What we’re reading, April 24, 2026
Wind farm aesthetics, AI's grant application flood, and street votes for housing
Hope your week is going well! Here's what caught our attention:
There are a lot more loose vibes than specific predictions when it comes to AI’s economic impacts these days, which is why I was grateful to see Metaculus, Renaissance Philanthropy, and the Schultz Family Foundation collaborate on a forecasting project on this topic. The Labor Automation Forecasting Hub collects the whole Metaculus community’s predictions on everything from the outlook of specific jobs (nursing seems like it’ll continue to do well) to the outlook of specific US states (Washington state looks like it’ll do okay, despite Microsoft and Amazon’s vulnerability to tech layoffs.) I expect the forecasts to be wrong a lot, but I appreciate the clarity with which they’re being made. — Dylan Matthews
As AI drives down the time & effort cost required to produce scientific grant proposals, funders are likely to be flooded with eligible, reasonable-seeming submissions that strain an already overburdened review system. It seems like we are starting to see the effects. Last week, the European Research Council (ERC) announced stricter application rules for 2027, not mentioning AI specifically but citing a “rapidly increasing number of applications.” The new policy increases the amount of time that unsuccessful applicants must wait before submitting another proposal (you cannot submit in 2027 if you received a “B” score on a proposal in the past two years, or a “C” score in the past three) and limits researchers to one application per year. Last summer, the NIH instituted a less strict but similarly motivated rule capping annual applications at six per investigator. Caps are one option for re-imposing application costs on researchers and encouraging self-screening, but they are a blunt instrument. Some other proposals for dealing with the influx include distributed peer review (which is being piloted by the UK Metascience Unit, and can produce more targeted application costs by scaling an applicant’s review requirements to the number of applications they submitted), tiered review (which could be a valuable tool, especially if combined with soft caps, but also runs the risk of increasing the number of applications by lowering the first-stage requirements), and AI-driven review (which is interesting and should be studied, though I remain wary of turning science into this cartoon). Figuring out the optimal solution is an open, and important, metascientific question. — Jordan Dworkin
In the interest of getting our fair share of aesthetic controversy, we’ve been reading Zane Kasher’s job market paper on wind farm construction and the effect on local residents. While the paper focuses on implications for taxing wind farms - Kashner finds taxes actually help wind farms get built in all but the most sparsely-populated areas, since it’s easier for developers to compensate locals for the downside - he also finds a 12% average drop in home prices within 3 miles of a wind farm. Since most auditory effects are within 2 miles, we wondered what the academic literature has to say about the aesthetics of wind farms. Beyond the entangled moral and political judgments that influence perception, there’s a few interesting takeaways. First, aesthetic responses vary a lot by geography (Spain and Sweden seem not to mind wind farms; Korea and Germany are less enthused on the aesthetics). Interestingly, some of that variation might trace back to the literal background - Molnarova et al. 2012 and Lothian 2008 both find that people like the look of wind farms better in degraded landscapes than in more scenic ones (an interesting wrinkle for the land-value externalities in Kashner’s wind paper, if more expensive homes cluster in baseline-prettier environments). Offshore, the question is much simpler: distance from shore dominates everything else (Ladenburg 2009; Cranmer et al. 2022 find indifference sets in beyond ~10 nautical miles). Finally, blinking aviation obstruction lights are a big - but pretty cheaply avoidable - driver of annoyance (aircraft-detection lighting systems can keep turbines dark ~95% of the night, and costs are estimated at just $1-2m for a single wind farm). — Willow Latham-Proenca
Our friends at the Centre for British Progress (with Labour Together) are out with a new paper on street votes—a mechanism that lets residents of a given street collectively hire an architect, agree on an upzoning plan, and vote to allow new housing, with value capture flowing to the consenting homeowners. The bottom-up mechanism design is interesting precisely because it doesn’t require preemption: you can imagine it mattering most in jurisdictions where state overrides have proven elusive, like Long Island in New York. In a galaxy brain sense, the most successful “street vote” of all time may be Senakw in Vancouver, where the Squamish Nation—exercising sovereign planning authority over ~10 acres near downtown—voted for thousands of rental units that will house tribal members and generate billions in revenue. They proved that single family zoning in high-opportunity areas is often NOT profit-maximizing, contra the simplest “pop” version of the Homevoter Hypothesis. With a tribal voting mechanism to allow it, the Squamish Nation’s people voted for abundance. UK Street Votes won’t be allowed to unlock as much density as Senakw, but Senakw shows how the incentives and voting could play out. — Alex Armlovich
Vincent Rollet’s MIT job market paper on zoning and the dynamics of urban redevelopment — which won Best Student Paper at the 2025 European UEA meeting and lands him at Stanford and then UChicago — is getting attention from actual city planning staff beyond academia. Rollet built a parcel-level panel of 833,000 NYC lots and estimates a dynamic spatial equilibrium model that captures something the static literature misses: redevelopment is slow, lumpy, and driven by fixed costs that rise sharply with existing building size. The fact that planners are paying attention matters: this is the kind of structural model that can discipline a rezoning’s projected supply impact in ways the usual back-of-envelope feasibility studies cannot. — Alex Armlovich
On a new episode of the Works in Progress podcast, Ruxandra Teslo, Ben Southwood and I chatted about everything you wanted to know about clinical trial reform — from why ethics reviews are so inefficient to how Australia has made it simpler and faster to run earlier stage trials. We also discuss our wishlist for policy reforms and some of the worst ideas we’ve heard. — Saloni Dattani
Separately, Niko McCarty and I wrote a round-up of news you might have missed in biotech and medicine. For example, did you know that organ donations have become vastly more efficient because of improvements in preserving organs from deceased donors? That semaglutide failed to slow the progression of Alzheimer’s in two large trials? Or that a transformative new drug to treat late stage pancreatic cancer succeeded in phase 3 trials, and will likely be expanded for earlier treatment and for other cancers too? Now you do! There’s much more in our post. — Saloni Dattani
And Matt is on holiday this week!
In grantee and team news:
The Roots of Progress 2026 conference has been announced. Coefficient Giving is a sponsor and the conference brings together people across the progress and abundance movements. Applications are open through May 31.
Progress Ireland’s Seán Keyes shared new ideas for Irish housing and infrastructure policy.
Andrew Stern of Open New York has been hired as press secretary for NYC’s Department of Housing Preservation & Development.
And in case you missed it, we published a digest of 79 things we read about abundance, a look back at the key readings and announcements from January through March.