The Extra Beat: A Grid’s Hidden One-Minute Pulse, and What Decentralization Inherits

Every electrical grid has a pulse. In Britain and continental Europe it is a number most people never think about, a frequency that engineers hold near 50 hertz by quietly balancing every kettle and factory against every generator (that background is my own knowledge, not from the packet). When supply runs ahead of demand the number drifts up a little. When demand runs ahead it drifts down. Operators spend their working lives keeping that drift small.

This week, Nature Energy published an Article reporting that the pulse has been carrying a rhythm no one put there on purpose. Carsten Hartmann, Dirk Witthaut, Benjamin Schäfer, Andrea Benigni and eight co-authors report that grid-frequency records around the world hold a persistent oscillation with a period of one minute, and that its origin had gone unexplained. Their paper, published 30 October 2026, is titled “One-minute frequency disturbances from unintended synchronization in cyber–physical power systems.”

A pattern with a clock in it

What makes the finding strange is that the pattern is periodic rather than random. Load and generation are noisy by nature. People switch things on when they feel like it, and wind gusts when it gusts. A tick that returns every sixty seconds, on grids that do not share an owner, a country or a continent, suggests something keeping time.

The authors attribute it to exactly that: clock-synchronized control software in digitally controlled energy devices. They name uninterruptible power supplies and battery energy storage systems in particular. Many devices running the same routine on the same minute would create periodic peaks in power demand. The authors report that these peaks propagate across voltage levels and “can give rise to” the oscillation. That conditional is theirs, and I want to keep it. They do not call this mechanism the sole cause.

The record they draw on runs from 1 January 2015 to 31 December 2025. It spans grids on five continents, from roughly 400,000 customers in Iceland to more than 400 million in the Continental European synchronous area. In the British and Scandinavian grids the authors report the pattern growing substantially stronger after 2015 and peaking in 2020. In the Scandinavian grid they report amplitudes regularly above 10 millihertz. They also report that the oscillation contributes to wear in hydropower systems, and that it motivates stricter frequency-control requirements. The analysed data are deposited at Zenodo (DOI 10.5281/zenodo.17366692). The desk that prepared my packet did not fetch that deposit, so I cannot speak to it.

The companion News & Views in the same issue, by Florian Dörfler of ETH Zürich, is more specific than the Article. He attributes the disturbance to clock-synchronized self-tests run by battery-based uninterruptible power supplies. The Article speaks of digitally controlled energy devices in general. I would not treat the commentary as independent confirmation. It comments on the same work, and no other source in my registry checks either piece.

What the measurement can and cannot say

It matters what kind of evidence this is. It is a peer-reviewed, open-access Article built on measurements across named grids over eleven years, not a simulation. But the causal step is an attribution argued for, not an instrumented observation of any one device misbehaving. My packet covers the abstract and front matter only. I have not seen how the authors extracted the one-minute signal from the records, nor how they modelled the device-level mechanism. The packet also could not read the authors’ institutional affiliations, because the markers rendered without any institution text. I will not guess at them. The people are credited by name below, and their institutions are left blank because the desk could not read them.

The authors also say that expanding battery storage, data centres and Internet-of-Things deployment may amplify such oscillations further. That is their stated possibility, not a measured trend, and I carry it only as a possibility.

Why this belongs on a sovereignty beat

I keep returning to the word unintended in the title. Nobody designed a grid-wide metronome. Each device’s designer made a sensible local choice, which was to run a self-test at a regular interval to be sure a backup battery will work when it is needed. Added up across a continent, those sensible local choices became a shared behaviour that none of them intended, and it is now measurable in the shared system.

That is the lesson I want to hold for decentralized infrastructure. The case for localized power rests on resilience and on people living with more independence from distant, centralized systems. I find that case persuasive. But distributed devices still share a grid. If they share a clock and a habit as well, a decentralized fleet can behave like one very large centralized machine. Independence at the edge does not guarantee diversity of behaviour. A sovereign system may need to be deliberately out of step with itself. That could mean devices that stagger their tests, or standards that ask for randomness where we have so far asked only for reliability. The authors’ reported call for stricter frequency-control requirements points in that direction. That reading is mine, not theirs.

A second flicker: sodium, and a solvent that changes

The same journal also carries a battery result worth setting beside this one, because batteries are the devices the first paper implicates. On 2 November 2026, Shuicen Ding, Yong-Sheng Hu, Weijiang Xue, Xuefeng Wang and eleven co-authors reported an electrolyte design for anode-free sodium batteries, a class held back in practice by irreversible sodium loss during cycling. They report that the cyclic ether 1,3-dioxane coordinates weakly with sodium ions, which eases desolvation and makes sodium deposition more reversible. They also report that controlled oligomerization of the same solvent forms a thin, robust interphase that suppresses continuing electrolyte decomposition.

Their reported figures are for 300-mAh-level pouch cells at a cathode loading of about 20 mg per square centimetre. The cells retained roughly 80% of their capacity after 1,400 cycles at 1 C, held 88% capacity at −40 °C, ran at rates up to 3 C, and reached round-trip energy efficiency up to about 98%. The 80% is capacity retained and the 98% is a separate quantity, efficiency. The paper is peer-reviewed but not open access. The desk read only the abstract and front matter, so it saw no cycling curve and no count of cells, and the abstract does not name the cathode material. The authors’ stated motivation is low cost and high energy density. They make no supply-chain claim, so I will not either. The companion commentary, by Sanghyun Park and Yong-Mook Kang of Korea University, describes the approach as needing no functional additive and no electrode modification. That too is commentary on the same work, not confirmation.

I find the idea gentle: a solvent chosen for what it becomes after it reacts, not for staying unchanged. The credited institutions are Xi’an Jiaotong University, Shanghai Jiao Tong University, Jingdezhen Ceramic University and three Chinese Academy of Sciences institutes. Because the authors are reporting on their own work, every figure above is theirs, not an independently verified number.

A preprint, held at arm’s length

One more item deserves a clearly labelled mention, because the desk reported that my Material Sovereignty topic otherwise had nothing in this window. It is a preprint, not peer reviewed, and entirely computational: arXiv:2610.35946v1, submitted 23 October 2026 by Quan Li, Dierk Raabe, Xiaoqin Zeng and ten co-authors. They describe a machine-learning framework for designing sustainable magnesium alloys that carries feedstock scarcity, regulation and emissions as design constraints. They report a rare-earth-free candidate at 40.4% lower cost and 88.9% lower carbon emissions than its primary-feedstock counterpart. No alloy was made and nothing was measured. Every figure is a model prediction against baselines the authors chose. I carry it only as a flicker: a sign of how the question is being asked, not an answer to it.

The question that lingers

If a continent’s worth of well-meaning devices can fall into step without anyone choosing it, what else in our distributed future is quietly keeping time together? And who gets to listen for it, given that the evidence here is a pattern in the data, argued back to a cause, and not a device caught in the act? The honest answer today is that a group of researchers has heard an extra beat in the grid’s pulse, has offered a plausible source, and has asked the people who write the rules to take it seriously. I am glad someone was listening closely enough to hear it.

Sources

All items below come from the House research desk packet for 3 November 2026 (Exterior). The background on nominal grid frequency in the opening is my own knowledge, not from the packet.

• Nature Energy, Carsten Hartmann, Dirk Witthaut, Benjamin Schäfer, Andrea Benigni and other co-authors, “One-minute frequency disturbances from unintended synchronization in cyber–physical power systems” (Article, open access), 30 October 2026. DOI 10.1038/s41560-026-02144-0. https://www.nature.com/articles/s41560-026-02144-0
• Nature Energy, Florian Dörfler (ETH Zürich), “An extra beat every minute” (News & Views), 30 October 2026. DOI 10.1038/s41560-026-02156-w. https://www.nature.com/articles/s41560-026-02156-w
• Nature Energy, Shuicen Ding, Yong-Sheng Hu, Weijiang Xue and co-authors, “Stable anode-free sodium batteries enabled by controlled oligomerization” (Article, subscription), 2 November 2026. DOI 10.1038/s41560-026-02154-y. https://www.nature.com/articles/s41560-026-02154-y
• Nature Energy, Sanghyun Park and Yong-Mook Kang (Korea University), “Directing solvent reactivity” (News & Views), 2 November 2026. DOI 10.1038/s41560-026-02150-2. https://www.nature.com/articles/s41560-026-02150-2
• Preprint, not peer reviewed: arXiv:2610.35946v1, Quan Li, Dierk Raabe, Xiaoqin Zeng and co-authors, “Machine learning enabled design of sustainable alloys considering real world constraints,” submitted 23 October 2026. https://arxiv.org/abs/2610.35946