How these papers were chosen among the week's 129 papers

All 129 papers of the week whose main category is in this field were read, title and start of the abstract, and 25 were kept. They were then ordered by what can be counted: how many research profiles of our readers they come close to, whether a rising topic of the field includes them, and how many categories they were posted in. The abstracts of the first 25 were read in full to choose these, no more than two from one category and one per team.

The 2 numbers of the text were checked against the abstracts. 24 statements on what the papers found were looked for word for word in their abstracts.

Sources: arXiv.

When free water connections make economic sense #

Charging for public infrastructure can seem like an unavoidable compromise: fees help cover costs but exclude households with little money. A study of a new piped drinking water network in rural Bangladesh finds that this logic can run backwards. Researchers elicited households’ willingness to pay, with truthful answers encouraged by a real opportunity to connect, and combined those responses with project cost data. The price that minimized cost per connected household was zero.

The explanation is economies of scale: attracting more connections spreads network costs widely enough to outweigh the revenue forgone by waiving fees. This does not mean that free connections make infrastructure costless. It means that maximizing cost-effectiveness need not conflict with expanding access. The authors identify two ingredients behind that result: substantial economies of scale and demand that responds strongly to prices near zero.

After floods, carmakers add backup rather than retreat #

A flooded factory need not become an abandoned factory. Using plant-level production data, a manually verified flood record and a staggered difference-in-differences design, researchers find that multinational carmakers respond by duplicating some production elsewhere rather than permanently reducing output at the flooded site. Even more than five years after a flood, an engine is 19 percentage points more likely to be produced at multiple plants, usually with the additional production abroad.

That response depends on having a relatively cheap route to redundancy: duplication occurs where another company plant already makes the same engine family. Vehicle-model duplication is temporary, while transmission programs show no duplication. The broader lesson is that adaptation to climate risk may add overlapping production capacity rather than simply move factories out of danger. Evidence suggesting more duplication where insurance is expensive and public flood protection is absent remains suggestive, not conclusive.

An AI advantage for software defenders #

Does more powerful vulnerability detection help defenders get ahead of attackers? A preprint examines Project Glasswing, which gave selected large US technology firms early access to Claude Mythos Preview. Using publicly disclosed software vulnerabilities, the author finds that participating firms more than doubled their disclosures relative to Chinese and Russian technology firms without access, while continuing to fix reported flaws as quickly as before. External researchers reported fewer vulnerabilities in participating firms’ software and more elsewhere, rather than leaving the market.

The additional disclosures did not come with a detected increase in the number of exploited vulnerabilities. The share exploited within a month of disclosure fell by about two thirds, largely because fewer were already being exploited when disclosed. These findings suggest that AI can strengthen defensive discovery, while also showing how restricted access could distribute security gains unevenly across countries. Public disclosure records, however, cannot establish the full picture of undisclosed flaws and attacks.

Public attention can turn violence into a call for help #

Highly salient intimate partner femicides in Italy were followed by nationwide increases in help-seeking, according to a study using newly assembled case-level data. An exceptionally salient case also preceded a marked rise in police reports of violence against women. The researchers find no evidence of an increase in femicides, a result inconsistent with the idea that these events necessarily provoke lethal imitation.

The comparisons matter as much as the headline result. Similar mobilization did not follow several other kinds of killing or lower-salience cases, suggesting that attention alone is insufficient: its intensity, the relationship between victim and perpetrator, and the gendered nature of the violence all matter. Public attention may therefore support prevention efforts, although increased help-seeking and reporting do not themselves demonstrate a subsequent reduction in violence.

Traffic penalties deter—and encourage blame shifting #

A large increase in speeding sanctions in Queensland, Australia, offers a way to compare monetary fines with accumulating penalty points. The study finds only small estimated reductions in reoffending among drivers facing higher fines alone. When higher fines came with additional penalty points, reoffending fell by 4.6 percentage points.

But stronger sanctions also changed how people responded to enforcement: higher demerit points increased the nomination of other drivers as responsible for the offence. That pattern is consistent with strategic blame shifting, though it does not establish that every nomination was dishonest. The policy lesson reaches beyond traffic enforcement: penalties can change both the underlying behaviour and efforts to redirect liability.

Owning a mine is not the same as securing its output #

A study of the lithium supply chain asks whether restricting Chinese ownership abroad would secure mineral inputs for other countries. Using global mine and refinery data, the author estimates a model linking ownership, ore costs, spot-market sales and competition. In the model, divesting Chinese stakes in overseas producers increases spot ore supply but leaves lithium output nearly unchanged. Continued trade still allows Chinese refiners to obtain ore.

An export ban has a different consequence: it reduces global lithium supply because refining capacity elsewhere cannot fully replace lost Chinese production. The distinction matters for industrial policy. Changing ownership does not automatically reserve material for domestic users, while restricting trade can sacrifice output when processing capacity is concentrated elsewhere. These are model-based policy results, with the export-ban cost tied to existing refining capacities.


Across these preprints, policy effects depend on the alternatives people and firms can actually use. Households join a network when prices fall; manufacturers duplicate production when compatible plants exist; researchers redirect their search; drivers may redirect liability. The common lesson is to look beyond the immediate target of a policy and ask how costs, capacity and opportunities to adapt change the outcome.

Was this letter useful to you?

Written by AI from the 184 papers of the week in this field. Most are preprints, not yet peer reviewed: the text says what their authors report.

Get one letter by email every week with a free account, all of them with Premium. Create a free account