Academic Articles & Supplements

An Incomplete History of Wolfram Research, Abridged, Ablated, and (only) a Bit Fabricated

Daniel Lichtblau

Author

Daniel Lichtblau

Title

An Incomplete History of Wolfram Research, Abridged, Ablated, and (only) a Bit Fabricated

Description

An Incomplete History of Wolfram Research, Abridged, Ablated, and (only) a Bit Fabricated

As written by Daniel Lichtblau in August 2009, for the occasion of Stephen Wolfram's fiftieth birthday

Prelude: Antiquity

Wolfram Research rolled out its flagship product, Mathematica 1, in 1988. It was regarded as a marvel of mathematical functionality for that time. Stephen Wolfram, principle founder of the company that happenstance bears his name, claimed that he often wrote 1000 lines of code per day. I don’t doubt that. Well, I do, but only by a modest factor. As the most experienced kernel code spelunker at WRI, I can attest that early Mathematica featured some of the worst software engineering on the planet. About this, the less said the better. Especially for my future at the company. I will only note that the original developers had to divert some of their time to the slaying of mastodons, both for food and office space. Me, I spent those nascent years at a nearby university, locked in mortal combat with an uncooperative dissertation.

Part 1: The Middle Years

It was the 90’s. Stephen was about one year away from finishing his "book" (this was to be the theme of that decade). Mathematica 2.0 emerged. It contained Jerry Keiper’s particularly brilliant flavor of approximate arithmetic on large numbers, something that, outside of Wolfram Research, remains a tad controversial to this day: those of us who know it, deeply appreciate it. I came to WRI in September of ’91. Stephen had, by then, removed himself to Berkeley to work on his opus, and I was one of the early employees to be interviewed by telephone. Some of you may remember the kind that had a cord; I think it was one of those.

In 15 month intervals, approximately, we came out with versions 2.1 and 2.2. Stephen remained busy with his research. I came in to work one day in 1993, only to learn that my computer, a spiffy NeXT box, had been lobotomized to give him more computation cycles. I spent my enforced down time of two-three weeks reading up on the finer points of Gröbner basis computation. I only hope I will some day have an opportunity to voice my disdain for this malappropriation of then-scarce hardware.

Following the release of version 2.2 was, well, nothing. Three-plus years of nothing. As in nada. Bubkes. Product stagnation, bit rot, the works. These were lean years. That is to say, we leaned on our Sales staff to keep the company afloat. Stephen claimed at each annual holiday meeting that the company had showed a profit. We had our suspicions he had developed “A New Kind of Accounting”. Finally, in October of 1996, we unleashed upon a very suspecting public the long-awaited Mathematica version 3. It came with a new automated typesetting user interface, accompanied by a quirky—some might say whimsical— notion of program termination.

Version 3 changed things around for the company. The Kernel Development Group, or whatever we were called back then, had matured, and this showed in the engineering, functionality, and overall capabilities of the new product. And the revamped User Interface was, how shall I describe it? Sublime. Like summer wine, made from freshly squeezed jelly fish.

We were quick to come out with 3.0.1, a “bug fix” release. (Stephen hates that term: “Bug fix, bug fix, bug fix!”) Then we all quietly slunk over to version 4, in the hope that the front end people would discover the niceties of software engineering.

Part 2: The Late Middle Ages

Versions 4 and 5 saw Mathematica getting bigger, badder, and better all at once. No mean feat, if you try to think of software of comparable complexity and scope. Stephen began to push development elsewhere in the company. Cup and tee shirt sales took off, reaching the low four figures annually. On a serious scale, Wolfram funded both MathWorld and work that led to the wonderful web resource

functions.wolfram.com.

It was during this period that the annual office pool finally hit pay dirt, with the emergence of A New Kind of Science. Stephen gradually turned his focus to possible applications. The first to escape the lab was Wolfram Tones, no doubt using some advanced sonic weaponry to stun the guard. It rose all the way to the middle of the Slag Heap Of Abandoned Technology (yes, Stephen, you really need to capitalize that “Of”). Lesson: you cannot hit everything out of the park.

But MathWorld and the Functions site were home runs. Or whatever the equivalent might be in cricket. And version 5.2 of Mathematica, the last release before the rewrite of all things graphical and documentational, was, in its way, also a gem.

Part 2.5: Interlude

Like the proverbial Prince turned into a frog, Stephen seems to enjoy jumping from stone to slippery stone. Duly unchastened from his foray into atonal cacaphony, he cast about for novel technology to develop from Mathematica, with or without NKS. Which brings us to…

Part 3: Renaissance

As version 6 took shape, Mathematica had evolving Dynamic update-&-display capabilities. I use “evolving” in the sense of “documented behavior in flux”; WRI insiders know exactly what is meant here. Mathematica 6 also brought forth the revamped graphics technology that had been a decade in development. And, contrary to the insistence of a certain person whose birthday we now celebrate, it elicited howls of despair over A New Kind of Documentation.

In short order we had two point releases, each bringing tremendous improvements to this already perfect documentation. Then we followed by releasing Mathematica 7, with its image processing, charting, discrete calculus, transcendentally meditative equation solving, and even more substantive improvements to the perfect documentation.

The new Do-It-Yourself dynamic interface widgetry of these recent versions motivated Stephen to invest both his own time, and company resources, in what became the Wolfram Demonstrations site. As with MathWorld and the Functions site, this was and remains pure philanthropy on the part of Wolfram. And let me now give credit where due. It has been said that the Demonstrations site, with all its freely accessible, educational, interactive manipulations—hey, they pay me to advertise this stuff every chance I get—was motivated by Stephen having four young children. This was at the time the project began, around 2006. Now their total age is 35, the oldest is seven years older than the youngest, and if he swims upstream at one mile per hour, when will the next one catch the train heading to Buffalo at 80 miles per hour? Wait a moment while I start up Mathematica...

Where was I? Oh yes, the Wolfram funded free sites. They all feature a quiet excellence that even critics do not deny. Though there have been sporadic complaints about the so-called Middle Manager of Demonstrations, for biting people who come into his office.

And then, following long, intense but very quiet development—and absolutely no flashy pre-release hype or media circus—came Wolfram|Alpha. As with almost everything else at this company except maybe holiday bonuses, it has seen more than its share of controversy. No matter. It is stunning in scope, it grows by leaps, bounds, and splats, and frequently it delivers surprising results (read: “It often works”). As we head into the next school year, I wonder how much homework traffic the servers will see. That, I suspect, is but the tip of the ice cube. I mean berg. Iceberg. The tip. Of the iceberg. Small part showing, big part submerged. You get the idea.

Epilogue: Into the Future

Don't ask me what comes next: when it comes to finding water, I'm a cactus, not a divining rod. So I have but little idea of what will follow for Wolfram Research. Ask Stephen. Chances are, he’ll be as stubbornly wrong as ever. And whatever emerges will still be brilliant.

Cite this as: Daniel Lichtblau, "An Incomplete History of Wolfram Research, Abridged, Ablated, and (only) a Bit Fabricated" from the Notebook Archive (2009), https://notebookarchive.org/2018-07-6z259qd