Quick take: Longer observation is doing useful work this morning: it is tempering a gene-therapy claim, exposing the physical limits of a moon-bound launch system, and forcing institutions to show whether their safeguards survive contact with reality. Meanwhile, flood warnings matter only when people can act on them.
Four-year data weaken the headline case for a Huntington’s gene therapy
New 48-month results from uniQure’s experimental Huntington’s gene therapy AMT-130 did not show a statistically significant slowing of disease progression on the trial’s main composite measure, Reuters reported on 29 September. The 12 high-dose participants showed an estimated 44% slowing versus an updated external control group, down from the 75% estimate at 36 months that supported the company’s US approval application. UniQure’s own update emphasises continued slowing and a more encouraging functional-capacity measure, so the result is mixed rather than a simple failure. The promise remains substantial: Huntington’s has no approved disease-modifying treatment. The limitation is equally important—this is a small, non-randomised comparison vulnerable to changes in the control group and survivor bias. Patients should not alter care around market headlines. The practical question for clinicians, regulators and families is whether full data show a consistent benefit across measures, doses and follow-up—not whether one percentage sounds dramatic.
Possible Alkemata article: When Longer Follow-Up Changes a Breakthrough Story
Floods across South and Southeast Asia turn warning capacity into a life-safety test
Storms and floods had killed at least 56 people in India and 14 in Nepal by 29 September, while Thailand reported eight deaths and disruption to more than 580,000 homes, Reuters reported that day. Nepalese authorities warned of possible flash floods in 49 of 77 districts; thousands of Bangkok residents moved into evacuation centres. The promising element is not the hazard itself but the use of district alerts, shelters, boats and mobile kitchens before waters peak. The danger is that a technically correct forecast may still fail people who cannot receive it, understand it or leave safely—especially older residents, disabled people and households without transport. Climate attribution for this specific event requires analysis; the immediate facts are extreme rainfall, saturated systems and exposure in flood-prone cities and valleys. If you are in or travelling through an affected area, use official local warnings, avoid moving water, identify higher ground and keep medicines, power and documents ready. Elsewhere, test whether your own alert chain ends in a feasible action.
Possible Alkemata article: A Warning Is Only Real When Someone Can Act
Google asks Europe’s court to redraw the boundary between competition and privacy
Google filed two challenges at the EU General Court on 28 September against Digital Markets Act orders requiring Android interoperability for rival AI services and access to anonymised search data for competing search providers, Reuters reported on 29 September. This is a material legal step beyond the Commission decisions covered elsewhere: a court will now test whether the remedies are proportionate. The Commission says its July decisions use layered anonymisation and privacy safeguards; Google argues highly personal queries and Android security could still be exposed. More access could reduce dependence on a single search and mobile ecosystem, but weak anonymisation would shift the cost to users whose queries reveal health, relationships or beliefs. Individuals cannot adjudicate the engineering claims, but they can limit retained search history, review account deletion settings and avoid assuming “anonymous” means impossible to re-identify. Watch for published technical safeguards and independent testing, not only courtroom assurances from either side.
Possible Alkemata article: Can Search Data Be Shared Without Sharing the Searcher?
Starship reaches orbit and deploys a satellite—but a propulsion failure cuts the test short
SpaceX’s uncrewed Flight 14 reached orbit and completed Starship’s first commercial satellite deployment on 28 September, but an upper-stage Raptor engine shut down unexpectedly and a later propulsion malfunction ended the planned ten-hour flight several hours early, Reuters reported on 29 September. That combination matters: the test demonstrated useful capability while also revealing a failure mode relevant to endurance and mission completion. Iterative flight testing is promising precisely because it finds problems before people fly. The uncertainty is schedule pressure. NASA is developing Starship HLS to carry astronauts between lunar orbit and the surface, with a crewed landing targeted for 2028; repeated redesign or qualification work could narrow the margin. The public should resist both “success” and “disaster” labels until SpaceX identifies the cause. The actionable standard is whether the investigation produces a disclosed corrective action, a repeatable ground test and a later flight that closes the same failure—not merely another launch.
Possible Alkemata article: What Counts as Progress in a High-Risk Flight Test?
