The Big Crunch: Interview with Dan Werthimer
The Big Crunch:
Interview with Dan Werthimer
Cosmologists model the end of the universe as The Big Crunch-they roughly reverse the simulations of how the expansive Big Bang might have spawned all the stars and planets. But for those interested in what is happening between the bang and the crunch on those uncountable number of planets, the big crunch can mean something different. The scientists who built the world’s largest supercomputer have a daily task of sorting the radio noise that showers our little corner of the galaxy. For this number crunching, their network can be as close as your own home.
No matter how much time you may think you are spending on your computer today, it is likely you are unable to overload its true capabilities. The average personal computer is running in second gear most of the time. Whatever is happening–whether surfing the internet or answering an email– you are not likely taxing the processor (CPU cycles) or its spare bandwidth. This ‘overhang’ is exactly the spare capacity that innovative scientists have figured out how to tap into: what if millions of PCs could be linked together in parallel to finish a job, one that normally would require a supercomputer and its mainframe price in the hundreds of millions of dollars? The concept is called distributed computing. The most well-known implementation is a massive search of the night sky for intelligent radio signals, the SETI@home project.
|Top inset: The massive Arecibo dish, world’s largest radio telescope. SETI@home instrument sits in the raised carriage housing; lower inset: SETI@home Chief Scientist, Dan Werthimer (right), and David Anderson, Project Leader (left).|
"We didn’t expect friendly competition [among users] to lead to such craziness." –D. Werthimer
Image Credit: Arecibo – Planetary Society
That has been one bold dream of the Berkeley team behind the SETI@home screensaver program. The project runs the world’s largest virtual supercomputer. In rough terms, if the planet’s total computer resources were tallied, the SETI@home volunteers contribute about one out of every thousand processing cycle potentially available everywhere. But the volunteer network is not just about crunching data. Arguably, the underlying screensaver program itself is the most widely distributed software in daily use ever conceived  –other than maybe web browsers– given that the screensaver is most active when the rest of the unused programs are idle.
In a very practical way, SETI@home has changed the world’s perception of what can be done while the rest of the world, and their computers, are asleep.
After more than five years running SETI@home, Werthimer, the Chief Scientist, and Anderson, the Project Leader, may appear not to need much sleep, given that they have tied together the entire world (226 countries) in a way that even the most optimistic futurists would have missed. But seamlessly changing time zones is something computers are good at. On a per capita basis, other than North America and Scandanavia, the most active region for contributing to SETI@home is the most remote: Antarctica.
If jumping 226 national borders isn’t challenging enough, moving data back and forth to 4.7 million individual computers is something you have to be a little audacious to even dream about. Most network administrators would wilt at the assignment. For this task at Berkeley, the home server is aptly named "SAGAN", and goes well beyond handling Carl Sagan’s hallmark equivalent by having powered ‘billions and billions’ of jobs.
As a radio astronomer, Werthimer is perhaps trained from graduate school to see the passage of time not in cyclical sunrises and sunsets– but in his own unique data units. Werthimer’s cycles are a bit different. Each day for the SETI@home network offers the equivalent to a thousand years of donated computing time. Each signal is listened to on billions of channels.
Each day, the network is touching the equivalent population of what might fill up the Rose Bowl’s 100,000 seating capacity– more than six times over.
Working together with a radio carriage mounted above the world’s largest radio telescope, Werthimer has managed to cover about ninety-five percent of the available sky from that location (Arecibo, Puerto Rico). The giant Arecibo dish spans 305 meters (more than three football fields across) in the Puerto Rican woodlands. To fill in any holes in their sky map, observations from the Southern Hemisphere are currently part of the Southern SERENDIP, a sky survey that covers stars unviewable from Arecibo. The team is also in the early planning stages with the Australian-based Parkes telescope in what is called the Southern SETI@home. [Among Parkes' many noteworthy science contributions, the telescope was the star of the Australian movie, "The Dish", since it proved the critical link that brought to the world the live images from the first moonwalk of Neil Armstrong. Without Parkes, one of the most stunning events of the century would have been just fuzzy TV static].
The first lesson from those listening for potentially intelligent radio signals may well be the most obvious one: the universe is very big place. The bad news is this gives a lot of places to point a receiver; that is the good news too, at least as far as considering the possible candidates for where a civilization might advance far enough technologically to communicate with our tiny corner of the galaxy.
The SETI@home database has logged about 100 million candidate signals. About two hundred have been filtered into a bin labelled ‘interesting’. Ultimately, Werthimer is the radio astronomer who knows what is interesting or not. Astrobiology Magazine had the opportunity to talk with Dr. Werthimer about what it takes to survive the big crunch.
Astrobiology Magazine (AM): Remarkably, a single modern desktop PC today is more powerful than a 15-year old supercomputer . So in a way, by patching together a volunteer network like SETI@home, that is the best way–or even the only way– to keep current in today’s supercomputing. Was an initial target to get several hundred thousand people participating with computers– to perhaps a peak of a million users?
Dan Werthimer (DW): When we launched SETI@home, we thought we’d be lucky to get ten thousand people to participate.
|Dan Werthimer, Chief Scientist, SETI@home|
Credit: Planetary Society
Near some star rather like the Sun there are civilizations with scientific interests and with technical possibilities much greater than those now available to us… No guesswork here is as good as finding the signal. Few will deny the profound importance, practical and philosophical, which the detection of interstellar communications would have. We therefore feel that a discriminating search for signals deserves a considerable effort. The probability of success is difficult to estimate; but if we never search the chance of success is zero.
–Giuseppe Cocconi, Philip Morrison, Nature Magazine, 1959
[Ed. Note: Teraflops refers to speed, or sometimes computer power. For benchmarks, the term 'FLOPS' means floating point operations per second -- an approximate measure of a computer's processing speed. Teraflops refer to a trillion (1012) such operations per second. On a relative scale it is interesting that by comparison, the combined calculating power of all the computers on the planet would only amount to several thousand teraflops, also a unit called a petaflop, making by this coarse measure, the volunteer SETI@home network very approximately 0.1% of the world total].
DW: The volunteers contribute 1000 years of computing time daily.
[Ed. Note: Terabytes refer to information storage. For comparison, a typical video store contains about 8 terabytes of video. The books in the largest library in the world, the U.S. Library of Congress, contain about 20 terabytes of text, making the SETI@home storage more than two and half times greater than the planet's printed volumes].
DW: Our database currently stores about 100 million candidate signals, and we’ve dubbed about 200 of these candidates "interesting".
AM: To allow other science projects to access the unique SETI@home architecture, there is a new plan to roll out a prototype expansion of the screensavers, called BOINC. What is the schedule for rolling out BOINC (Berkeley Open Infrastructure for Network Computing)?
DW: We are beta-testing BOINC now and hope to have it working well towards the end of the year.
BOINC allows volunteers to participate in a number of different projects; volunteers will be able to specify what fraction of their spare CPU cycles should go toward SETI@home (including SETI@home candidate followup with the higher resolution 8 bit data), Astropulse (searching for primordial black holes or radio pulses from ET), Protein Folding (Folding@home at Stanford), Climate Modeling (Oxford, climateprediction.net) and hopefully other projects as well.
BOINC is developed by my colleague Dr. David Anderson. David is a computer scientist and an expert at distributed computing. David developed the SETI@home code and directs both the SETI@home and BOINC projects.
AM: The Planetary Society sponsored some of the early 1998 work to create the world’s largest virtual supercomputer, based on calls with you and David Anderson. From the outset of SETI@home, what was common atmosphere at that time about its ultimate size and workability? Audacious, outlandish or quixotic?
DW: Most people were skeptical we could make it work, or that anybody would sign up. The Planetary Society was the only group willing to provide seed funding, take the risk, and try something unconventional.
|The National Academy of Sciences awarded Sagan the 1994 Public Welfare Medal for "communicating the wonder and importance of science."|
DW: Carl was very excited about SETI@home in the early stages, but unfortunately he never got to see SETI@home grow. We named SETI@home’s server "SAGAN" in Carl’s honor. SAGAN "communicates" daily to millions of participants in a global science project.
The real Sagan communicated science to over 500 million people.
Carl’s wife and collaborator, Ann Druyan, has helped enormously by raising crucial funding for SETI@home.
AM: I have heard the original founding group feels that the goal of SETI–discovery of an intelligent signal–was the unique draw, what got so many people to join in, along with the surprisingly competitive aspects of exactly who could provide the most data units . That became a kind of attractive sport for serious users. Do you think a science project in another field than SETI would have had such a successful roll-out?
DW: Other distributed computing projects have attracted tens of thousands of volunteers.
But SETI@home is the largest; almost all earthlings are interested in the question "Are We Alone?"; people are eager to help.
AM: How did this draw of users competing, this competition itself, become so prominent? Some clubs have formed with up to nearly 6,000 people pooling big computers, with single clubs doing the equivalent of tens of thousands of years of computing time. Was all that expected, or did it just evolve from the usage patterns of the early adopters?
DW: We didn’t expect friendly competition to lead to such craziness.
You can buy work units on eBay — for merely a few hundred dollars, an eBay seller will transfer their work units to your account, so you’ll receive the credit and your name will rise close to the top of the lists on the SETI@home web site.
AM: Let’s talk about the SETI search itself–where all that data is coming from. Currently the Berkeley home server is sending out and receiving back radio data that was originally collected at Arecibo, in Puerto Rico–the world record-holder for single dish size. That is like our planet’s ‘biggest ear’, listening for signals. So for collecting all the radio data that SETI@home is crunching, what percentage of the night sky has been covered from this location– to be subsequently analyzed by SETI@home?
DW: The Arecibo telescope can ‘see’ about a third of the whole sky (the dish covers declinations from -2 to +38 degrees).
We’ve covered 95% of that sky (or about 30% of the whole sky). We’ve observed most beams several times. It’s important to observe the same point in the sky several times because extraterrestrial transmitters might not be on or be visible all the time, and many of our candidate detection algorithms require a signal be repeatable.
AM: Will the SETI@home computers process any future radio data not from the Arecibo Observatory?
|What would an interesting signal look like? The three main types are: pulses, gaussians and triplets. Click here for computing client that distributes the signal processes. Or ‘how to tune in and drop out at Berkeley?’|
"If you had asked me 100 years ago what to look for, I might have said smoke signals." -D. Werthimer
Image Credit: seti@home
DW: Perhaps, but we are more excited about collecting and analyzing data from the upcoming multi-beam receiver at Arecibo – there will be spectacular all-sky surveys using this new 7 beam receiver starting in 2005.
AM: With private donors, the SETI Institute and Berkeley are constructing a telescope that is kind of distributed, like the screensavers, in that it will be an array of 350 commercial satellite dishes, individually 20-foot antennae, that combine their reception to make a huge telescope grid. That is called the Allen Telescope Array planned for 2005 or beyond, correct?
DW: The Allen Telescope Array (ATA) is a giant telescope made from hundreds of small dishes; it’s a joint project of the SETI Institute and the University of California, Berkeley. It’s terrific for targetted SETI searches, and is paving the way for future giant telescopes.
AM: Berkeley has been involved since the beginning of the SETI project called by the acronym, SERENDIP, for the "Search for Extraterrestrial Radio Emissions from Nearby Developed Intelligent Populations". The project is the world’s only "piggyback" SETI system, operating alongside simultaneously conducted conventional radio astronomy observations. How is SERENDIP part of the giant 305-meter (1000-foot) Arecibo radio dish? Is this a continuous data feed that is selective for the radio frequencies that are either most accessible or of most interest to SETI research?
DW: Most observers are lucky to get two days a year on the world’s largest radio telescope, but SERENDIP has a dedicated feed and receiver on Arecibo’s carriage house, enabling us to conduct SETI observations 24 hours a day, year round.
Observing at the same time other scientists are using the telescope. That is what makes the SERENDIP and SETI@home searches so powerful. We developed this "Piggyback SETI" technique in the late 1970′s.
AM: To be clear, one needs to tune in on a SETI source, just like you can tune through the various radio bands — AM, FM, short-wave, ham, police, weather, etc.–, so where exactly on the radio dial does modern SETI and SERENDIP tune in?
DW:The dedicated feed/receiver at Arecibo covers a 100 megahertz (MHz, one million Hertz, or vibration cycles per second) band centered at 1420 MHz (the Hydrogen line).
SERENDIP analyzes the full 100 megahertz (MHz) band available from the receiver.
[Ed. Note: This SETI radio search involves a frequency that is relatively quiet, between 1000 and 10,000 MHz--just above the frequencies used by electronic pagers and some wireless cell phones at 900 MHz. The most abundant molecules are hydrogen, either neutral gas at 1420 MHz or combined with oxygen at 1640 MHz. In the spectrum of background that rains on our planet from interstellar space, this quiet region is called the 'water-hole', because water molecules (necessary for life as we know it terrestrially), has this vibration. Modern SETI efforts began with a paper written by physicists Giuseppe Cocconi and Philip Morrison. They published in the science press in 1959. Cocconi and Morrison suggested that the microwave frequencies between 1000 and 10,000 megahertz would be best suited for interstellar communications. ]
AM: What is the primary challenge for SETI researchers today?
DW: Earthlings are just beginning to learn how to hunt for ET. We can only observe small parts of electromagnetic spectrum, or small parts of the sky, and we are limited in search sensitivity and signals types.
But capabilities are growing exponentially – we used to listen to 100 channels at once (SERENDIP I); now we listen to 168 million channels at once (SERENDIP IV), and we are designing 2 and 20 billion channel spectrometers.
AM: So is it better to look on many narrowly tuned channels, or a broad spectrum? In your experience, technically, is the assumption of another civilization sending a signal as some kind of a transmission distinct from natural radio background (noise), is that a better bet than the particular choices of single, narrow beacon-like transmission?
DW: It’s hard to predict what ET might be doing. My guess is that it’s more likely we’ll discover an artifact of a civilization’s technology rather than a signal directed to us for the purpose of interstellar communication – perhaps we’ll discover navigational beacon, an asteroid radar system, radio signals leaking off their planet, or something completely unexpected.
We should be looking in the visible, infrared and radio for a variety of signal types, and mining data bases from other astronomy surveys, on the lookout for something unusual.
If you had asked me 100 years ago what to look for, I might have said smoke signals.
|Allen Telescope Array (ATA)|
Our Berkeley SETI group is conducting five different searches for radio and optical signals, both pulsed and continuous.
The more variety, the better the chances of finding ET.
AM: So is it a pulse or kind of siren that scores well in SETI searches?
DW: SETI@home covers spectral resolutions from 0.07 Hz to 1 KHz, so we can detect pulses and broad band signals.
AM: For instance, what about signal drift? What happens in cases when both the receiving and the transmitting planet are also moving, if that causes multiple, unknown Doppler drifts? Is this an uncertainty in what is the maximum achievable resolution? One problem with massive Fast-Fourier-Transforms is bin-size (say, discrete 1 hertz (Hz) wide for sensitivity) but those narrow signals have only a six second integration time because the Earth’s motion causes a doppler shift out of this signal.
DW: Because of the massive computing power contributed by the SETI@home volunteers, SETI@home is able to conduct a sensitive search for drifting coherent signals from -50 Hz/sec to +50 Hz/sec (eg: transmitters on the surface of planets or in orbit around planets).
AM: What do you think would make the biggest difference for technically upgrading the search?
DW: No SETI experimenter has searched for short radio pulses before, because it’s extremely computationally intensive. We are launching a new distributed computing search, "Astropulse", which searches for dispersed radio pulses as narrow as 400 nanoseconds (nS). Such pulses might come from extraterrestrial civilizations; but also might come from primordial black holes.
|In a universe brimming with stars, the search is on if life exists elsewhere|
Steven Hawking figured out that black holes evaporate; when the they finally reach zero mass, the black holes are likely to give off a radio pulse. Astropulse will mine the 50 Terabytes of SETI@home data to search for such pulses.
[Ed. Note: This type of pulsed signal is different from those which would be caught by SETI@home. Since the pulses are so fast, they are broad-band signals. Such pulses travelling through the interstellar medium (the thin gas which fills the space between stars in our galaxy) become "dispersed," or stretched out in time. SETI searches can correct for this effect with a specialized algorithm (known as "coherent de-dispersion"), but it is very computation intensive, which is why this is a good distributed computing project.]
AM: Do you have a personal favorite for what you have seen in seti@home data as a false positive? Something like the Ohio State ‘WOW‘ signal that meets all the detection criteria but turned out to have a mundane cause?
DW: All the powerful radio signals have turned out to be radio interference.
We maintain a list of the most interesting candidate signals, and we recently reobserved these at Arecibo, but there’s nothing that sticks way out of the distribution.
AM: What are the plans for the future?
DW: Astropulse, the multibeam sky survey at Arecibo [5-10 times more sensitive than previous surveys], more optical SETI [the search for visible or infrared signals from extraterrestrial civilizations], ever larger bandwidths (SERENDIP V and beyond), the Allen Telescope Array, the SKA (square kilometer array), and onward, funding permitting.
AM: ‘Onward’ seems to be a kind of motto for the thinking behind SETI@home.
DW: Capabilities are doubling every year.
Homo sapiens. What an inventive, invincible species. It’s only been a few million years since they crawled up out of the mud and learned to walk. Puny, defenseless bipeds. …And now, here they are, out among the stars, waiting to begin a new life. Ready to outsit eternity.
Dr. Who, 1963
I’m optimistic and predict that Earthlings will discover extraterrestrial civilizations in the next 50 or 100 years. Don’t hold your breath.
The goal of this vast SETI network is to pick out the one world-changing radio transmission from the billions and billions of possible worlds. No consideration of SETI can be complete without an accounting of how life has prospered, or potentially propagated- the run-down of the chase, strictly by the numbers. The number of stars in the visible universe, for instance, is estimated to be 70 sextillion, or 70,000,000,000,000,000,000,000 [seven followed by twenty-two zeros].
Such a vast population can be compared in a list of the very biggest numbers imaginable, with some terrestrial references borrowed from a combination of science and poetry:
- ten times more than the number of grains of sand on Earth
- eleven times the number of cups of water in all the Earth’s oceans
- ten thousand times the number of wheat kernels that have ever been produced on Earth
- one hundred million times more than the number of ants in all the world
- one hundred million times the dollar value of all the market-priced assets in the world
- ten billion times the number of cells in a human being
- one hundred billion times the number of letters in the 14 million books in the Library of Congress
In the realm of astrobiology, it may be said that most meaningful terrestrial analogies to the number of stars in the known universe are indeed biological: only a fertile biosphere can yield such large numbers. One may ask how many living things the Earth itself can accomodate in its volume. If one cubic inch can hold ten billion animal or plant cells, and if one stacked these cells across both the land and oceans to a thickness of fifteen feet, the planet would be a vast teeming mass of biology–literally, life as far as the eye could see. The thickness of fifteen feet, while extreme overpopulation on the land, is likely an underestimate given the depth of the more three-dimensional ocean biosphere or the realms of winged species. In this way, the ceiling on the carrying capacity of Earth for cellular life is vast, since about ten million times the number of plant or animal cells could pack the planet than the number of stars in the visible universe. Compared to 70 sextillion, the cellular capacity terrestrially is estimated to be what can be called one undecillion, or ten raised to the power of 30.
In addition to The Planetary Society, other major funders include Sun Microsystems, the University of California, Quantum Corp., Fujifilm Computer Products and Network Appliance.
 Popularity is not an easy or uncontroversial thing to assess. For instance, even the most popular music network, Napster, had over 1.5 million simultaneous users logged on and was by far the largest exchange network of its time. Its use peaked however with 13.6 million total users in February 2001 (source: Media Metrix). See also statistics for its successors, like Kazaa. For processing power, as of early 2002, the fastest supercomputer is the Earth Simulator at the Yokohama Institute for Earth Sciences. It is a cluster of 640 custom-designed 8-processor vector processor computers based on the NEC SX-6 architecture (a total of 5120 processors).
Related Web Pages
Terrestrial Planet Finder Home Page
What Does ET Look Like from 40 Light Years Away?
Anybody Out There? Part I
Anybody Out There? Part II
Search for Life in the Universe: Neil deGrasse Tyson Interview
Aliens Depend on Time to Grow Brains
Rare Earth? Are we so special
Stellar Countdown Yields Skymap