Inside the most sophisticated plan for solar geoengineering

The premise of the episode is economic, and it is Shayle’s. Almost every climate technology is a collective action problem where the effort only works if nearly everyone participates. Solar radiation management inverts that. Shayle’s opening figure is that roughly three million tons of reflective particles in the stratosphere could cool the planet by one and a half degrees Celsius for about $30 billion, which is less than a single hyperscale data center costs. He quotes the economist Scott Barrett’s formulation: climate change is a free rider problem, where everyone waits for someone else to pay, while solar geoengineering is a free driver problem, where it only takes one. One country, one company, conceivably one individual. On that framing the low cost is not a favorable detail, it is the entire governance problem, and it is the frame Yedvab is answering against throughout.

His technical case is more modest than the episode’s title suggests, and he volunteers the modesty. The concept is old, and the default candidate has always been sulfuric acid, because large volcanic eruptions do the same thing naturally. Sulfate’s drawbacks are that it is toxic, causes acid rain, harms the ozone layer and carries high uncertainty. Stardust’s answer is two particles: amorphous silica of the kind used in toothpaste and as a food additive, and a composite with a silica core and a shell of calcite, the material in limestone and eggshells. Asked whether they work better, Yedvab says no. Effectiveness is comparable to sulfate within a few tenths of a percent, neither much better nor much worse, and he adds unprompted that Stardust is not competing on cost and that sulfur may come in somewhat cheaper. What he claims instead is that the materials are naturally occurring and known to be safe, that they are inert and biodegradable so they do not accumulate in the environment over decades, and that they treat the ozone layer better. The one performance advantage he names is optical: sulfate absorbs outgoing infrared radiation from the Earth, which the composite particle avoids, and avoiding it is what would let you push to higher levels of cooling if that were ever wanted. For cooling comparable to the warming of the past fifty years, he says the simpler silica particle alone is enough.

The genuinely novel argument is about testability, and it is the hinge between the technical question and the political one. The stratosphere already carries a few hundred thousand tons of sulfate, and that background fluctuates, so the smallest sulfate experiment that could be distinguished from noise is on the order of a million tons. That, as Yedvab puts it, is not an experiment, it is deployment. A novel particle has no such background, so you can begin orders of magnitude below the level that produces any climatic or environmental effect and ramp up stepwise against explicit success criteria, which he compares to a clinical trial for a new drug. Two other pieces support it: a tagging technology that gives each batch, and each injection site, a unique fingerprint trackable in real time as the particles move around the globe, plus monitoring of the radiation balance; and a claim that the resulting layer can be tailored more precisely than a sulfate layer. Around this sits a safety framework published as eight papers, sorting the concerns into impacts on human health and the biosphere, impacts on stratospheric chemistry, and climatic impacts. The three sit at visibly different stages of proof, which he does not hide, and only the first is anywhere near settled; climatic impacts cannot be lab tested at all, which is precisely why the small-scale ramp is his answer. He states that the testing is not done, that the papers were released so the scientific community can review and externally validate them, and that he wants three or four other organizations pursuing their own options within a few years so that governments are never choosing from a menu of one.

Governance is where he declines to go further, and the decline is the substance. Every decision, on research, on testing and on deployment, should be made by governments in the plural; Stardust would not participate in anything lacking clear and strict regulation; this is on the company’s website and its investors are aligned with it. Asked what body that would be, he says he does not know, and adds that the company should be careful not to try to shape the framework, because a technology developer steering its own regulation would be bad for several reasons. The most he offers is a precedent, the Montreal Protocol, which he describes as a US-led sequence in which academia identified the problem, US industry produced the substitute, and the US government assembled a multilateral coalition within three or four years, producing a treaty that both banned the harmful gases and set criteria for acceptable replacements. Shayle agrees it is the obvious analogy and then pushes on it: the same analogy has been reached for across every other climate effort, and the later agreement built in its image has clearly not done the trick, which is why the conversation about geoengineering is happening at all. Two further objections are aired and not resolved. On moral hazard, that a working option reduces the incentive to decarbonize, Yedvab raises the concern himself and calls it valid and serious, sets against it a moral imperative that governments have good options available in five or ten years, and says plainly that he is not claiming this is an easy problem. On the free driver risk he softens Shayle’s “anyone could do it” to “many players, not anyone,” and argues that a safe option lowers the chance someone chooses a reckless one, since dispersing toxic material into the sky requires no sophistication and no governance at all.