Today, we recognize these and many other signs of genius, but when he died, Bach was not especially revered. He spent the last twenty-seven years of his life as the cantor at St. Thomas Church in Leipzig, composing music for weekly services attended by only a few hundred parishioners. His handwritten originals were worth something as music, but only at prices comparable to any church cantor of the era, and unfortunately, his descendants were more often in need of money than legacy.
Bach’s widow, Anna Magdalena, still had a number of younger children to care for, so after his death she traded her portion of her husband’s music back to St. Thomas Church, in exchange for an extra six months in the cantor’s residence. The church made formal copies for republishing, so most of the songs themselves survive, but they had no particular use for the originals, and eventually began selling them as scrap paper to wrap fish and other market goods in.
Another collection of Bach’s music went to his adult son, Carl Philipp Emanuel, who was already one of the most respected performers of his day—much more so than his father, at the time. C. P. E. Bach, as he was known, was personal chamber musician to Frederick the Great, and his work was praised by no less than Mozart, Beethoven, and Haydn. He had the financial stability to protect what he was given, and the wisdom to recognize his father’s greatness when no one else did. Nearly all of the pages bequeathed to C. P. E. Bach are held in museums today.
The final stack of compositions, however, went to Bach’s eldest and least reputable son. Like his brother, Wilhelm Friedemann Bach was a talented musician, and he taught many pupils who would go on to be famous composers in their own right. But a string of conflicts with his employers and alleged problems with alcohol left him perpetually in debt and on the move. Some of his inheritance was sold for cash, while other pieces were lost, accidentally destroyed, or even given away to his students. No one knows exactly how many he disposed of, but Bach’s obituary referred to five seasonal cycles at St. Thomas Church, which would imply a total of four hundred cantatas composed during his tenure. Only about two hundred of them survive today. Meanwhile, other records indicate the existence of several masses, concertos, fugues, and other works that have never been found.
When I learned this bit of history, I was genuinely devastated. Hundreds of musical pieces from the greatest Baroque composer in all of Europe, gone forever. Just imagining what the seventh Brandenburg Concerto might have sounded like, and knowing we would never hear it, was deeply painful.
It’s a little hard to explain why I find Bach’s music to be so transcendent. The sense I get when I listen to his work is that he’s not telling me his story, but humanity’s story. He’s sharing the joys and sorrows of his life in a more universal sense, a language that doesn’t require me to understand the specifics of his situation. I can read a book from eighteenth-century Germany, and find some amount of empathy with the historical figures inside, but there will always be a forced translation of culture, society, and a thousand other details that I can never truly understand. Bach isn’t bogged down in those things—he’s cutting straight to the heart of what we already have in common. He can reach across three hundred years and make me, a man who manipulates electromagnetic circuits with my fingertips on a keyboard, feel just as profoundly as he made an impoverished farmer feel during a traditional rural celebration. He includes me in the story, just as I wanted to include my players in my games; we make the story together. Bach’s music is a perfect illustration of the idea that it’s not the artist that matters, but the connection between us.
I couldn’t bring back Bach. But what if I could harness artificial intelligence to generate more music like his, calculating harmonies and rhythms and contrapuntal phrases with the same ruleset that he would have followed? If he could create a puzzle canon with only one right answer, then so could a computer.
This was, admittedly, dangerous territory. People take it personally when you suggest computers can create art, let alone art that rivals our best. They see it as a reduction of humanity, rather than an elevation of technology. But Garry Kasparov didn’t cease to be a human the day the Deep Blue* computer beat him at chess, and the beauty of Bach’s work wouldn’t diminish even if I did manage to mimic his style.
Besides, I think time has proven that the “humans are special” folks don’t have much to worry about. We’ve made so much progress in the confluence of art and technology over the last twenty-five years, and yet are still so far from completion. Every time we solve a problem like chess, we find three more impossibly fuzzy and human-dependent problems like humor, love, or running on two feet without falling over. So, I don’t think we’re in any danger of making either art or humanity obsolete. In fact, I’d say that creating a computer that creates art counts as a form of artistic expression itself. It’s participation, not hubris. Regardless of whether my musical experiments were successful, or even passable, nothing could be more human than the act of trying in the first place.
Plus, it was about as orthogonal to Civ as I could get.
I started with the fugue, since it was one of the more rigid formats that Bach wrote in. Like a sonnet in poetry, there are rules to what a fugue is, no matter who’s composing. It gave me a good benchmark to see how close I was getting, first to a fugue at all, and then hopefully to a Bach fugue.
I enlisted my coworker Jeff Briggs for advice. He had been hired at MicroProse as a composer—the third member of our growing sound department—for a game called Sword of the Samurai. (“Like Pirates!, but with samurai.”) But he also had a background in board game design, and functioned as a kind of everyman on a number of projects. He did playtesting and documentation for several games, including F-15 Strike Eagle II, and was a project leader on various ports of Pirates! and F-19 Stealth Fighter. He wrote music for Railroad Tycoon and Covert Action, among others, and most recently he had worked with me on Civ, composing music and contributing to the massive Civilopedia reference tool that Bruce had started.
While I may have read Walter Piston’s textbook, Jeff could have written a version himself. He helped me break down the obvious rules, the secret rules, and the broken rules of Bach’s music. We talked about what made him unique and stylistically recognizable from other composers, and I resisted Jeff’s attempts to convince me that some of those other composers might, in their own way, be as talented as Bach himself. Maybe that was true, I thought, but I wasn’t interested in anyone else’s music.
Jeff also helped me file the patent for the game’s algorithm, which stretched to 12,000 words by the time the lawyers were done with it. The idea of a computer generating music was certainly not new—our patent referenced fifteen similar claims, as well as technical books from as far back as 1956, and was subsequently mentioned by 117 others, the most recent from Yamaha in 2016. But the way in which we did it was different enough that we felt it might be a nice thing to commemorate. We included a dense, three-page flowchart explaining the program’s logic tree, and outlined major acceptance rules, like “Leaps of more than a fifth are always followed by a step back” and “A step followed by a leap in the same direction, if the first note is a sixteenth note, is prohibited.” I also programmed statistical tendencies that would discourage things like dissonance, but not prohibit them entirely—in other words, rules on how and when to break the rules, just like Bach did.
I named my creation C.P.U. Bach, as a portmanteau of his most responsible son and the central processing unit of a computer. The melodies might not have been inspired by numerology or emotions, but it worked well enough to convince a layperson. Even a Cornell University music professor acknowledged that it was, at least on occasion, “uncannily plausible.” MicroProse agreed to publish it, though I’m not entirely sure why. Mostly, I think I was being given a free pass, since I had just made them a whole ton of money with Civilization. And hey, they hadn’t thought that game would be a big success either, so who knew? Maybe I was about to prove them all wrong again.
I wasn’t.
&nbs
p; The obscure subject matter and minimal interactivity definitely played a role, but those weren’t the only reasons C.P.U. Bach was a commercial failure. The other major pitfall was the console we chose for the game’s release, a new machine called the 3DO.
For the most part, the 3DO was simply ahead of its time. The early 90s were full of technological optimism, fueled by the upcoming turn of the millennium—everyone just knew that once the years had twos in front of them, we’d be living in a science fiction paradise. Things like virtual reality and internet connectivity were barely in their infancy, but advertisements and news profiles promised us that they would completely infiltrate our lives any day now. What had previously been thought of as merely inevitable suddenly seemed imminent, and everyone agreed that a comprehensive media center was at the top of the innovation list. Music, movies, games, telephone calls, and more would come from a single, universal box—along with a propeller function to blow everyone’s hair back, if you believed the ads.
Electronic Arts founder Trip Hawkins strongly supported this noun-not-adjective dream of “a multiplayer,” and when the EA board was hesitant to enter the hardware market, he stepped down to pursue it on his own. Supposedly, he chose the machine’s name as a reference to the new rhyming triumvirate of media: audio, video, and “3D-o.” It would serve all your needs, and replace all your devices, with better graphics than your PC, and better speakers than your stereo. The 3DO wouldn’t even play common floppy disks, only CD-ROMs, a move which prodded developers to either take advantage of its cinematic capabilities, or else waste 99 percent of their disk space.
Like C.P.U. Bach, a number of factors contributed to the 3DO’s ultimate failure in the market, including a high price tag, inconsistent manufacturing quality, and a lack of support from game developers. Without games, even the greatest console in the world can’t amount to anything. But all of this was only evident in hindsight, and as of 1993, there was not yet any writing on the wall. The only thing everyone agreed on was the size of the 3DO’s impact, which one stock analyst described to the New York Times as “a binary event.” It would either be the biggest hit, or the biggest failure, the industry had ever seen.
As usual, I tried to stay away from decisions based on money, and consider only what was best for the players. C.P.U. Bach was a music generator, and it didn’t make sense to release it on a platform with substandard audio output, which unfortunately included the vast majority of consumer PCs. High-quality audio cards existed, but they weren’t common, and I didn’t want people listening to our music in eight-bit mono and thinking it was the best we could do.
All of the evidence, not to mention the heavy marketing, pointed toward the 3DO as the best choice for an artistic, media-centric project like ours. Plus, it came with an algorithm that could generate colorful, abstract visuals in time with the beat, in case users got tired of watching our little animated Bach accurately playing the harpsichord. My friend Noah Falstein, who worked for 3DO at the time, has ruefully admitted to “convincing” me to release my game with them, as has Trip Hawkins, but I don’t remember getting a hard sell from anyone in particular. I went with the 3DO because it seemed like the best format for the game, and for all I know, the sales for C.P.U. Bach would have been the same on a different platform anyway. My only regret is that the game is essentially unplayable today, now that the physical console has become a lost relic.
But I still have a 3DO at home.
* Achievement Unlocked: Watson’s Pals—Discuss Deep Blue and Sherlock Holmes.
14
SEQUEL-ISH
Sid Meier’s Colonization (1994)
*
Sid Meier’s Civilization II (1996)
CIVILIZATION MAY HAVE LEFT ME burnt out for several years, but other designers in the building were just getting started. We’d made the world safe for strategy games, and there were plenty of ideas that we hadn’t had the time or wisdom to fit into ours. Some of the best belonged to a young designer named Brian Reynolds. He had been hired for a strange project called Rex Nebular and the Cosmic Gender Bender, which was apparently MicroProse’s answer to the Leisure Suit Larry franchise of adult-comedy adventure games. Though I wasn’t personally involved in its development, we still have a number of gaudy red baseball caps from this game floating around the office—unlike today’s cornucopia of flash drives, fidget toys, travel mugs, and reusable grocery bags, promotional merchandise back in the early nineties was almost always clothing—and we wear them for good luck when a project is going through its final testing phases. I don’t know how this marketing swag from a bygone era came to be seen as lucky, but it probably has something to do with the idea that if we released that game, we can release anything.
Fortunately for everyone, Brian’s heart was in the strategy genre, and now that he had his foot in the door, he was eager to prove it. Without being asked, he created a working prototype of a game he called Colonization, which he pitched to the executive team as a narrower, deeper version of Civ. Set during the European discovery of the Americas in 1492, the game would focus less on expansion and more on resource gathering within the player’s society, testing its robustness primarily through the economic challenges of the era. Any colonist could grow tobacco, for example, but a Master Tobacco Planter would do it twice as fast, especially if your colony bordered on the proper grassland. Meanwhile, a separate colonist might be trained to convert that tobacco into exportable cigars, and an Expert Farmer could grow enough to feed all three of them. Once you established a sufficient population and achieved dominance over the other colonizers, the game ended by staging an alternate version of the American Revolution, allowing you to rebel against the king of whatever nationality you had originally chosen and secure your independence.
MicroProse merchandise ad.
© 1987 MICROPROSE, WWW.MICROPROSE.COM.
The corporate suits had seen the error of their ways, at least to a certain degree, and were willing to support a strategy game that might continue the recent sales numbers of Civilization. I think it was expected that I would help Brian on his fledgling project, maybe even taking over if it turned out he’d bitten off more than he could chew. But he didn’t need it, which was fortunate, because he was flying without a net whether anyone knew it or not. I was not about to end up coding another strategy game so soon after Civ. I gave guidance early on, mostly of the “here’s how you figure that problem out for yourself” variety, and then checked back in at the end to help him tweak a few final details, but the eighteen months in between I spent as a committed disciple of Bach.
Colonization and C.P.U. Bach actually released at the same time, but the last few months of my game were taken over by a programmer named Kerry Wilkinson, who did the work of converting the finished PC code onto the 3DO. Both games were on display at CES that year, but while Brian’s was in its final, almost-bug-free stages, the conversion of C.P.U. Bach was not a gradual process. Either it was done, or it wasn’t—and when the day came, it wasn’t. So instead, we put a decoy 3DO on display, then hid a PC inside the cabinet underneath to actually run the program. We didn’t lie about it if asked, but we hoped we wouldn’t get asked too many times.
Even after I started offering more concrete direction at the end of Colonization, I worked very hard not to alter the spirit of the game Brian had created. I was willing, for example, to suggest cutting the radius of the cities in half—something we pulled off at the last moment, just as we had with the world map in Civ—because it further highlighted the job-specialization mechanics that Brian had developed. But I didn’t argue against ending the game with the American Revolution, even though it was a grandiose, win-or-lose proposition with the potential to invalidate hours of successful gameplay. Generally speaking, I would never risk alienating the player to that degree. It was historically accurate, however, and Brian saw it as a satisfying boss battle rather than a last-minute bait and switch, so I deferred to him. Good games don’t get made by committee.
The question
of how much influence I’d really had on the game brought us to a major crossroads, both for MicroProse and the future of my career. I’m not sure if they’d planned it from the beginning, or held off until I crossed some imaginary threshold of hours spent, but at some point marketing began floating the title Sid Meier’s Colonization.
Truthfully, my name had already gone on one game I didn’t do much for, Railroad Tycoon Deluxe. But that had been a mostly cosmetic upgrade to my original code, and there were no new designers whose contributions were being diminished. Colonization, on the other hand, was not “Civilization Deluxe.” It was a unique world that had drawn only loose inspiration from mine, and Brian had written every line of code—I’d made sure of it. Yes, I made suggestions along the way, but it had been up to Brian whether to accept them. Colonization was not Sid Meier’s game.
From a marketing standpoint, though, none of that mattered. Over the course of five games and one remake, my name had somehow become a brand. My entire philosophy of gaming was that the player should be the star and the designer should be invisible, yet I was the guy who kept ending up on the box. I should clarify that no one was trying to maliciously exploit me—the marketing team’s position was purely utilitarian. And I couldn’t deny that selling more copies would mean a stronger company, which I did still care about even if it wasn’t technically my company anymore.
Sid Meier's Memoir! Page 15