SETI@home

SETI@home
SETI@home logo
The SETI@home screensaver displaying radio signal analysis in progress
Project
Status	Hibernating
Category	Astrophysics / SETI
Compute	CPU & GPU
Requires	None
Development
Developer	University of California, Berkeley
Author	David Gedye, David P. Anderson, Dan Werthimer
Sponsor	National Science Foundation, NASA, volunteer donations
Maintainer	Eric Korpela (current director)
Initial release	May 17, 1999  (27 years ago)
Software
Written in	C, C++
Operating system	Windows, Linux, macOS, Android, Solaris, FreeBSD, IBM AIX, HP-UX, IRIX, OS/2 Warp
BOINC statistics
Stats as of	March 2020
Active users	91,454
Total users	1,803,163
Active hosts	144,779
Total hosts	165,178
Metadata
Website	https://setiathome.berkeley.edu/
License	GPL-2.0-or-later

SETI@home (pronounced "SETI at home") is a volunteer computing project run by the Berkeley SETI Research Center and hosted by the Space Sciences Laboratory at the University of California, Berkeley. Its goal is to analyze radio telescope data in search of signals that could indicate extraterrestrial intelligence, making it one of many activities in the worldwide SETI effort. SETI@home ran as an active volunteer computing project from May 17, 1999, until March 31, 2020, when it entered an indefinite period of hibernation while the research team focused on analyzing accumulated data.

The Arecibo Observatory in Puerto Rico was SETI@home's primary data source throughout most of the project's history. Data was recorded onto magnetic tapes and physically mailed to Berkeley for processing.

SETI@home was the third large-scale use of volunteer computing over the Internet for research purposes, following the Great Internet Mersenne Prime Search (GIMPS), launched in 1996, and distributed.net, launched in 1997. With over 5.2 million participants worldwide at its peak, it was the volunteer computing project with the most participants recorded to date, and was recognized by the 2008 edition of the Guinness World Records as the largest computation in history. Since its launch, the project has logged over two million years of aggregate computing time across more than 145,000 active and 1.4 million total computers in 233 countries.

Background and origins

The concept for SETI@home emerged in 1995 when David Gedye, then a project manager at Starwave Corp., proposed using a virtual supercomputer composed of large numbers of Internet-connected computers to perform radio SETI analysis. Prior to SETI@home, radio SETI projects relied on special-purpose supercomputers located at the telescope facility itself. Gedye partnered with University of Washington astronomer Woody Sullivan, who suggested contacting Dan Werthimer, whose SERENDIP project was already conducting SETI observations at Arecibo, and with David P. Anderson, a specialist in distributed computing at UC Berkeley's Space Sciences Laboratory. The original project description was presented at the Fifth International Conference on Bioastronomy in July 1996, and prototype versions of the distributed computation mechanism, signal analysis program, and screensaver graphics were developed that same year.

The project was initially funded with just $100,000 from The Planetary Society and Paramount Pictures, and in its early years received donated server hardware from companies such as Sun Microsystems and Intel before transitioning to a purely donation-supported model.

SETI@home was publicly launched on May 17, 1999. Within the first week, nearly 300,000 computers were already processing data from Arecibo. Within a few months, more than one million volunteers had signed up across 223 countries.

Scientific goals

SETI@home was established with two primary goals:

1. To conduct useful scientific work in an observational analysis aimed at detecting intelligent life beyond Earth.
2. To demonstrate the viability and practicality of the "volunteer computing" concept.

The second goal is considered to have succeeded fully: the BOINC platform, developed from technology pioneered by SETI@home, now supports dozens of computationally intensive scientific projects across many disciplines. The first goal, as of 2026, has produced no conclusive evidence of extraterrestrial intelligence, though the project identified a number of scientifically interesting candidate signals for follow-up observation. The most prominent of these early candidates, announced in 2004, was a recurring signal designated Radio source SHGb02+14a, which has since been attributed to instrumental or terrestrial interference rather than an artificial source. The vast majority of the sky, over 98% as of the project's most recent assessment, remains unsurveyed, and SETI researchers generally hold that any given point in the sky would need to be surveyed many times over before a large fraction of possible signal types could be excluded.

How it worked

Data acquisition

SETI@home collected observational data "piggyback" or "passively" while Arecibo (and later the Green Bank Telescope) were being used for other scientific programs. At Arecibo, data was sampled and written to high-density DLT cartridges at a rate of approximately one 35 GB tape per day. Because Arecibo lacked a broadband Internet connection, tapes were physically mailed to the SETI@home laboratory at UC Berkeley. In later years, observational data was instead recorded onto 2 terabyte SATA hard disk drives, each holding approximately 2.5 days of observations, which were likewise shipped to Berkeley by postal mail rather than transmitted over the Internet. Once received, the data was divided both in the time domain and frequency domain into work units of approximately 107 seconds in duration, each roughly 0.35 MB in size, overlapping in time but not in frequency.

The Robert C. Byrd Green Bank Telescope in West Virginia, the world's largest fully steerable radio telescope, became a secondary data source for SETI@home after it began operation in the early 2000s.

Beginning in 2016, SETI@home also began processing data recorded by the Breakthrough Listen initiative, a separately funded SETI survey that uses both the Green Bank Telescope and the Parkes Observatory in Australia, broadening the range of observational data available to the project's analysis pipeline beyond what Arecibo alone could supply.

Client software and screensaver

Volunteers installed a free client program on their computers. When the machine was otherwise idle, the program downloaded a work unit from the SETI@home server, processed it, and returned the results automatically upon the next Internet connection. The software also featured an optional screensaver that displayed a real-time visualization of the signal analysis in progress, showing spectrograms and signal-strength graphs. The screensaver became one of the project's most recognizable features and a common sight on home computers throughout the early 2000s, doubling as a form of outreach that helped popularize the broader concept of volunteer computing.

Signal detection algorithms

The client software searched for five categories of signals that distinguish genuine candidates from background noise:

1. Spikes in power spectra
2. Gaussians: rises and falls in transmission power that may represent a telescope beam's main lobe passing over a radio source
3. Triplets: three power spikes in a row
4. Pulses: repeating signals possibly representing narrowband digital-style transmissions
5. Autocorrelations: matching signal waveforms using autocorrelation

The core technique involved applying large numbers of discrete Fourier transforms (DFTs) at various chirp rates and durations, essentially equivalent to simultaneously tuning many narrow radio channels and looking for unexplained excess power. Formally, for a sampled time-series signal $x[n]$, the discrete Fourier transform at frequency bin $k$ is:

$X[k]=\sum _{n=0}^{N-1}x[n]\cdot e^{-i2\pi kn/N}$

Each work unit was analyzed across hundreds of frequency sub-bands and drift rates to account for the Doppler frequency drift that would result from the relative motion between a transmitting planet and Earth. Because a hypothetical extraterrestrial transmitter and Earth would generally be in relative motion, with their separation changing nonlinearly over time, the search also incorporated chirp rates as an additional parameter representing the predicted rate of change of the Doppler-shifted frequency, so that a constant-frequency signal at the source could still be recovered after distortion by relative acceleration.

SETI@home running in the BOINC Manager

Work unit validation

To guard against fraudulent or erroneous results, every work unit was sent to multiple computers (a practice called "initial replication," typically set to a value of 2). Credit was only awarded once a minimum number of returned results agreed with one another (the "minimum quorum"). If disagreement occurred, additional copies of the work unit were distributed until a quorum was reached. The final credit granted to all machines returning the correct result was set to the lowest value claimed among them. This validation scheme became a foundational feature of the BOINC platform generally, since any project built on volunteered, unverified hardware faces the same risk of erroneous or deliberately falsified results.

From Classic to BOINC

The initial software platform, known as "SETI@home Classic," ran from May 17, 1999, to December 15, 2005. This program was capable only of running SETI@home tasks and required volunteers to manually download new software with each algorithm update. The discontinuation of SETI@home Classic also rendered older Macintosh computers running the classic Mac OS (versions released before December 2001) unable to continue participating in the project, since the successor platform required a more modern operating environment.

In 2005, the project transitioned to the BOINC platform, which had been developed by David P. Anderson with funding from the National Science Foundation. BOINC allowed algorithm updates without requiring user intervention, enabled volunteers to contribute to multiple scientific projects simultaneously, and opened the door to new types of signal analysis. The BOINC server software itself continues to be actively maintained as an open-source project on GitHub, where development and release management of the server-side components is led by a team headed by Laurence Field of CERN, who also oversees the BOINC-based LHC@home project used for particle-physics simulations.

SETI@home Enhanced

On May 3, 2006, distribution of a new version called "SETI@home Enhanced" began. Taking advantage of increased desktop computing power since 1999, this version was approximately twice as sensitive to Gaussian signals and certain classes of pulsed signals as the original BOINC-based release. Application builds were also produced with processor-specific optimizations, especially for Intel instruction sets, allowing faster execution on compatible hardware.

GPU acceleration

With assistance from NVIDIA, SETI@home developed a client application using the CUDA parallel computing platform. This GPU-accelerated version achieved speeds from 2 to 10 times faster than the CPU-only version, depending on hardware. For example, a GeForce GTX 280 was more than twice as fast as a high-end 3.2 GHz Intel Core i7 965 CPU running the same analysis. GPU support via CUDA was formally incorporated into SETI@home starting in 2015, and the client was subsequently extended to support Intel GPUs as well, broadening the range of consumer hardware capable of accelerated processing.

AstroPulse

AstroPulse was a companion application to SETI@home designed to search for short, broadband radio pulses in the same Arecibo data. Where SETI@home concentrated on narrowband continuous signals, AstroPulse used coherent dedispersion to search for brief but powerful bursts that could indicate rapidly rotating pulsars, evaporating primordial black holes, or previously unknown astrophysical phenomena, as well as another possible signature of extraterrestrial intelligence. AstroPulse was one of the earliest test applications for BOINC. Beta testing of its final public release was completed in July 2008, and distribution of work units to qualifying machines with sufficient processing power began in mid-July 2008, continuing alongside the main SETI@home application until both ceased distributing new work in 2020.

Hibernation and renewed scientific analysis

The SETI@home project logo, used throughout the project's active phase and retained for the hibernating website and message boards.

On March 31, 2020, the SETI@home team stopped distributing new work units to volunteers, citing two main reasons: the project had reached "the point of diminishing returns," having already gathered as much usable data from Arecibo and Green Bank as the team judged necessary for the time being, and the considerable ongoing labor cost of operating the distributed-computing infrastructure itself, which the team wished to redirect toward analysis and publication. The project's organizers were explicit that this was a hibernation rather than a shutdown: the SETI@home website and community message boards continued to operate, and the team left open the possibility that volunteer computing could resume in the future, potentially using data from other instruments such as MeerKAT or the Five-hundred-meter Aperture Spherical Telescope (FAST) in China.

Following the cessation of new work distribution, the science team shifted its effort to combing through the roughly two decades of accumulated data using a back-end analysis tool called "Nebula." By around November 2021, the team had used Nebula to remove radio-frequency interference (RFI) from the dataset and had begun selecting the highest-scoring 100 or so signal "multiplets" for further scrutiny, for which they had been granted 24 hours of dedicated observing time on the FAST telescope. This follow-up phase culminated in two companion papers, accepted for publication in June 2025 and appearing later that year in The Astronomical Journal: one paper, "SETI@home: Data Acquisition and Front-End Processing," documents the project's data recorder, splitter, and client software, including the parameters and statistics of the five detection types and the algorithms used to identify them, while the second, "SETI@home: Data Analysis and Findings," describes the Nebula back end and its results, including the RFI-removal process, the candidate-ranking methodology, and the use of artificially injected "birdie" signals to calibrate detection algorithms and estimate the project's overall sensitivity. The analysis identified approximately 100 candidate signals warranting closer scrutiny, of which around 92 were prioritized for re-observation using FAST; no candidate has, to date, been confirmed as an artificial extraterrestrial signal, with the majority instead attributed to radio-frequency interference or known astrophysical sources. The findings and the renewed candidate follow-up were discussed publicly by SETI@home director Eric Korpela on the SETI Institute's "SETI Live" podcast in January 2026.

Challenges and project culture

Because SETI@home's analysis ran on hardware donated by volunteers rather than on infrastructure the project controlled directly, the team faced recurring engineering challenges distinct from those of a conventional supercomputing project. Competitive aspects of the project emerged organically among participants, who formed teams to combine the computing output of individual members; this competition intensified after the platform moved to BOINC and credit statistics became easier to track and compare across projects. As with many systems relying on self-reported, unverifiable computation, some participants attempted to manipulate the credit system, which was one of the motivations behind the replication and quorum-based validation scheme described above. Separately, the project's popularity led some individuals to install the SETI@home client on workplace or institutional computers without authorization, an act sometimes referred to in the community as "Borging" after the Borg of Star Trek; in at least two documented cases in the United States, this resulted in the termination of an employee for misusing company or government computing resources.

The Arecibo Observatory itself, SETI@home's original and longtime primary data source, sustained irreparable structural damage in 2020 and fully collapsed that December, several months after SETI@home had already entered hibernation. The collapse underscored the rationale for SETI@home's general approach of treating its telescope-based data sources as interchangeable inputs to a portable analysis pipeline, since the project's computing infrastructure and signal-detection software were not permanently tied to any single facility.

Legacy

SETI@home is widely credited with demonstrating that large-scale, Internet-based volunteer computing could be a practical tool for scientific analysis, laying the technical and organizational groundwork for the BOINC platform that now underlies dozens of unrelated research projects spanning astrophysics, biology, mathematics, and climate science. A full list of scientific publications arising from SETI@home and other BOINC-based projects, including the papers discussed above, is maintained on the official BOINC publications page.

Publications

The following is a list of scientific publications produced by the SETI@home project team, compiled in part from the official BOINC publications page maintained by UC Berkeley.

• Anderson, David P., Eric J. Korpela, Dan Werthimer, Jeff Cobb and Bruce Allen. SETI@home: Data Analysis and Findings. The Astronomical Journal (2025). DOI: 10.48550/ARXIV.2506.14737.
• Korpela, Eric J., David P. Anderson, Jeff Cobb, Matt Lebofsky, Wei Liu and Dan Werthimer. SETI@home: Data Acquisition and Front-End Processing. The Astronomical Journal (2025). DOI: 10.48550/arXiv.2506.14718.
• Zhang, Zhi-Song, Dan Werthimer, Tong-Jie Zhang et al. First SETI Observations with China's Five-hundred-meter Aperture Spherical Radio Telescope (FAST). The Astrophysical Journal (2020). DOI: 10.3847/1538-4357/ab7376.
• Korpela, Eric J., Andrew P. V. Siemion, Dan Werthimer, Matt Lebofsky, Jeff Cobb, Steve Croft and David Anderson. The next phases of SETI@home. SPIE Optical Engineering + Applications (2015). DOI: 10.1117/12.2188619.
• Korpela, Eric J. SETI@home, BOINC, and Volunteer Distributed Computing. Annual Review of Earth and Planetary Sciences (2012). DOI: 10.1146/annurev-earth-040809-152348.
• Korpela, Eric. Distributed Processing of SETI Data. Searching for Extraterrestrial Intelligence (2011).
• Korpela, Eric J., David P. Anderson, Robert Bankay et al. Status of the UC-Berkeley SETI Efforts. (2011). DOI: 10.48550/ARXIV.1108.3134.
• Korpela, Eric J., Jeff Cobb, Matt Lebofsky, Andrew Siemion, Joshua von Korff, Robert C. Bankay, Dan Werthimer and David Anderson. Candidate Identification and Interference Removal in SETI@home. (2011). DOI: 10.48550/ARXIV.1109.1595.
• Korpela, E. J., D. P. Anderson, R. Bankay et al. SETI with Help from Five Million Volunteers: The Berkeley SETI Efforts. Bioastronomy 2007: Molecules, Microbes and Extraterrestrial Life (2009).
• Korpela, Eric J., Jeff Cobb, Steve Fulton et al. Three Years of SETI@home: A Status Report. Symposium — International Astronomical Union (2004). DOI: 10.1017/S0074180900193635.
• Demorest, Paul, Aaron Golden, Eric Korpela, Dan Werthimer and Ron Ekers. Serendipitous Detection of Radio Pulses from Evaporating Black Holes, GRBs and Extragalactic Supernova Using SETI@home. Astronomy, Cosmology and Fundamental Physics (2003).
• Anderson, David P., Jeff Cobb, Eric Korpela, Matt Lebofsky and Dan Werthimer. SETI@home: an experiment in public-resource computing. Communications of the ACM (2002). DOI: 10.1145/581571.581573.
• Werthimer, Dan, David Anderson, C. Stuart Bowyer et al. Berkeley radio and optical SETI programs: SETI@home, SERENDIP, and SEVENDIP. Photonics West 2001 — LASE (2001). DOI: 10.1117/12.435384.
• Sullivan, Woodruff T., III, Dan Werthimer, Stuart Bowyer, Jeff Cobb, David Gedye and David Anderson. A new major SETI project based on project SERENDIP data and 100,000 personal computers. (1997).

For the complete and continually updated list of publications across all BOINC-based projects, see the official BOINC Publications page.

See also

• BOINC
• AstroPulse
• SERENDIP
• Breakthrough Listen
• Einstein@Home
• Radio source SHGb02+14a