George Marsaglia

1924 - 2011

George Marsaglia: The Architect of Modern Randomness (1924–2011)

In the digital age, randomness is the invisible engine of the modern world. It secures our bank transactions, powers complex weather simulations, and ensures the fairness of online gaming. For decades, the undisputed master of this mathematical frontier was George Marsaglia. A mathematician and computer scientist whose career spanned the transition from vacuum tubes to modern supercomputing, Marsaglia’s work fundamentally changed how we generate and test the "random" numbers that computers produce.

1. Biography: From the Navy to the Ivory Tower

George Marsaglia was born on February 12, 1924, in Denver, Colorado. Like many of his generation, his early adulthood was shaped by World War II; he served in the U.S. Navy from 1943 to 1946. Upon his return, he pursued a rigorous mathematical education at Ohio State University, where he earned his B.S. (1948), M.S. (1949), and Ph.D. (1950) in just a few years. His doctoral research, supervised by the noted statistician Henry Mann, focused on the power of rank tests.

Marsaglia’s career was a blend of high-level industrial research and prestigious academic appointments. He spent significant time at the Boeing Scientific Research Laboratories in the 1960s, a hub for computational innovation. His academic journey took him to McGill University and Washington State University, but he spent the final and perhaps most productive decades of his career at Florida State University (FSU). At FSU, he served as a Professor of Statistics and Computer Science, eventually becoming Professor Emeritus until his death on February 15, 2011, in Tallahassee.

2. Major Contributions: Finding Order in Chaos

Marsaglia’s primary contribution was the rigorous analysis and improvement of Pseudo-Random Number Generators (PRNGs). Since computers are deterministic machines, they cannot produce truly random numbers; they use mathematical formulas to mimic randomness. Marsaglia proved that many of these formulas were dangerously flawed.

The Marsaglia Effect (1968): This was his "bombshell" moment. In a landmark paper, he demonstrated that the most common method of generating random numbers at the time—Linear Congruential Generators (LCGs)—produced numbers that, when plotted in multi-dimensional space, fell onto a relatively small number of parallel planes. He famously summarized this with the phrase:
"Random numbers fall mainly in the planes."
This meant that for high-dimensional simulations (like physics or finance), the "randomness" was an illusion that could lead to massive errors.
The Ziggurat Algorithm: Marsaglia developed this highly efficient method for generating samples from non-uniform probability distributions (specifically the Normal and Exponential distributions). It remains one of the fastest algorithms for this purpose and is still used in software libraries today.
The DIEHARD Battery of Tests: Realizing that developers needed a way to "grade" their random numbers, Marsaglia created the DIEHARD Battery. This was a suite of 15 rigorous statistical tests designed to sniff out patterns in supposedly random sequences. For years, passing the DIEHARD tests was the gold standard for any new PRNG.
Algorithm Development (KISS, Xorshift, MWC): He developed several famous families of generators, including Keep It Simple Stupid (KISS), Multiply-With-Carry (MWC), and Xorshift. Xorshift, in particular, is celebrated for its extreme speed and simplicity, requiring only bitwise operations.

3. Notable Publications

Marsaglia was a prolific writer whose papers often had a direct, almost defiant tone toward sloppy mathematics.

"Random numbers fall mainly in the planes" (1968, Science): His most influential work, which forced the scientific community to rethink how computers simulated reality.
"The Ziggurat Method for Generating Random Variables" (2000, Journal of Statistical Software): Co-authored with Wai Wan Tsang, this provided the definitive guide to his high-speed sampling method.
"Xorshift RNGs" (2003, Journal of Statistical Software): A late-career masterpiece that introduced a new class of incredibly fast generators that are still widely implemented in modern programming languages.
"A Current View of Random Number Generators" (1985): A comprehensive overview that served as a textbook for a generation of computational statisticians.

4. Awards and Recognition

While Marsaglia did not seek the limelight, his peers recognized him as a titan of computational statistics.

Fellow of the American Statistical Association (ASA): An honor reserved for those who have made outstanding contributions to statistical science.
Fulbright Scholar: He served as a Fulbright researcher at the University of Amsterdam.
The "Marsaglia CD-ROM": In 1995, FSU and the National Science Foundation recognized his work by helping him distribute a famous CD-ROM containing 4.8 billion "truly random" bits, produced by combining various hardware and software sources. It was a landmark resource for researchers at the time.

5. Impact and Legacy

Marsaglia’s legacy is embedded in the code we use every day. If you use the statistical language R, the Python library NumPy, or C++ standard libraries, you are likely using algorithms either designed by Marsaglia or designed to fix the problems he identified.

His work on the "lattice structure" of random numbers didn't just improve simulations; it had implications for cryptography. By showing how predictable certain sequences were, he helped define the requirements for cryptographically secure pseudo-random number generators (CSPRNGs), which protect modern internet traffic.

6. Collaborations

Marsaglia was a central figure in a network of computational mathematicians.

Wai Wan Tsang: His most frequent late-career collaborator, with whom he perfected the Ziggurat and various MWC algorithms.
Arif Zaman: A colleague at FSU with whom he developed the "Add-with-carry" and "Subtract-with-borrow" generators, which allowed for incredibly long periods (the number of steps before a sequence repeats).
The Computing Industry: His time at Boeing and his consulting with various tech firms ensured that his theoretical work was always grounded in practical, high-performance computing needs.

7. Lesser-Known Facts

The "Mother of All" Generator: Marsaglia had a penchant for colorful names. He developed a generator he called "The Mother of All Random Number Generators," a name that reflected both its robustness and his own dry sense of humor.
The Physicality of Randomness: Before the internet made data sharing easy, Marsaglia was known for mailing physical tapes and disks of random bits to researchers around the world who didn't have the computing power to generate high-quality randomness themselves.
A Late-Life Innovator: Unlike many mathematicians who do their best work in their 20s, Marsaglia published some of his most influential papers—including the Xorshift paper—well into his 70s. He remained intellectually sharp and active in the research community until his final days.