CNET también está disponible en español.

Ir a español

Don't show this again

Christmas Gift Guide
Internet

The quest for near-perfect compression

The CEO of start-up ZeoSync says he has hit on a unique method that successfully solves a software problem that many consider unsolvable. Skeptics abound.

In a tiny office in West Palm Beach, Fla., a handful of clunky computers are crunching through a software problem that many consider unsolvable.

These are the offices of ZeoSync, a start-up that made a name for itself in January by claiming to have discovered a way to shrink virtually any digital file to a hundredth of its size--and then restore the file to its original size without error.

If true, it would signify a huge leap forward in computer science, comparable to the invention of a water-powered engine for automakers or cold fusion for power companies. With such a technology, a full-length feature film could be sent easily over a dial-up modem or stored on a floppy disk, for example. The discovery would ultimately transform data storage, networking and virtually any computing task.

But as ZeoSync hunkers down to test its claims, few believe they'll emerge successful. The company is just one in a long list of seekers after this technological El Dorado. But just as with the mythological city of gold, few believe perfect compression exists, or will ever be attainable.

Eric Scheirer, a former Forrester Research analyst who developed a significant portion of the MPEG-4 standard for compressing audio, doesn't mince words when talking about ZeoSync's compression challenge.

"It's impossible," Scheirer said. "This comes up over and over again...The right analogy is a perpetual motion machine."

As with any problem where the stakes are high enough, ZeoSync's claims are drawing scrutiny. A small group of passionate online researchers have dedicated themselves to debunking ZeoSync's theory, as has been done with previous claims of near-perfect compression.

ZeoSync now says it's close to testing the technology with a handful of big-name chipmakers, including Intel, although a representative for Intel Labs could not confirm this.

Some of the company's public statements have given critics more fuel for their fire, however. An early press release offered a list of technical advisors, some of whom have admitted publicly that they had no connection with the company, or could not verify its claims.

ZeoSync Chief Executive Peter St. George admits to some misstatements in past releases. But he vigorously defends the basic validity of his findings, and promises to soon provide independent verification.

Critics have missed the point, he said. St. George isn't refuting the mathematical laws that prove perfect compression using traditional techniques is impossible. Rather--having dedicated 12 years to the project--he says he has hit on a unique method that successfully solves the problem.

"All the skeptics are absolutely right," St. George said in a telephone interview with CNET News.com. "But they never anticipated that we would sidestep (the problem)...This is the natural course of evolution in science."

The incredible shrinking machine
Data compression is the science of shrinking. Compression makes the data that creates images, sounds, video and even text small enough to be sent through telephone wires--the network of the Internet. Without data compression, the Web simply wouldn't exist.

In "lossless" compression, used by programs such as WinZip, it's possible to shrink files to a certain size and then recover them without any degradation, or data loss. To compress, the software will pull out repeating bits of data. For example: In the sentence, "The crook took a long look at the book," every "ook" could be replaced by a "1."

Basic mathematical theory, however, limits such a technique. Once you take out every repeating chunk, you wind up with a random string of data. For the past 50 years, most mathematicians have come to agree that completely random data can't be further compressed.

Many people have claimed to come up with ideas that could shrink any file--including a random string of digits--down to a fraction of its size. In every case they've been proven wrong. The skeptic community is now after ZeoSync and St. George, calling his 12-year quest for the perfect data-shrinking machine a digital version of the Don Quixote story.


Gartner analyst Martin Reynolds says that "lossless" compression--the kind used for data files--is limited by the need to perfectly restore every bit of data.

see commentary

Robert Bristow-Johnson, an engineer at digital audio company Wave Mechanics, is one of those critics. He recently sent a letter to the United States Patent and Trademark Office warning them about ZeoSync's claims, saying that giving the company a patent would be as foolhardy as patenting a perpetual motion machine.

"Even I know better," said Bristow-Johnson. "There should be hundreds of thousands of people that know better. There should be dozens of patent examiners who should know better."

St. George might be considered an unlikely candidate to discover a mathematical breakthrough. First working as a vice president for technology services at a financial services company, he shifted gears to become an entrepreneur. He served as CEO of three separate but now-defunct companies before starting ZeoSync in 1999. On his resume, St. George also says he's served as a "bandwidth delivery solutions consultant" to second- and third-world countries.

Despite his pedigree, the entrepreneur said the main problem is critics don't truly understand what he's working on.

The CEO says he's simply discovered a different way of compressing data, calling it "compacting" or "filtering." Where other methods chop off a string of binary digits representing a picture or song or text, he looks at the entire string as one huge number. This number can then be represented as a separate mathematical expression, he said.

Since typical digital files are represented by hundreds of thousands of zeros and ones, St. George said that crunching numbers using his method requires substantial computing power. For now, ZeoSync's shoestring budget has meant testing on computers using old Pentium II chips--although partners may help pitch in for more powerful computers down the line.

At the time of the company's first press release, it took more than a day to squash a random 128-bit file--about 16 letters in ASCII, the most basic computer character set--into just 100 bits.

It's not a product yet, St. George insists. But the fact that the "compacting" can be done at all means ZeoSync has made a theoretical breakthrough, he said.

Question marks
Skeptics simply want proof of the breakthrough. St. George said the company is close to making deals with Intel and other chipmakers for testing on state-of-the-art machines. Intel would not independently confirm its participation.

In the absence of technical proof, critics have taken a hard look at the company, which is in the process of trying to raise $40 million in a private stock offering. Some non-technical questions about the company have emerged.

Early on, the company provided a long list of distinguished scientists whom it said had served as technical advisors or consultants on the project. A few weeks later, this list was chopped to a third of its size.

Several of the professors on the original list contacted by telephone or e-mail now say they had little or nothing to do with the company.

"I had never even heard of them before they made the announcement," said University of Washington Professor Richard Stanley, one of the people dropped from ZeoSync's list of advisors. "I'd never talked to them. I just assumed it was some kind of hoax or scam or something."

St. George said the original list included clerical errors, and some names of people with whom the company was in contract negotiations. He is issuing a personal apology for the mistakes, he said.

According to the company's financial prospectus, ZeoSync also is involved in a lawsuit with an early investor, a company called First Frontier Capital. ZeoSync has said it will return more than $2 million in early funding. In addition, the company said it plans to file suit against the venture capital firm and an Indian software company over trade secret issues.

Skeptics expect that St. George and his company will fade into history along with other seekers of the impossible. He'd like to prove them wrong.

"I don't blame people," St. George said. "The day the Wright Brothers flew, there was a group of scientists nearby saying in concert that man couldn't fly. It's human nature to describe the world in terms of lack, in terms of what we can't do."