2016-03-27

Born from speed-up kits
Many of Pac-Man's historical tidbits came to light five years ago when series creator Toru Iwatani spoke at the original game's 2011 post-mortem panel. (I had the honor of asking Iwatini about the game's name-change from the original Puck-Man, to which he laughed and said, "Don't you already have your answer?") But Ms. Pac-Man's genesis story is quite possibly more interesting—most notably because it wasn't created by Namco's Japanese staff.



Sam Machkovech
How arcade games' "speed-up kits" were originally advertised.



Sam Machkovech
How arcade games' "speed-up kits" were originally advertised.


You didn't need any soldering know-how to connect Asteroids' most popular speed-up kit.

Sam Machkovech
Steve Golson, one of the creators of Crazy Otto, which eventually became Ms. Pac-Man.

Before Ms. Pac-Man, Golson and his friends worked on a speed-up kit for Missile Command.

Assets for GCC's mod, Super Missile Attack, had to be mapped out by hand so that the sprites could be coded into the mod.

Atari didn't take too kindly to Super Missile Attack.

As arcade-gaming historians know, Ms. Pac-Man began life as a "speed-up kit" developed by a crew of MIT drop-outs. One of those drop-outs, Steve Golson, was at this year's GDC to tell us the story of its creation. He and his MIT buddies began their game-industry careers by acquiring and operating a few coin-op arcade machines in their dorm; that enterprise spread to three dorms before they left the school. This didn't just give them easy access to hot games between classes. It exposed them to a brutal reality of the arcade market. The problem with arcade machines, Golson said, was that their "expensive cost of entry" wasn't sustained by long tails of profit. "Players got really good at the games and figured out how to beat them. They'd play for longer and longer on a single quarter." Golson and his friends soon discovered the world of speed-up kits, which were marketed by non-official companies as add-on boards to be connected to otherwise perfectly functional games. (The term "speed-up kit" came from one of the earliest such hacks, which raised Asteroids' processor-interruption rate to very simply speed that game up and make it more challenging.)
"Missile Command was really popular at MIT," Golson added. "Nuclear war! Explosions!" But unlike other hits of the era, that game didn't yet have a speed-up kit to help with its lagging-profit issues. "Suddenly, we realize: we go to MIT! I bet we could write a kit." Thus, General Computer Corp was born in spring 1981—a name chosen with hopes that it would sound like a huge company (like General Electric or General Motors). "We started during spring break. It was much more fun writing games software than going to the beach."
Golson talked about the simple tweaks needed to tweak existing arcade games—particularly the data tables just sitting on an arcade board's ROM chips—but also spelled out the grueling process necessary to disassemble the game's full code and figure out exactly how to make a working speed-up kit. Armed with a GenRad 6502 microprocessor emulator (which they bought for $25,000 thanks to a parent's loan) and a TRS-80, Golson and his team stayed up all hours copying disassembled code from Missile Command by scratch. "You had to sit next to the TRS-80 and type it up live," Golson remembered.
Unlike other speed-up kit creators, GCC went to great lengths to ensure that its kits couldn't be copied—even scratching PAL markings off of its chips—and soon launched its Super Missile Attack kit. That pricey GenRad investment was quickly paid off and then some, thanks to over a thousand units selling within a matter of weeks for $250 a pop. However, with success came the watchful gaze of Atari.
"Atari back then... was the high-tech company," Golson said. "If you took Google, Apple, Microsoft, Sony, and half the companies on GDC floor and put them in one, that was Atari." That might explain why Atari's lawyers slammed GCC with a $15 million lawsuit (which is already a crazy number before adjusting for inflation since 1981). When GCC chose "not to back down" and instead responded by sending depositions to Ray Kassar and various Atari engineers, the big company panicked. "Legal protection for games at this point was very unsettled," Golson said. "Copyright for games, for audio-visual presentation? Not clear at all. Atari didn’t want to set a precedent," especially if the case didn't go Atari's way.
"Atari was getting annoyed," Golson added. "Naturally, they hired us."

You Otto leave some Easter eggs...
GCC was paid a combined $50,000 a month to become Atari staffers on a two-year contract. The lawsuit was dropped on a few conditions, including that GCC cease work on Super Missile Attack and—more importantly to Golson—not produce any more speed-up kits "without permission from the manufacturer." (As Golson joked, Atari let that one slide: "Who's gonna let these clowns" make a kit?)
That language was crucial, Golson noted, because before Atari dropped its lawsuit, GCC had already been hard at work on another speed-up kit.

Sam Machkovech
GCC added more unpredictability to Pac-Man by adding more mazes. This image shows the original Crazy Otto sprites.

Sam Machkovech
GCC added more unpredictability to Pac-Man by adding more mazes. This image shows the original Crazy Otto sprites.

Getting character data out of the original ROM was easier this time...

...and the devs found that Lite Brites made sprite artistry a lot easier.

Sam Machkovech
Steve Golson holds up a copy of the Pac-Man source code.

The one original song made for Crazy Otto.

From Otto to Ms. Pac-Man—and some weird-looking red hair in between. Notice that Otto had way more frames of animation so he could be seen from behind while standing up.

OG!

GCC was proud of its lawsuit and contract warpath.

"There were obvious problems with Pac-Man," Golson said about his company's choice of follow-up project. The biggest was predictability, as proven by popular books that had analyzed and solved the game's obvious patterns. "You could play Pac-Man forever, so, okay. We'll kill the predictability." GCC bought itself a new decompiling machine, the Tektronix 8550—complete with a line printer, which saved the team the fatigue issues of copying code by hand—and combined that with repair and schematic manuals to get to work. Among other things, the 8550 allowed the team to put the game into slow motion so they could check out cool tidbits in the game, particularly Pac-Man's "skidding" animation when turning corners and the ghosts' eyes leading the way a few frames before they would turn in a direction.
An algorithm-driven counter enabled "truly random" ghost movement, Golson said. That unpredictability was bolstered by four brand-new maze designs, which could be added via external ROM chips since Pac-Man, unlike Missile Command, had no constraint on memory space. Some game tweaks were easy to institute, like new fruit designs and fruit movement in the maze. Other tweaks, particularly vertical tunnels at the tops and bottoms of mazes, confused the ghosts too much, so GCC scrapped them.
Speaking of fruit: GCC took the lawsuit's lessons to heart and removed elements that might prove legally dicey, such as the Galaxian ship as a bonus "fruit" icon (which was replaced with a pretzel as a tribute to team member Kevin Curran's love of salty snacks). GCC rebranded almost everything else: their version of Pac-Man had a new name (Crazy Otto), new character designs, and a requirement that "existing Midway ROMs provide the new GCC code as an overlay."

With GCC's technical and legal ducks—or ghosts—in a row, the 21-year-old Curran called Midway president Dave Marofske on October 9, 1981, the day after the Atari deal had been inked. This was before anybody knew what the deal actually meant. "Kevin says, 'You may have heard that we trounced and pounded Atari in court and they dropped their lawsuit. We're about to do the same thing with you. If you have any disagreements, let’s settle them now. We don’t wanna have to take you to court, too.'"
The cold-call bluff worked.
"In one month, we’ve struck deals with the top two game manufacturers in the country," Golson said with a laugh. "Dude, it's looking better and better—that dropping out of school was the right thing to do."
Midway bought the rights to Crazy Otto and contracted GCC to rework its kit as a full-on Pac-Man sequel, which was originally going to be called Super Pac-Man. A Midway rep then asked the team to turn Pac-Man into a woman, but the suggested name, Miss Pac-Man, was shot down thanks to a cinema scene showing a Pac baby. "Out of wedlock video game birth? No!" After that, "Pac-Woman" became Ms. Pac-Man.
Golson reminded the crowd "not to believe everything you read on the Internet" before dispelling an old rumor. Turns out, Namco was involved in, and approved, Ms. Pac-Man's production from the moment Midway found out about it. In particular, Namco president Masaya Nakamura asked the developers to change the character from a red-headed creature to a simple, bow-sporting, beauty marked Pac lady.
GCC's 10k Otto ROM went into production as an auxiliary add-on chip for the original 16k game—whose anti-piracy measures Golson was particularly proud of. "It wasn't cloned or pirated for six months," he said, "which helped Namco ramp up production. That was my code, the anti-copying hardware." GCC left two Easter eggs in the ROM: the company's acronym as a logo, and the phrase "HELLO NAKAMURA" tucked into the source code—both of which were eventually used by GCC staffers to prove in court that its code had been lifted for the game's 20th anniversary arcade edition.
Golson concluded his talk by describing GCC's other output, both within Atari and in the decades after its collapse. He then suggested that he and his colleagues come back to GDC some day for another post-mortem panel—on how they created the Atari 7800 in 1984. "That would be a cool post-mortem talk," Golson said. We couldn't agree more, GDC.

Diablo as a... turn-based game? You kidding me?
GDC 2016's final panel saw Diablo lead programmer and concept creator David Brevik wax nostalgic on the game ahead of its 20th anniversary—which he swears is in 2016. "Some people say it came out in 1997, but if you were on the West coast, you could buy it on December 31," he insisted.
Diablo began life as a "party-based, Wizardry kind of game" concept in Brevik's junior year of high school, when he lived in Danville, CA, at the base of Mount Diablo. "I didn't speak a lick of Spanish when I moved there," Brevik said. "Just thought, yeah, that name rocks! I'm gonna use that someday."

After a brief tenure at Iguana Studios (NBA Jam), Brevik wrote Diablo's first design doc under the moniker "Project Condor" during his lunch breaks at another studio. While he drew inspiration from Nethack, Rogue, and other roguelike RPGs, Brevik's first plan had many different elements from the final game, including expansion pack discs that only contained new items to add to the game, a la Magic: The Gathering booster packs, along with turn-based action, permadeath, and a very brief dalliance with Primal Rage-styled claymation characters. "We looked into [claymation] and said, 'oh my god, this is a disaster. There's no way we can make and photograph these models. We tried stop-motion for five seconds."
Brevik met the devs who would eventually form Blizzard when his company (then known as Condor) and their company (then known as Silicon & Synapse) met at booths for their games during CES 1994. Both companies were under contract to make Justice League-themed brawling games, one for the Sega Genesis and the other for Super Nintendo. Neither team even knew the other existed. The S&S team gave Brevik an "under the radar" scoop: they were bailing on console game development to make their own fantasy-themed PC games. "That's my dream, too!" Brevik exclaimed. Less than a year later, they crossed paths once more and signed a deal so that Condor would make Diablo for Blizzard.

Unfortunately, Condor didn't think that January 1995 contract through. "We were so excited about making Diablo that we signed a contract without realizing we’d decided to do Diablo for $300,000." Brevik paused for emphasis. "Fifteen people in the studio! What? Twenty grand a piece for the next year to get this done? Plus with office space and everything, that’s down to $12,000 a month? Wow, that was dumb. But we’re making Diablo!"
Brevik scrambled to find more money. He landed Condor an "almost $1 million" side contract gig for a 3DO football game prototype ("NFLPA Superstars, six-on-six backyard football") running on the (never-released) M2 console. By the end of 1995, 3DO expressed interest in acquiring Condor, which kicked off a "bidding war" between 3DO and Blizzard. Brevik ended the war by choosing to stick with his fantasy-loving peers. "3DO offered us twice as much money. We turned them down. Really, because we felt that Blizzard really got us and got the game. We were so close in company culture and beliefs."

(Brevik mentioned one other lousy business decision: when he turned down his friend Sabeer Bhatia's 1996 request for free office space in exchange for 10 percent of his eventual startup project. Brevik was slammed with Diablo work and kicked Bhatia out. A year later, Bhatia's creation, Hotmail, sold to Microsoft for $400 million. "It's a stupid idea," Brevik had originally said about Hotmail. "I already have e-mail on the Internet!")

NHL '95 with a loot tree

Among the other design decisions illuminated by the panel: Brevik and his team (who were eventually rechristened Blizzard North) copied XCOM's "actual tile square basis." Brevik admitted that Diablo's design "was basically built upon a screengrab from XCOM." Also, Brevik was the dev team's last holdout on two of Diablo's most iconic elements: real-time control and online multiplayer.

When he finally caved on removing Diablo's original turn-based system, Brevik demanded a milestone payment from Blizzard, since altering such a core part of the game would take a long time. He got the payment. A few hours later, he implemented fully functional real-time combat. "Everything [in the game already] took different amounts of 'turns,' so I just made the turns happen 20 times a second, and it all worked magically.

"I remember the moment that this happened," Brevik continued. "It was one of the most vivid moments in my career. I remember taking the mouse—I clicked on the mouse, and the warrior walked over and smacked the skeleton down. I was like, 'oh my god! That was awesome!' That’s when the ARPG was born, at that moment, and I was lucky to be there."

On the multiplayer side, Brevik's Blizzard bosses were astonished to learn that roughly seven months before shipping, the game didn't have a single line of multiplayer code written yet. "I didn’t know anything about multiplayer or networking at the time," Brevik admitted. He gave Arena.net founder Mike O'Brien credit for righting Diablo's multiplayer and Battle.net ships from the get-go. (He also declared his mea culpa for one major result of the game's peer-to-peer structure: "[Once Diablo] came out, instantly we were like, 'oh my god, they can upload the cheats and everyone can cheat. We didn’t even think about that!")

Brevik broke down some other Diablo minutiae at length, including the game's lighting and shading systems ("we had up to 16 light levels... that hadn't been seen at the time") and its NHL '95-inspired startup process ("in RPGs, there was always 25 minutes of character creation—what color are your eyes, where are you from—but in NHL '95, just, clicky-click, I am skating"). The game's funniest development story came from a $100 bounty offered to the first staffer to kill Diablo in the near-final version: "We had a skill in the game, 'Blood Exchange,' which swapped your hit points with a monster's... Yeah, we didn’t think that through, did we. We promptly yanked the skill."

During the Q&A session, Brevik admitted that Battle.net originally ran on a single PC: "We weren't handling bandwidth; we just handled connections." He also minced no words criticizing Blizzard's decision to kill a major character in Diablo III: "I think it sucks. I love that character, too... but they're gonna do what they're gonna do." He was even harsher in "disavowing" Diablo's Hellfire expansion pack, developed by one of Sierra's in-house studios: "They made a teddy-bear quest." (In a much warmer scene, a fan approached the microphone, apologized for pirating the game in the '90s, and walked up to Brevik to hand him $60.)

And Brevik admitted that Diablo's final cinematic scene, in which a gem is taken out of Diablo's head and wedged into the player's own (no spoiler warning, you've had 20 years to see this), caught the development team by surprise—since it was created by Blizzard's separate cinematic team instead of Blizzard North. "We crowded around to look at this thing, and then... he jams the gem in his forehead! We were all, what the hell did we just watch? What was that? That doesn’t make any sense at all! Is that cool?" After some internal hemming and hawing, the team made its peace with that ending. Brevik admits he likes it even more in hindsight "especially [because of] the ability to play [the game] over and over again."

Due to the panel's focused nature, Brevik didn't talk about development on Diablo II, nor did he wade into his work with Marvel Heroes and Gazillion Entertainment. He left that Diablo-like franchise behind this January to "go indie" and form a new company called Graybeard Games.

Show more