You probably haven't thought about linear algebra since college. Honestly, you might have spent that entire semester wondering when you'd ever use a matrix in the real world.
But every time a commercial airplane lands safely, an iPhone processes a photo, or a financial analyst prices a complex derivative, a hidden piece of software is doing the heavy lifting. That software is MATLAB. And the man who built its very first version, Cleve Moler, passed away on May 20, 2026, at the age of 86.
Moler didn't plan to build a massive software empire. He wasn't trying to change the tech world or become a billionaire corporate executive. He just wanted to help his students pass their math exams without pulling their hair out over tedious programming bugs.
It's a classic example of an accidental tech revolution. By trying to solve a tiny classroom friction point, Moler accidentally ended up giving scientists and engineers the keys to modern computing.
The Fortran Nightmare and the Birth of a Matrix Lab
In the 1970s, if you wanted a computer to solve complex mathematical equations, you had to write the code in Fortran.
Fortran was powerful, but it was brutal. If you made a single mistake with a comma or a deck of punch cards, the whole program crashed. Moler was a professor of mathematics and computer science at the University of New Mexico. He had spent years collaborating with other top minds to build LINPACK and EISPACK—highly advanced Fortran code libraries designed for matrix calculations.
The libraries worked beautifully. The problem? His students couldn't use them.
The friction was too high. Students spent 90% of their time debugging messy Fortran syntax and only 10% actually learning the underlying mathematics. Moler hated seeing that waste of intellectual energy.
So, around 1977, he sat down to write a simple, interactive terminal program. He called it MATLAB, short for "Matrix Laboratory."
His goal was straightforward. He wanted a tool where a student could type a simple matrix, hit enter, and get an immediate answer without compiling a single line of Fortran. He didn't sell it. He gave it away for free. He copied it onto floppy disks for anyone who asked.
The Pivot From Classroom Tool to Global Infrastructure
Moler thought MATLAB would remain a niche academic utility. The computer science establishment at the time didn't think much of it either. To them, it was just a simplified wrapper for underlying code libraries, not a "real" programming language.
They completely missed the point.
Engineers didn't want to be computer scientists; they wanted to solve engineering problems. In 1983, an engineer named Jack Little saw the software during a visit to Stanford, where Moler was teaching. Little recognized something huge: the newly launched IBM PC was about to put computing power on every desk in America, and those desktop machines would need software that engineers could actually understand.
Little teamed up with Moler and another programmer, Steve Bangert. They rewrote MATLAB in C, added control flow features, and founded MathWorks in 1984.
The timing was impeccable. The aerospace industry grabbed it immediately. Control systems engineers realized they could model the flight dynamics of a spacecraft or an aircraft in minutes instead of spending weeks writing custom code.
From there, the software spread like wildfire into:
- Automotive design: Simulating braking systems and suspension physics.
- Genetics: Processing massive datasets of gene sequences.
- Telecommunications: Designing the signal processing algorithms that power cellular networks.
- Finance: Running risk assessment models on Wall Street.
The Coined Phrase That Defined Modern Supercomputing
Moler wasn't just a one-trick pony with MATLAB. His career spanned across academia and corporate hardware development. During a stint working with Intel's Hypercube division and Ardent Computer Corporation in the late 1980s, he focused heavily on parallel computing.
It was during this era that Moler coined a phrase you still hear in computer science departments and AI labs today: "embarrassingly parallel."
The term describes a computational problem that can be easily broken down into a bunch of smaller, independent tasks that run at the same time without needing to talk to each other. Think of it like 100 people painting 100 separate fence posts. No one needs to wait for the other person to finish. Today, this exact concept forms the bedrock of how modern graphics processing units (GPUs) train massive artificial intelligence networks.
Why His Legacy Still Shapes Your Daily Life
It's easy to look at software like Python, R, or Julia and assume MATLAB is a relic of the past. Python is free, open-source, and dominates the current AI conversation.
But don't underestimate how deeply entrenched Moler's creation remains.
The elegance of his design was that it treated everything as an array or a matrix by default. You didn't have to define data types or set up complex loops just to add two sets of numbers together. That syntax choice changed the mental model of how a generation of scientists interacted with hardware.
Take a look at the classic MATLAB logo. It looks like a deformed, wavy fluid surface. That shape isn't just a random graphic design asset; it's a visual representation of the wave equation inside an L-shaped domain. Moler calculated the underlying math for that specific shape during a research stay at ETH Zurich back in 1965.
That logo serves as a permanent reminder of what Moler stood for: turning cold, abstract numbers into clear, visual intuition.
What to Do Next if You Want to Learn Numerical Computing
If you want to honor Moler's legacy, the best thing you can do is start building things yourself. You don't need a corporate license to experiment with matrix laboratory concepts.
- Try an open-source alternative: If you don't have access to a university or corporate account for MATLAB, download GNU Octave. It uses a nearly identical syntax and lets you experiment with matrix math for free.
- Read his open textbooks: Before his passing, Moler wrote two excellent online textbooks that MathWorks still hosts for free: Numerical Computing with MATLAB and Experiments with MATLAB. They're written in his signature conversational, accessible style.
- Shift your programming mindset: Next time you write code to handle a large dataset, don't immediately jump to writing nested loops. Try to frame the problem as a matrix operation. It's faster, cleaner, and exactly how Cleve Moler taught the world to think.
