is AGI-pilled. Good. Many to follow. But this response is completely inappropriate. A moratorium on data centers isn't the answer! 95% of this video is him listing quotes from people about how AI will be a huge transformation and very dangerous. True. Imagine a similar call before the Agricultural Revolution (I know, I know, just run with the thought experiment for a moment.): - "This will be the most disruptive change to our way of life, ever" - "We'll become tied to the land, irreversibly changing our social structures" - "People's health may actually grow poorer, due to the loss of nutrient diversity" - "Agriculture will support large cities, which will support large armies and large-scale war" - "Diseases we've never experienced will decimate our population" - "Our societies will become highly unequal" - ... Yes, true! Alright then, what shall we do? A sowing moratorium? Those who sowed eventually outcompeted those who did not. (Yes, even the Mongols and others who, for a while, were an exception.) Similar calls could've been made during the Industrial Revolution, based on similarly real concerns. Those who industrialized outcompeted those who did not. Make no mistake: nations that stunt their intelligence and restrict their labor supply -- potentially by orders of magnitude -- will be outcompeted by those that do not. "Outcompeted," in each of the prior cases, is a euphemism. Weak nations don't have it easy in this world. This is not to say that we should just let it rip "because China." But think of the concrete thing being proposed: a bill to stop building AI data centers in the US. Nowhere else. What will this look like if it succeeds? - The US becomes compute-poor - China overtakes it - Then the scary AI stuff starts happening, first in China, then everywhere else - But now the US has no way of dealing with it at the source (and neither does any other democracy) - The US is degraded to the role the EU has today: expressing concern, and wishing, for its own survival, that her powerful neighbors display more wisdom than ambition The problem is real. But the answer is not a moratorium on US data centers. Unfortunately, a real answer would require more nuance than Bernie can fit into this kind of video, starting with: - Actually funding CAISI, with ~$75M more than the pennies it got for FY26 - Setting up a large R&D program for AI verification, to make sure any regulatory actions the US wants to take in the future can be matched, and verified, in China - Strengthening export controls on AI chips and SME, because you don't want the most transformative technology ever developed to spawn outside democratic control - Real actions on biosecurity: DNA synthesis screening (Cotton and Klobuchar are already on the ball), metagenomic sequencing, 10x the PPE strategic national stockpile - Making the few leading AI labs spend some portion of their compute on AI alignment and control - Launching a lare prize competition for achieving interpretability milestones - Funding serious research on post-AGI economics (super vague but I mean, better to have it than not). The UBI trials were a start, but they fall far short of what we need - Hardening open-source critical infrastructure code against AI-accelerated cyberattacks, - Funding automated vulnerability patching for critical infrastructure, to keep pace with AI-accelerated offensive cyber - Making sure data centers can be built in the US with cheap energy, because if the US can't build them, others will fill the gap - (and many, many more things) There are answers, there are things to do. This stunt is not part of the winning strategy.

This is supposed to be the “Future Land Use Map” but it coincidentally looks exactly the same as the “Current Land Use Map.”

Per @AndyMasley's new tool, alfalfa growers in Colorado alone use 16 times as much water each year as all the data centers in the United States.

tariffs are bad. there is no plan

Neat profile of Taiwan's "Industrial Technology Research Institute", a government-funded lab which helped spin out TSMC and much of Taiwan's chip industry. One of the most impressive industrial policy plays of all time. asteriskmag.com/issues/13/the-…

Great post by @BenSchifman on a blind spot in the SPEED Act that passed the House. NEPA reform needs to limit vacatur (court remedy at final judgment) AND set limits on preliminary injunctions (court delays at the beginning of litigation). x.com/ThomasHochman/…

@BenSchifman makes an important point here: the SPEED Act claims to fully solve the litigation delays caused by NEPA, but without changes, it won't succeed. The SPEED Act: a) doesn't solve preliminary injunctions, a major source of delay (while being so harsh at the final decision stage that it can't pass the Senate) b) relies on judicially-imposed deadlines, which haven't worked in the past- agencies won't listen