Reopened, not solved: five questions science pried back open this week

Here is what I took away from this week: five stories, five unrelated fields, and every one of them is an attempt to reopen a question that was considered closed. Not solved. Reopened. That distinction is the whole point, and it is why I want to walk you through them.

A kidney was chilled below freezing without freezing, then transplanted

Organ transplantation runs on a clock. From the moment an organ leaves the body it starts to deteriorate, so surgeons have a window of a few hours. Miss it and the organ is no longer usable. Everything about transplant logistics — the flights, the matching, the phone call in the middle of the night — exists because of that clock.

This week, researchers reported transplanting a supercooled kidney into a pig. Supercooling means holding something below zero degrees without letting ice crystals form, and ice is exactly what has always ruined organs at low temperatures. MIT Technology Review called the result a landmark achievement.

Now the honest part. This was a pig. Nobody has done this in a human, and the gap between a pig kidney and a person waiting on a transplant list is not a small one. But if it holds up, the thing that changes is not the surgery. It is the clock. Medicine that has always been chasing a deadline would get room to breathe.

AI has started designing the next generation of medicines

Drug development is brutally expensive and slow. Years of work, enormous cost, and most candidates never reach a patient. Biologic drugs — medicines built from biological material, such as antibody therapies — are especially hard, because you are engineering something that has to fold and behave correctly inside a living system.

AI is now being used to narrow the field before the expensive work starts: sorting through vast numbers of candidate molecules and proposing which ones are worth building. It is not inventing medicine. It is doing triage.

I find that framing more interesting than the usual hype. A process that used to be exhaustive brute force now has one more experienced eye on it early. That does not shorten a clinical trial, and it does not change what the regulators require. It changes which molecules get to the starting line at all.

A space telescope that changes the shape of its own mirrors

NASA’s new space telescope carries a coronagraph — an instrument that deliberately blocks the light of a star. The reason is straightforward: a star is so bright that any planet next to it disappears in the glare. Block the star, and the planet can emerge.

The interesting part is how precisely it blocks. The instrument uses deformable mirrors — mirrors that physically change shape to cancel out stray light with far finer control than a fixed mask allows. The goal is to image Jupiter-like planets around other stars directly, rather than inferring them from wobbles and dips in brightness.

There is something worth sitting with in this one. To see the faint thing, you have to hide the bright thing. That is not a bad description of a lot of research.

Animals smell in stereo, and someone finally did the math

Different flavor of story here, and it is the one I keep coming back to. You have two nostrils. Two eyes give you depth perception, and it turns out two nostrils do something comparable — animals extract directional information about an odor from the difference between the left and right side. Stereo smell.

That much was known. What this preprint (a paper shared publicly before peer review) does is ask a narrower question: under what conditions is using both nostrils actually better than using one? Not whether it works, but when it is worth it.

This will not cure anything. It will not appear in a product. It is a quiet piece of work that takes something everyone already assumed and checks whether the numbers support it. That kind of paper rarely gets attention, and I think it should get more.

The hippocampus is not one structure repeated over and over

Today’s AI is built by repeating a component. The same basic mechanism, stacked and scaled, handles text, images, and audio. Uniformity is the design.

Brains, it seems, did not take that route. A preprint this week argues that the hippocampus — the region involved in forming and retrieving memory — should be understood not as one structural motif repeated at scale, but as a collection of subsystems doing genuinely different jobs. The authors’ phrase is a shift from structural monoculture to a system of systems.

Both of these are preprints, so treat them as arguments rather than settled findings. But the contrast is worth holding onto if you work with AI, as I do: one architecture wins by repeating a part, the other by combining unlike parts. They can look similar from a distance and be built on opposite principles.

What I’m taking from this week

Short version:

• The kidney result moves a deadline, not a procedure — and it is still a pig.
• AI in drug design is triage at the front of the pipeline, not a shortcut through it.
• Blocking the bright thing is how you see the faint thing.
• Confirming an assumption with numbers is real work, even when nothing ships.
• Repeating one part and combining unlike parts are different bets. We have mostly made the first one.

Five fields, no connection between them, and the same posture underneath: something long treated as fixed — the transplant clock, the cost of finding a molecule, the glare of a star, an assumption nobody had measured, the idea that a brain region is uniform — got picked back up and turned over.

None of these arrived finished. Most of them will not matter for years, and some will not matter at all. I still think the reopening is the part worth watching, and I have not decided yet which of the five will turn out to have been the important one.

Sources

This piece is built from the science and medical headlines collected on 24 July 2026. The stories above came from:

• Supercooled kidneys have been transplanted into pigs in a “landmark achievement” (MIT Technology Review)
• How AI helps scientists design the next generation of medicines (MIT Technology Review)
• Shape-shifting mirrors on NASA’s new space telescope could unveil Jupiters like our own (MIT Technology Review)
• When to Smell in Stereo (arXiv preprint)
• The Giant Hippocampus: From Structural Monoculture to a System of Systems (arXiv preprint)