Firm Touts 'Perfect Compression'
2:00 a.m. Jan. 16, 2002 PST
WASHINGTON -- Physicists do not question the laws of thermodynamics.
Chemistry researchers unwaveringly cite Boyle's Law to describe the relationship between gas pressure and temperature.
Computer scientists also have their own fundamental laws, perhaps not as
well known, but arguably even more solid. One of those laws says a perfect compression mechanism is impossible.
A slightly expanded version of that law says it is
mathematically impossible to write a computer program that can compress all files by at least one bit. Sure, it's possible
to write a program to compress typical data by far more
than one bit -- that assignment is commonly handed to computer science sophomores, and the technique is used in
.jpg and .zip files.
But those general techniques, while useful, don't work on all
files; otherwise, you could repeatedly compress a .zip, .gzip or .sit file to nothingness. Put another way, compression
techniques can't work with random data that follow no known patterns.
So when a little-known company named ZeoSync announced last week it had achieved
perfect compression -- a breakthrough that would be a bombshell roughly as big as e=mc2 --
it was greeted with derision. Their
press release was roundly mocked for having more
than a Walt Disney store, not to mention the more serious sin of being devoid of
any technical content or evidence of
A Reuters article was far more credulous, saying in the lead
paragraph that "a Florida research startup working with a team of renowned mathematicians said on Monday it had
achieved a breakthrough that overcomes the previously known limits of compression used to
store and transmit
For compression buffs, responding to such assertions ranks
somewhere between a teeth-gnashing migraine and a full-contact sport.
The comp.compression FAQ has an entire section devoted
to debunking: "From time to time some people claim to have invented a new algorithm for (perfect compression). Such
algorithms are claimed to compress random data and to be applicable recursively; that is,
applying the compressor to the compressed output of the previous run, possibly multiple
Several comp.compression fans have even offered rewards
up to $5,000 for independently verifiable proof of perfect compression. They've never been claimed.
Perfect compression, or even compression of a few hundred
times -- what ZeoSync claims -- would revolutionize the storage, broadband and digital entertainment industry. It
would mean that modems would be as fast as DSL, and DSL
speeds would be blinding. A 40-GB hard drive would hold a terabyte, and so on.
That is, if it works.
During a telephone interview this week, ZeoSync chairman
and CEO Peter St. George refused to answer any specific questions about his company's product. He could not point
to one independent researcher who had reviewed his product but promised an announcement as
Wired News: When did you start working on this technology?
Peter St. George: I started developing the technology
about a dozen years ago. I worked on this one problem for 12 years consecutively. This is a project that I dedicated
my life to a dozen years ago.
WN: What respected independent reviewer can verify your claim?
PSG: We're going to be announcing later this year that we're going to be setting up a global
test bed for scientists
around the world to participate in.
WN: Let's go into the details. Tell me how it works. It can compress random data?
PSG: If you say absolutely random, it's going to be very hard to agree what absolutely
random is. We can compress
sequences that Stanford University professor Don Knuth described as uncompressible (in his Art of Computer Programming).
WN: Are you saying you can compress random data by 100 times?
PSG: What happens if you take the existing compression technologies of the world, they all
have the principal method
of mapping and then secondly encoding. Once the technology is operating to its maximum efficiency, you
cannot send it through for a second or third pass. Because we have the ability to handle these random sequences, we
can encode it once and then go around for a second pass. We have the ability to perform operations that aren't
available with traditional compression technology. That allows us a magnitude of performance
with traditional techniques.
WN: You have this working right now?
PSG: It's only operating on a limited bitstream of a few hundred bits. It needs a lot of work to
make it a
WN: How do you get around the conventional wisdom that says simple mathematics says
PSG: That's what's being proposed as the reason that our technology won't work. We plan to
issue head on. What hasn't been previously proven, we're proving. We can compress every single
permutation of an N-member set. These are going to be the details that we're going to be announcing in a few
If you have a propeller-driven airplane, nobody will deny that that engine will lift you up into
the air. Now
you have a turbine-driven jet engine. They operate by completely different methodologies. With our
technology, you have to have a minimum set of base two binary carriers to create a multidimensional
construct. Once that construct has been created, we can create a random sequence, a pattern sequence. It
WN: You say base two binary carriers -- don't you just mean a bit?
PSG: Yes, base two is synonymous with binary. You need at least 100 bits, let's say, to
multidimensional construct. Everything in an N-member set can be expressed in an
N-1 set. You can
reconstruct the H set without losing a bit in the process.
PSG: That's what we're going to be asking the world to wait and see.
WN: Why didn't you submit this to be peer-reviewed and published in a conference
PSG: You can't have a scientific test without us describing how it all works. In a peer-review
you have to describe the technology.
The thing about it is we've invested a lot of time and energy (and don't want to) describe the
fundamentals of the technology yet. We won't be bringing it to the scientific community for some time
yet. We'll have an announcement on Wednesday or Thursday.
WN: What will you say?
PSG: I'm not going to describe this today because it's so substantial. I'm going to let the
themselves describe it. You've got my official statements for now.
WN: You're saying that you can compress an arbitrary amount of data, say one terabyte,
down to a few
PSG: You're asking questions that I can't answer. I will answer these as we go. It's like
asking someone who
flew the airplane for the first time how many passengers will fit. I've got a group of people who have
financed this research for the last 10 years or so. We have proprietary patents pending (with 50 to be filed
this year). We're addressing these questions through the best way we know how.
WN: In your demo, will someone upload a file and get a compressed version back? Will they
PSG: You get compressed stuff on your machine. It will satisfy everyone, I'm sure. It will be
I have one quote I'd like to share with you: "The person who says it cannot be done should
the person doing it." We will make the demonstration of the technology available
as quickly as we can get