By all accounts, David and Gregory Chudnovsky are brilliant number theorists. Their quest to calculate pi to the greatest number of decimal places was warmly chronicled in a long New Yorker profile in 1992 titled "The Mountains of Pi." Three years later, New York Magazine ranked the Chudnovskys among the 100 smartest New Yorkers. And two years after that, Esquire included the brothers (as a unit) in a grouping of "The 100 Best People in the World."
The brothers have all sorts of inventions worthy of both patent protection and commercialization. There is only one problem. They are not fans of the patent system and are reluctant capitalists. This is a story about how the Chudnovskys came to believe in the value of patenting and commercializing their work. It is not an objectively told story; I am their patent lawyer.
Born in Kiev after World War II, the brothers discovered the joys of mathematics as children. They earned Ph.D.s by their mid-20s and international reputations shortly thereafter. Both have a well-earned distrust of authority. At age 12, Gregory developed myasthenia gravis, an autoimmune disorder that leaves him bedridden on most days. In the 1970s, they tried to leave the Soviet Union for the United States so that Gregory could receive medical treatment. But the KGB had other plans for the brothers, who are Jewish. David says that KGB operatives beat him and cracked his skull and also attacked his parents. Only through the intervention of U.S. Sen. Henry "Scoop" Jackson were they able to emigrate, eventually settling in New York in 1978.
While many high-level mathematicians in the 1980s gravitated to Wall Street, where they revolutionized high finance, the Chudnovskys headed for academia, living in a modest apartment near Columbia University. Unable to acquire tenured teaching jobs -- Gregory does not often leave his apartment, and David usually stays by his brother's side -- the brothers survived on government grants, the kindness of their employed wives, and a research affiliation with Columbia University.
During that time the brothers built a sprawling, Rube Goldberg-style supercomputer for less than $100,000. They ran the machine, which they named "m zero," out of Gregory's top-floor apartment. The computer's 25 cooling fans ran day and night to prevent overheating, taxing the apartment's circuitry. Without an academic agenda to pursue, the brothers set about using m zero to win an international competition, the object of which was to calculate pi -- the number of times a circle's diameter travels around its circumference.
Most contemporary mathematicians believe pi to be a transcendental number capable of generating decimal places into infinity without displaying any regular or predictable pattern. For several years the Chudnovskys and their m zero engaged in a somewhat quixotic but captivating search for a pattern. But by 1991 the Chudnovskys had calculated pi to 2,260,321,363 decimal places -- a numeral that, set in normal type, would stretch from Los Angeles to New York. Their lead established, the brothers gave up the pi chase for other projects.
M zero might have transformed the world of supercomputing if they had protected their invention. Surely the Chudnovskys could have secured patents on aspects of their machine and lived a life of comfort. But their interests prompted them to move on to other things.
Drawing on their experience with m zero, Gregory and David invented and tested in the early and mid-1990s a novel switching network that performed faster than even their expectations. But as with m zero, they did not patent the invention. The prototype switch sat on an end table in Gregory's apartment for five years, while a comparable invention from someone else later helped propel the Internet explosion.
Why no patents? Gregory has said more than once that he does not like intellectual property. His reasons are practical. Patents are expensive. They are time-consuming and a bother to write. The benefits of the protection are often outweighed by the drawbacks of disclosure. Patents can be circumvented too easily. Large companies with huge legal budgets will find ways to break them or fight over them in court forever. Patents don't work, because licensees cheat on royalties; they cost too much to litigate; and judgments can be difficult to collect. Finally, they are pointless, because technology changes so quickly in the computer field.
The Chudnovskys do not believe that intellectual creations naturally belong in the public domain. The brothers are conservative, hard-headed, not overly self-effacing, and not especially philanthropic men who believe in looking out for themselves and their families first. Their attitudes seem logical and reasonable -- the question, then, is how to make their ideas profitable.
Even in the current, battered stock market, there are angel investors and venture capitalists looking for people like the brothers with interesting and useful projects to underwrite. There have been quite a few potential investors in the Chudnovskys' lives during the past ten years. Except for one small hardware venture, none of them took out a checkbook.
The typical stumbling block is that investors are looking for assets to sponsor, not people. Inventors are a sometimes cranky, prickly, eccentric and paranoid bunch. A patent is different. A contractual right to own a chunk of a patent is an asset. If a patent goes beyond novelty to pioneering quality, it can be an absolute bull's-eye. It lands on the balance sheet, and, in due course, shows up on an income statement. But all of this must be agreed on in advance and put in writing. Anyone who has read "Moby-Dick" knows you divide the whale before you leave port.
Inventors often see this last point differently. By instinct, they consider themselves to be the asset that investors seek. And they are confident of their own judgment as to how to allocate returns fairly between inventor and investor. Some of these generalizations can be fairly applied to the Chudnovskys. Their embrace of capitalist principles is qualified in other ways. Gregory abhors debt. Debt, to him, is not a financing technique; it is a surrender of fragile, hard-won liberty. Every new business failure in the media confirms his intuitive conviction that a startup enterprise should borrow nothing.
The Chudnovskys are equally reluctant to issue equity. Their instincts and experience tell them there are always master puppeteers somewhere, manipulating the system.
None of this suggests a happy outcome for the Chudnovskys' inventions. Yet on April 30, the brothers were issued a patent for a "multibank, fault-tolerant, high-performance memory addressing system and method." Given the brothers' beliefs about patents, capitalism, and investors, how did this happen? The answer lies initially with the persuasive powers of the late S. Leslie Misrock.
In November 1998 the Chudnovskys accompanied a mathematician friend on a visit to New York's Pennie & Edmonds. The mathematician wanted to patent a geonomics discovery. Misrock, a well-known and charismatic technology impresario, was seriously ill at the time with prostate cancer. He happened to walk into the conference room during our discussion. Without prompting, Leslie went into a soliloquy about his passion for patents and how his health problems had contributed to his interest in medical patents. He told David and Gregory that, even at this stage of his life, he was energized by working with younger people on new and interesting technologies. I believe that Misrock inspired David and Gregory to understand that important patents can change industries and have positive effects on peoples' lives on many dimensions.
The Chudnovskys started thinking again about the possibilities of patenting. One thing led to another, and they let it be known that there were several inventions in their pipeline, one nearly ready for patenting.
The Chudnovskys outlined an invention that, like all of their others, turns on number theory to make computers run faster or cheaper or both. The first part of the invention speeds computer memory chips called DRAMs (dynamic random access memory). These chips are made of multiple banks of memory cells, and when one of the banks is accessed more frequently, there is a logjam and loss of efficiency.
To use an analogy from a bricks-and-mortar bank, several customers may be lined up in front of one teller, while few are lined up in front of the remaining tellers. Banks have gotten around this problem by creating a common teller line that serves all of the tellers. The Chudnovskys' solution is somewhat similar. They designed high-efficiency circuitry for memory address translation (also termed "remapping") that smooths distribution across the memory banks. The computer is thus tricked into avoiding an overused memory bank. The essence of the invention is that the actual physical memory addresses no longer correspond to the "logical" addresses with which the processor believes it is communicating. The logjam disappears.
This remapping trick has a second advantage. Normally if a memory bank cell fails, the system will likely crash or suffer memory loss of at least 50 percent. But remapping allows the computer to route around the failure without an abrupt reduction of memory capacity. The Chudnovskys believe that this fault-tolerance feature will become increasingly important as the industry migrates to so-called systems on a chip. During manufacturing it will no longer be necessary to reject expensive silicon wafers with only small defects. The Chudnovskys estimate that the savings in production cost may be in the range of 20-30 percent.
Misrock took the brothers on as clients, and I became their lead lawyer. (Misrock died in 2001, before the patent ultimately issued.) I am a litigator. The firm wanted a litigator in this role so that a patent application could withstand not just the scrutiny of the U.S. Patent and Trademark Office examiners, but also an attack in court.
I needed to find the right Pennie & Edmonds attorney to work with the brothers in drafting the application. The firm has many computer science specialists, comfortable with higher mathematics. David in particular wanted a lawyer who understood number theory. I introduced to the Chudnovskys three candidates: a partner and two senior associates. The brothers are known as difficult taskmasters; they are accustomed to collaborating mainly with each other. They ultimately picked senior associate Henry Lebowitz to write the patent application.
My interest in the brothers became personal as well as professional. I decided to explore ways to help them find financing. I know a group of private investors, composed mainly of active and retired lawyers. The investors agreed to meet the brothers. Members of the group, which is headed by retired Davis Polk & Wardwell partner Richard Spizzirri, listened to the Chudnovskys explain their invention. They liked its potential for becoming an industry standard.
Neither the firm nor the investor group did much for the next several months. In April 1999 the pressure of an impending publication date for a Chudnovsky paper brought the parties back. We stitched together eight investors, including Spizzirri and me. An agreement that might have taken a year to negotiate with a venture capital firm was wrapped up in three weeks. The agreement provided for funding patents on two inventions, the DRAM invention and another project that was not so far along.
It took longer than that to get the patent application into acceptable form. The claims underwent a couple of false starts and countless drafts. But on Dec. 27, 1999 -- two days after the birth of Gregory's first child -- we filed the brothers' first patent application.
Other than routine filings, nothing happened until October 2001, when Pennie & Edmonds received a preliminary decision from the patent office. The good news was that the broadest and final claim had been allowed. The bad news was that the 45 preceding narrower claims had been rejected.
Thankfully, the examiner had not discovered prior art -- published information describing the invention or some aspect of it. Rather, he wanted certain terms and mathematical concepts explained more clearly. Lebowitz, himself a former patent office examiner, resolved the matter fairly quickly. On Jan. 2, the patent office said it would approve the patent. The patent itself was issued in April.
Now it is time to make a deal. The natural licensees or purchasers are the comparative handful of manufacturers in the commodity DRAM chip industry. There are five major producers. Those companies account for 80 percent of the market. We don't know if all of the producers might want to take a license or whether just a couple of them will want to use the patent to try to establish a competitive edge. There are intriguing questions about whether to carve up the rights under the patent, doling out licenses based on type of use. But we can go only so far without the friction of an actual negotiation. That is where the brothers are now headed.
In the meantime, Gregory says he has in mind a second patent on a novel approach to data organization. With the hurdle of the first patent crossed, the world may see a string of Chudnovsky inventions in years to come.