Berkeley SETI 101: How UC Berkeley Listens for ET

What Berkeley is listening for

⏱ 4 min read

SETI — the Search for Extraterrestrial Intelligence — doesn't look for life directly. It looks for technosignatures: evidence of technology that nature can't easily produce. The Berkeley SETI Research Center (BSRC) runs searches across radio, visible and infrared light, but its signature work has always been in radio.

Why narrowband radio?

Natural radio sources — pulsars, gas clouds, quasars — spread their energy across wide ranges of frequency. The narrowest natural emission lines are still thousands of hertz wide. A transmitter built by engineers, by contrast, can pack its power into a channel about 1 Hz wide. Our own spacecraft, including Voyager 1 and 2, talk to Earth this way. So a very narrow spike of power is a strong hint that something was built rather than born.

That's why Berkeley's instruments are, at heart, giant spectrometers: they split incoming radio into hundreds of millions of tiny frequency channels and look for any channel that's unusually loud.

The "water hole"

Much of Berkeley's search clusters near 1.42 GHz, the 21 cm line of neutral hydrogen, and 1.66 GHz, where hydroxyl (OH) radiates. Hydrogen plus hydroxyl makes water, hence the nickname the water hole. The argument is that any radio astronomer anywhere would know these frequencies, and the band between them is relatively quiet. It's a reasonable guess, not a law — which is why modern searches now sweep from under 1 GHz to over 10 GHz.

Drift: the motion fingerprint

Planets spin and orbit. A transmitter on another planet is accelerating relative to our telescope, so its frequency slowly slides over time — a Doppler drift. Earth's rotation adds its own drift. A signal with zero drift is often suspicious: it's probably a transmitter sitting still on Earth, right next to the telescope.

Two strategies

Berkeley uses two complementary approaches. Targeted searches point a telescope at specific stars and stare. Sky surveys cover as much sky as possible, usually by riding along with other astronomers' observations. Targeted searches go deeper; surveys go wider. You'll see both in the next three modules.

SERENDIP: piggyback SETI

⏱ 4 min read

In 1979–80, Berkeley astronomer Stuart Bowyer started SERENDIP — the Search for Extraterrestrial Radio Emissions from Nearby Developed Intelligent Populations. Its founding idea was simple and cheap: don't compete for telescope time. Instead, attach a receiver to a big telescope and analyze whatever it happens to be pointing at while other astronomers do their work.

That approach is called commensal observing (from the Latin for "sharing a table"). It turns SETI from a project that needs dedicated telescope hours into one that runs nearly all the time, for free, on the world's best dishes.

The trade-off

Commensal SETI can't choose its targets. It sees what the primary observer chooses, for as long as they choose. The payoff is enormous sky coverage over years. The cost is that any interesting signal has to wait for a reobservation, sometimes much later.

Hardware that rode Moore's Law

Each SERENDIP generation was a bigger, faster spectrometer. The first had 100 channels. By the 2000s, Berkeley was building FPGA-based instruments with over 100 million channels — and the number of channels roughly doubled every year for decades. Explore the generations below.

The Arecibo years

From 1992 until 2020, SERENDIP's main home was the 305-meter Arecibo Observatory in Puerto Rico — then the world's largest single-dish radio telescope. Arecibo's huge collecting area made SERENDIP one of the most sensitive SETI searches in operation, mostly near the water hole. The same data stream later fed SETI@home. Arecibo's platform collapsed on December 1, 2020, ending that era.

SERENDIP's piggyback philosophy outlived the dish. It's the direct ancestor of today's commensal systems on MeerKAT and the Very Large Array (Module 4).

SETI@home: the crowd supercomputer

⏱ 4 min read

SERENDIP collected more data than Berkeley's own computers could analyze deeply. In the mid-1990s, David Anderson, Dan Werthimer and colleagues had a radical idea: home computers were becoming fast and internet-connected, so why not borrow their idle time? With seed funding from The Planetary Society, SETI@home launched in May 1999.

The team planned for around 50,000 volunteers. Within a year it had more than 2 million, in over 100 countries. Its screensaver became one of the most recognizable images of early internet culture.

How it worked

Arecibo data was recorded, split into small "work units" and sent to volunteers. Their software searched each chunk far more thoroughly than SERENDIP could, testing many drift rates and signal shapes — steady tones, pulses, triplets and Gaussian-shaped signals that rise and fall as the telescope beam sweeps past a point in the sky. SETI@home was about 10 times more sensitive than SERENDIP, but over a narrow 2.5 MHz band centred on 1420 MHz.

The software platform built for it, BOINC, became open-source infrastructure for dozens of other volunteer-computing projects in medicine, climate and physics.

Hibernation and the long analysis

In March 2020, SETI@home stopped sending out new work. The team had enough results and needed to focus on the hard part: sifting them. That back-end, called Nebula, removed interference, grouped related detections and ranked candidates. Walk through the funnel below.

"SETI@Home Has Finally Been Completed After 27 Years, Here's What Was Found" — Anton Petrov. Open on YouTube ↗

The 2025 papers

In 2025 the team published two papers in The Astronomical Journal: one on data acquisition and the client software, one on Nebula's analysis and findings. A key method was injecting fake signals — "birdies" — into the data to measure how well the pipeline could find them. That turns a vague "we didn't find anything" into a precise statement of what the search would have found.

Breakthrough Listen: SETI at full scale

⏱ 4 min read

On July 20, 2015, at the Royal Society in London, Stephen Hawking, Martin Rees, Ann Druyan and Frank Drake announced Breakthrough Listen: a 10-year, $100 million program funded by Yuri Milner's Breakthrough Prize Foundation, with UC Berkeley as the science lead. It was by far the largest SETI effort ever.

Berkeley's job was the hard engineering: high-speed digital electronics able to process billions of voltage samples per second, plus the data pipelines and training for a new generation of researchers.

The targets

The plan: survey the nearest million stars, the centre and plane of the Milky Way, and the nearest 100 galaxies. Initial facilities were the Green Bank Telescope in West Virginia, the Parkes "Murriyang" telescope in Australia, and the Automated Planet Finder at Lick Observatory for optical SETI.

The commensal comeback

Arrays changed the math. In December 2022, Listen began searching data from MeerKAT, which can form beams on up to 64 targets simultaneously and sees an area of sky about 50 times larger than the GBT at once. The team said it would expand the number of targets searched by roughly a factor of 1,000. On the VLA, the COSMIC system has recorded data from all 27 antennas since March 2023, riding along with the VLA Sky Survey. Both are SERENDIP's piggyback idea, rebuilt for the 2020s.

"A Tour of our Computer Hardware, Part II" — BerkeleySETI. Engineer David MacMahon shows the machines behind Breakthrough Listen. Open on YouTube ↗

Open data

Listen commits to making its data public through an open archive, and its search software, such as turboSETI, is open source. That means anyone — including students — can rerun searches with new methods. The 2026 retrospective, "The First Decade of Breakthrough Listen," calls it the most comprehensive technosignature search ever conducted.

Vetting signals: the BLC1 lesson

⏱ 4 min read

The biggest enemy of SETI isn't silence. It's RFI — radio frequency interference from phones, satellites, aircraft radar, Wi-Fi, microwave ovens and the telescope's own electronics. Human technology produces exactly the kind of narrowband signals SETI looks for. Every search generates millions of hits, and nearly all are us.

The cadence test

Breakthrough Listen's main defence is the ON–OFF cadence. The telescope alternates: point at the target star (ON), then at a nearby patch of sky (OFF), repeated several times. A real signal from the star should appear in every ON scan and vanish in every OFF scan. Interference near the telescope usually shows up in both, because it isn't coming from one spot in the sky.

BLC1: the best false alarm yet

On April 29, 2019, Parkes observed Proxima Centauri — the nearest star — for a flare study, with Listen recording. An undergraduate intern later found a signal near 982 MHz. It was narrowband, it drifted (about 0.038 Hz per second), it lasted hours, and it appeared only in the ON scans. It was named BLC1: Breakthrough Listen Candidate 1.

Then came the detailed work. Sofia Sheikh's team found dozens of lookalike signals at frequencies mathematically related to common clock oscillators — and those lookalikes did appear in OFF scans. Their conclusion, published in Nature Astronomy in October 2021: BLC1 was an intermodulation product of local interference that happened to line up with the observing cadence. Guilty by association.

Null results and what's next

⏱ 4 min read

In January 2026, Berkeley announced SETI@home's final shortlist: about 100 candidate signals out of 12 billion detections. Since July 2025, China's 500-meter FAST telescope — the only instrument able to cover Arecibo's sky with similar power — has been reobserving them, about 15 minutes per target. By early 2026, roughly half had been observed with no repeat.

"SETI@home Update: 21 Years of Citizen Science—and 100 Signals to Investigate" — SETI Institute (SETI Live with Eric Korpela, UC Berkeley). Open on YouTube ↗

What a null result really says

"We found nothing" is not the same as "there's nothing there." A well-designed null result sets an upper limit: no transmitter brighter than X, at frequencies Y, in sky regions Z. Breakthrough Listen's early estimate was that its telescopes could detect something as strong as aircraft radar from any of the 1,000 nearest star systems. That rules out some very loud neighbours — and leaves almost everything else open.

Project director Eric Korpela has said SETI@home was the most sensitive narrowband search of large parts of the sky, so it had the best chance of finding something. The team also published a candid list of what future surveys should do differently, including planning candidate vetting from day one and asking whether aggressive filters throw out real signals with the noise.

How big is the haystack?

The Drake Equation, written by Frank Drake in 1961, estimates how many communicating civilizations might exist in our galaxy. It isn't a calculation so much as a list of what we don't know. Try your own assumptions.

The road ahead

The next decade looks very different from the Arecibo era. Commensal systems on MeerKAT (over 1.2 million pointings since mid-2022) and the VLA cover far more stars than all earlier radio SETI combined. Machine-learning pipelines now run in real time on the revived Allen Telescope Array. In 2026 Breakthrough Listen announced a new campaign targeting one million nearby stars, and the field is preparing for the Square Kilometre Array.

Test yourself (10 questions)

⏱ ~6 min

Glossary

Technosignature

Observable evidence of technology, such as a narrowband radio signal or laser pulse, that nature is unlikely to produce.

Narrowband signal

A signal concentrated in a very small frequency range, around 1 Hz, far narrower than natural emission.

Doppler drift

The slow change in a signal's frequency caused by acceleration between transmitter and receiver.

Water hole

The relatively quiet radio band between hydrogen (1.42 GHz) and hydroxyl (1.66 GHz) lines.

Commensal observing

Recording and analyzing data in parallel with another astronomer's observation, without controlling where the telescope points.

RFI

Radio frequency interference — human-made signals that contaminate astronomical data.

ON–OFF cadence

Alternating between a target and nearby empty sky so that a real signal shows only in ON scans.

BOINC

Berkeley Open Infrastructure for Network Computing — the open-source volunteer-computing platform built for SETI@home.

Nebula

SETI@home's back-end pipeline that removed RFI and ranked candidates from billions of detections.

Upper limit

The strongest signal a search could have missed — what a null result rules out.

Sources

• Berkeley News — "For 21 years, enthusiasts used their home computers to search for ET…" (Jan 2026)
• Anderson et al. — "SETI@home: Data Analysis and Findings," The Astronomical Journal 170 (2025)
• SETI@home project site and news archive — UC Berkeley
• Bowyer — "A brief history of the search for extraterrestrial intelligence…" (arXiv)
• Korpela et al. — "Status of the UC-Berkeley SETI Efforts" (arXiv, 2011)
• SERENDIP — Wikipedia
• Smith et al. — "A radio technosignature search towards Proxima Centauri resulting in a signal of interest," Nature Astronomy (2021)
• Sheikh et al. — "Analysis of the Breakthrough Listen signal of interest blc1…," Nature Astronomy (2021)
• MacMahon & Czech — "The First Decade of Breakthrough Listen" (arXiv, 2026)
• Live Science — Breakthrough Listen expands search with MeerKAT

All consulted November 2026. Figures reflect published sources at that time; FAST follow-up of SETI@home candidates was still in progress.