The SETI@home screensaver identifies four different types of events: spikes, Gaussians, pulses, and triplets. Our science team analyzes and removes events clearly caused by Earth broadcasts (also called RFI), leaving us with noise and (hopefully) ET. Distinguishing an ET signal from noise is not easy. One way to distinguish the two is to assume that extraterrestrials send signals persistently over time. (These signals could be intentional attempts at communication or merely leakage. Our own civilization on Earth has been leaking radio signals into space for about 60 years now, beginning with radio and television programs such as "The I Love Lucy Show". That show may seem outdated now, but it's only entering its first season at locations 60 light-years away from us.) Over the past 3 years, SETI@home has accumulated enough data to cover each part of the sky approximately 3-4 times, giving us the ability to match up signals arriving from the same direction in space. If these matching signals have similar frequencies, then it's quite possible they were emitted by the same source.

I Love Lucy—Earth's Ambassador to the Stars: An example of radio leakage from Earth.

But what constitutes a "similar" frequency match? From our position here on Earth, it's not always obvious which incoming signals have matching frequencies because of a phenomenon called "the Doppler effect". Here is the dictionary definition: "The Doppler effect is a change in the frequency at which waves [radio waves, in SETI@home's case] from a given source reach an observer when the source and the observer are in motion with respect to each other so that the frequency increases or decreases according to the speed at which the distance is decreasing or increasing." For example, when you hear a car honking as it passes you, the frequency—or pitch—of the sound changes as the car passes. Similarly, if the Earth is moving toward an incoming signal, that signal's frequency will appear higher than if the Earth is moving away from that same signal. Most bodies in space are spinning and moving relative to one another, and this movement changes incoming frequencies detected at the Arecibo radio telescope. Furthermore, if a body is in an orbit, over time it will move toward and away from us at different rates. Thus, if we detect the same signal from the same source at three different times, the detection frequency for that signal might be different each time. (See Figure 1.) Figure 1: If an orbiting source (such as a planet) emits a persistent signal, a telescope on Earth might detect three completely different frequencies for that signal, depending on the source's orbital location. Incoming frequencies are also affected by Earth's own spin and rotation around the sun. We can compensate for this movement by calculating barycentric frequency, which is the frequency we would detect if were at rest relative to the gravitational center of the solar system (or "barycenter"). We can also compensate for "drifting" signals (signals that rise and fall in frequency over short periods of time) by calculating chirp rates. However, we can't control for the movements of an incoming signal's source. (We have no way of estimating how an unknown source was moving when it emitted the signal.) Thus, we will still detect undirected extraterrestrial radio leakage at different barycentric frequencies over time. To be certain of catching signals that might come from the same source, we need to allow for a wide frequency range of +/-50,000 Hz. This wide range covers the maximum Doppler shifts we would expect from a source. We currently have hundreds of thousands of signal matches using this criteria.

But what if an extraterrestrial source is actually taking its own movements into account when emitting signals? For example, when sending a signal, an extraterrestrial civilization might control for its home planet's orbit around a star, its movement within a galaxy, or even its movement relative to the center of the universe. As a result, incoming barycentric frequencies detected at Earth would be constant over time. The probability of such an unwavering frequency occuring by chance is extremely small, and so its detection would provide strong evidence for extraterrestrial intelligence. We want to be able to detect and prioritize these cases, and so we specifically search for signal candidates with "magical frames of reference"—sets of signals sent by a source who was using a frame of reference relative to its position in a solar system, galaxy, or universe. More than just "leakage", such a discovery would imply an intentional attempt by an extraterrestrial civilization to send intelligible communication into space—truly a "magical" revelation. Signals coming from the same location in space at tightly matching frequencies (within 125 Hz, to account for a small margin of error) are considered candidates with a magical frame of reference. (See Figure 2.) Figure 2: If a source emits a signal while controlling for its own orbital movements, a telescope on Earth will detect the same barycentric frequency regardless of the source's orbital position. Such a detection would be identified as having a "magical frame of reference". We're currently compiling lists of both general and "magical frame of reference" candidates. At this time we are only identifying candidates that match within individual signal types. (In the near future we will be matching across different types as well.) The table below shows the approximate number of candidates in each current list. Special reference frame (+/-125 Hz) General (+/-50,000 Hz) Gaussian candidates ~3,000 ~150,000 Pulse candidates ~60,000 ~150,000 Triplet candidates ~60,000 ~150,000 These numbers will rise as more data from SETI@home screensavers come in, and fall as more RFI is detected and removed. Eventually, we expect that most, if not all, of these candidates will be conclusively identified as RFI or noise. We will publish the best matches from each list on our website soon. (Some of the Gaussian candidates are already available.) Until then, everyone keep up the good work crunching data, and we will be reporting our progress as it happens.