I hardly need to run through the math to point out how utterly absurd it would be to have two civilizations develop within a few light years of each other at roughly the same time. The notion that we might pick up a SETI signal from a culture more or less like our own fails on almost every level, but especially on the idea of time. A glance at how briefly we have had a technological society makes the point eloquently. We can contrast it to how many aeons Earth has seen since its formation 4.6 billion years ago.

Brian Lacki (UC-Berkeley) looked into the matter in detail at a Breakthrough Discuss meeting in 2021. Lacki points out that our use of radio takes up 100,000,000th of the lifespan of the Sun. We must think, he believes, in terms of temporal coincidence, as the graph he presented at the meeting shows. Note the arbitrary placement of a civilization at Centauri B, and others at Centauri A and C, along with our own timeline. The thin line representing our civilization actually corresponds to a lifetime of 10 million years. What are the odds that the lines of any two stars coincide? Faint indeed, unless societies can persist for not just millions but even billions of years. We don’t know if they can, but we need to think about it in terms of what we might receive.

Image: Brian Lacki’s slide illustrating temporal coincidence. Credit: Brian Lacki.

But there is another point. Should we assume that stars near the Sun are roughly the same age as ours? You might think so at first glance, given the likely formation of our star in a stellar cluster, but in fact clusters separate and diverge over time, so that finding the Sun’s birthplace and its siblings is challenging in itself (though some astronomers are trying). As we’re also learning, slowly but surely, stars around us in the Milky Way’s so-called ‘thin disk’ – within which the Sun moves – actually show a wider range of ages than we first thought.

A planet-hosting star a billion years older than ours might be a more interesting SETI target than one considerably younger, simply because life has had more time to start emerging on its planets. But untangling all the factors that help us understand stellar age and movement is not easy. What is now happening is that we are developing what are known as chrono-chemo-kinematical maps, which track these factors along with the chemical composition of the stars under study. Here we’re combining spectroscopic analysis with models of stellar evolution and radial velocity analysis.

This multi-dimensional approach is greatly aided by ESA’s Gaia mission and its extensive datasets on stars within a few thousand light years of the Sun. Gaia is remarkably helpful at using astrometry to pin down stellar motion and distance. Then we can factor in metallicity, for the oldest stars in the galaxy were formed at a time when hydrogen and helium were about the only ingredients the cosmos had to work with. A chrono-chemo-kinematical map can interrelate these factors, and with the help of neural networks tease out some conclusions that have surprised astronomers.

Thus a new paper out of the Leibniz-Institut für Astrophysik Potsdam. Here Samir Nepal and colleagues have been using machine learning (with what they call a ‘hybrid convolutional neural network’) to attack one million spectra from the Radial Velocity Spectrometer (RVS) in Gaia’s Data Release 3. Altogether, they are working with a sample of 565,606 stars to determine their parameters. Here the metallicity of stars is significant because the thin disk, which extends in the plane of the galaxy out to its edges, has been thought to consist primarily of younger Population I stars. Thus we should find higher metallicity, as we do, in a region of ongoing star formation.

But the Gaia mission is helping us understand that there is a surprising portion of the thin disk that consists of ancient stars on orbits that are similar to the Sun. And while the thin disk has largely been thought to have begun forming some 8 to 10 billion years ago, the maps that are emerging from the Potsdam work show that the majority of ancient stars in the Gaia sample (within 3200 light years) are far older than this. Most are metal-poor, but some have higher metal content than our Sun, which implies that early in the Milky Way’s development metal enrichment could already take place.

Let me quote Samir Nepal directly on this:

“These ancient stars in the disc suggest that the formation of the Milky Way’s thin disc began much earlier than previously believed, by about 4-5 billion years. This study also highlights that our galaxy had an intense star formation at early epochs leading to very fast metal enrichment in the inner regions and the formation of the disc. This discovery aligns the Milky Way’s disc formation timeline with those of high-redshift galaxies observed by the James Webb Space Telescope (JWST) and Atacama Large Millimeter Array (ALMA) Radio Telescope. It indicates that cold discs can form and stabilize very early in the universe’s history, providing new insights into the evolution of galaxies.“

Image: An artist’s impression of our Milky Way galaxy, a roughly 13 billon-year-old ‘barred spiral galaxy’ that is home to a few hundred billion stars. On the left, a face-on view shows the spiral structure of the Galactic Disc, where the majority of stars are located, interspersed with a diffuse mixture of gas and cosmic dust. The disc measures about 100 000 light-years across, and the Sun sits about half way between its centre and periphery. On the right, an edge-on view reveals the flattened shape of the disc. Observations point to a substructure: a thin disc some 700 light-years high embedded in a thick disc, about 3000 light-years high and populated with older stars. Credit: Left: NASA/JPL-Caltech; right: ESA; layout: ESA/ATG medialab.

Here is an image showing the movement of stars near the Sun around galactic center, as informed by the Potsdam work:

Image: Rotational motion of young (blue) and old (red) stars similar to the Sun (orange). Credit: Background image by NASA/JPL-Caltech/R. Hurt (SSC/Caltech).

A few thoughts: Combining data from different sources using the neural networks deployed in this study, and empowered by the Gaia DR3 RVS results, the authors are able to cover a wide range of stellar parameters, from gravity, temperature and metal content to distances, kinematics and stellar age. It’s going to take that kind of depth to begin to untangle the interacting structures of the Milky Way and place them into the context of their early formation.

Secondly, these results really seem surprising given that while the majority of the metal-poor stars in thin-disk orbits are older than 10 billion years, fully 50 percent are older than 13 billion years. The thin disk began forming less than a billion years after the Big Bang – that’s 4 billion years earlier than previous estimates. We also learn that while metallicity is a key factor, it varies considerably throughout this older population. In other words, intense star formation made metal enrichment possible, working swiftly from the inner regions of the galaxy and pushing outwards.

So our Solar System is moving through regions containing a higher proportion of ancient stars than we knew, and upcoming work extending these machine learning techniques, now in the planning stages and using data from the 4-metre Multi-Object Spectroscopic Telescope (4MOST) should refine the results of the Potsdam team in 2025. I return to what this may tell us from a SETI perspective. Ancient stars, especially those with higher than expected metallicity, should be interesting targets given the opportunities for life and technology to develop on their planets.

Maybe we’re making the Fermi question even tougher to answer. Because many such stars in the nearby cosmic environment are older — far older — than we had realized.

The paper is Nepal et al., “Discovery of the local counterpart of disc galaxies at z > 4: The oldest thin disc of the Milky Way using Gaia-RVS,” accepted for publication in Astronomy & Astrophysics (preprint).