Dear Aventine Readers and Ocean Lovers, 

The ocean has always been something of an unsung hero when it comes to carbon dioxide removal, absorbing about a quarter of all the CO₂ humans generate every year. Some scientists believe that through various engineering and chemical interventions it could absorb even more, helping slow the pace of global warming. But there are plenty of questions about the unintended consequences that could come from hacking an ecological system that has been serving the planet for millenia. In this issue we interview experts about one proposed intervention — seeding the ocean with iron in order to encourage phytoplankton growth — to get their thoughts on whether the benefits outweigh the risks. 

Also in this issue: 

Thanks so much for reading, 

Danielle Mattoon
Executive Director

Subscribe

Subscribe to our newsletter and be kept up to date on upcoming Aventine projects

The ocean has been called the world’s greatest ally against climate change. Every year, it absorbs about a quarter of all carbon dioxide emissions produced by humans.

But what if the ocean could be engineered to pull even more carbon from the atmosphere? Marine carbon dioxide removal, which leverages the ocean’s natural ability to store carbon, is being eyed as a potential new tool in the fight against global warming. Since September 2023, more than 400 scientists have signed a letter endorsing expanded ocean-based carbon dioxide removal research and development; additionally, several of the world’s leading scientific bodies, including the Intergovernmental Panel on Climate Change (IPCC) and the National Academies of Sciences, Engineering, and Medicine, have identified marine carbon dioxide removal as an emerging climate solution. 

One of the most studied of the six approaches to doing this is known as  ocean iron fertilization, or OIF, which may soon be tested on as much as 10,000 square kilometers of the northeastern Pacific Ocean. According to Scientific American, the trial — slated to start as soon as 2026 and organized by Exploring Ocean Iron Solutions (ExOIS), a nonprofit consortium of scientists — hopes to establish just how much carbon dioxide the approach can sequester, and what impact it will have on ocean biodiversity.

The strategy involves seeding the ocean with iron to promote the growth of marine plants, or phytoplankton, which grow near the ocean’s surface and remove CO2 through photosynthesis. As the phytoplankton die, some of that carbon sinks to the ocean’s floor. OIF occurs naturally when winds blow desert dust or ash from a volcanic eruption over the ocean’s surface; in fact large volcanic eruptions have resulted in phytoplankton blooms big enough to be seen from space.

However, like other geoengineering approaches to tackle climate change, OIF is controversial. In 2023, a group of scientists published an article warning that ocean manipulation may harm deep-sea ecosystems. For example, the explosion of phytoplankton growth could inadvertently rob the surroundings of nutrients or oxygen. And a 2020 study that ran simulations on the interactions between phytoplankton, iron and other nutrients found that iron fertilization would not lead to a significant increase in global carbon dioxide drawdown. The analysis found that the total amount of iron that phytoplankton receive is already “just enough” and adding more would not have long term beneficial effects.

Aventine spoke with five experts about what current data can (and cannot) tell us, and whether the risk of iron fertilization is worth the potential reward. Ultimately, they agree that more research is needed before deploying OIF at scale or attempting to commercialize the technology.

The El Capitan computer at the Lawrence Livermore National Laboratory is 22 times more powerful than LLNL's previous fastest supercomputer, Sierra. Garry McLeod/LLNL

The world’s new fastest supercomputer, is designed to simulate nuclear bombs. Inside the Lawrence Livermore National Laboratory (LLNL) in Livermore, California, sits an impressive new computer with mind-boggling performance: the El Capitan supercomputer, which contains 11,039,616 processor units and can perform 1.742 quintillion calculations per second. That hard-to-fathom number-crunching ability makes El Capitan, which went into service only this year, the fastest supercomputer in the world, taking the top spot on the international top 500 supercomputer ranking. It pushed aside the formerly top-ranked Frontier, which is located at Oak Ridge National Laboratory in Oak Ridge, Tennessee, in addition to the number-two-ranked Aurora, which can be found at Argonne National Laboratory in Lemont, Illinois. Worth noting: These computers are known as exascale computers, meaning that they are capable of performing more than one quintillion calculations per second, and the only three on the list. It is believed, however, that China also has computers with similar capabilities that aren’t publicized. Computers such as these are used by various national labs for fairly specific applications, and El Capitan will be put to work by the National Nuclear Security Administration for a variety of duties, including simulating the effects of nuclear explosions, which will obviate the need to perform underground detonations. Additionally, the computer will be able to assess the state of the nation’s nuclear stockpile by modelling its ageing and reliability. El Capitan’s performance allows researchers at LLNL to “do things in a few days that would have taken a few months” previously, one researcher told IEEE Spectrum.

AI can recreate your personality. All it takes for software to replicate your worldview is a two-hour interview about your life and some judicious use of large language models. That’s according to new research out of Stanford and Google DeepMind, published on the pre-press site arXiv and not yet peer reviewed. The study used data from two-hour interviews with just over 1,000 people to train an AI agent for each participant. The participants and their corresponding AI agents were then tested on a series of personality and social surveys, and the results were 85 percent similar. The goal of the research, MIT Technology Review reports, is to build cohorts of digital people that can be used to carry out studies in the social sciences that may be too time-consuming, expensive, unethical or even impossible to undertake in real life. Yet there is clearly a huge reason for concern here, too, as the research points to a possible future in which nefarious actors could clone a person’s thoughts and opinions in order to impersonate them. At the moment the technology isn’t quite sophisticated enough for that. One reason not to worry too deeply, for now at least, is that many of the tests used to compare participants to their models, while commonly used in social science, don’t attempt to fully capture all facets of the human condition — so the AI may, thankfully, still have some gaping holes in its understanding that a human could easily recognize. 

Sensing what’s in sweat could help to track your health. Most of us are aware that perspiration contains salts as well as water. But the fluid is also packed with other stuff: proteins, hormones, neurotransmitters, vitamins and more, reaching a total of more than 30,000 different biomolecules. New Scientist reports that several startups and labs are seizing on this to develop continuous, noninvasive methods of monitoring health, aided by advances in technologies such as microfluidics and batteries that have made it possible to shrink sensing devices so that they can be worn easily. We can already measure how much sodium is lost through sweat so that athletes can know how best to rehydrate, and several startups offer this as technology. More interesting are forms of sensing currently in development: measuring pH as a proxy for lactic acid concentration could reveal how intensely someone is exercising; uric acid levels could reveal early signs of gout; estrogen levels could help track ovulation cycles; the presence of recreational drugs could be monitored in people required to abstain. Sampling sweat will never replace the need to sample other fluids for a fuller picture of one’s health — blood, in particular, remains a gold standard — but for certain insights the information in sweat could prove useful. 

Get in, Loser—We’re Chasing a Waymo Into the Future, from Wired
5,800 words, or about 22 minutes

For most people in the U.S., autonomous cars remain a sci-fi concept. But for those who spend time in one of a handful of cities — San Francisco, Los Angeles or Phoenix — driverless cars have quietly become part of everyday life. To explore how driverless cars have gradually advanced, and to evaluate their near future in other parts of the world, Wired decided to tail some of Waymo’s fleet around the streets of San Francisco. If there’s a key takeaway here, it’s that the people who rely on these things have come to love them, both for the smooth ride and the lack of a pesky human driver to make small talk with. Even the longtime driver Wired enlisted to follow the driverless cars around was impressed by the driving style of Waymo’s vehicles. It’s still unclear exactly what impact autonomous vehicles will have on our cities, but this story makes it clear that finding out will be a case of when, not if.

How the world’s biggest laser smashed a nuclear-fusion record, from Nature
2,800 words, or about 11 minutes

There is only one place where a nuclear-fusion reaction has generated more energy than it has consumed: the National Ignition Facility at Lawrence Livermore National Laboratory in California. The equipment there, which covers an area the size of three football fields, uses 192 laser beams that are amplified more than a quadrillion times before being focused to create intense heat around a pellet of hydrogen isotopes the size of a peppercorn. The isotopes then fuse, creating huge quantities of power. Or at least, they do now. The first successful break-even experiment took place in December 2022, 13 years after the facility first opened, but they have now become commonplace. While the NIF will never be a commercial power-generating facility, it is now at the forefront of the research that’s required to make such facilities a possibility. This piece is essential reading for anyone who wants to understand the complexities of making nuclear fusion a reality.

The Therapist in the Machine, from The Baffler
4,800 words, or about 19 minutes

Can a large language model help with your mental health? A growing number of startups building AI therapists certainly think so, all of them offering variations of talking cures mediated by tools that resemble ChatGPT. There’s clearly a charm to the idea: An AI bot is always around to talk, unlike a regular therapist, and far more affordable than speaking to a trained human. No surprise, then, that it's being taken seriously by health care providers in some countries as a potential solution that won’t break the bank. But there are, as this piece points out, obvious limitations to the abilities of AI therapy because the underlying models still lack the nuance, insights and flexibility of a human. Also, genuine empathy is clearly a critical part of the therapeutic process and it is far from clear how even improved technologies might replicate it.