The current world record for the Atari 2600 racing game Dragster is 5.51 seconds. Thirty-five years after it was set, nobody has beaten or even tied it on official leaderboards. One speedrunner who has examined the game’s inner workings believes that world record to be impossible, but the game’s creator thinks it’s legitimate. For now, the Dragster record is one of the great unsolved mysteries of the golden age of video games.

Released in 1980 by Activision, Dragster was the pioneering publisher’s first game (and therefore the first third-party game ever released for any console). Created by company co-founder and Pitfall designer David Crane, Dragster is a surprisingly robust simulation of drag car racing in which players must engage their car’s clutch and shift into multiple gears as their speed increases. The fastest scores require near-perfect timing of the gear changes.

Advertisement

The scorekeeping organization Twin Galaxies recognizes a competitive gamer named Todd Rogers as recording a time of 5.51 seconds in Dragster. The Guinness Book of World Records officially recognizes Rogers’ achievement as the longest-standing game record—Rogers is said to have set the score in 1982, although Twin Galaxies did not verify the achievement until, Rogers says, around the year 2000. Twin Galaxies’ leaderboards also recognize five players with scores of 5.61, all tied for second place. But none of them have been able to duplicate Rogers’ feat.

Todd Rogers is one of the most well-known expert gamers of the 1980s.

Todd Rogers made a name for himself in the 1980s with many high scores and gaming achievements, including a 65 million-point score in Centipede. He appeared in the documentary King of Kong, which chronicled the race for the high score in Donkey Kong. He later went on to work at Twin Galaxies from 1999 to 2012.

Advertisement

Besides Dragster, Rogers also held high scores in games like Skiiing, Stampede, and Grand Prix. As one of the celebrated expert gamers of the time, Rogers was often featured in magazines, or flown out to gaming events like the Consumer Electronics Show. “He impressed us at Activision to the point that we paid his expenses to attend CES to demonstrate games, and he came up to our sales suite after show hours and demonstrated his play,” said David Crane in an email.

Rogers says that he replicated the 5.51 score multiple times in 1982: once at the Chicago Consumer Electronics Show and another time at The Electronic Thing show in Detroit. “That was more than enough for the developers,” Rogers told Kotaku. “You can’t alter history, no matter who speculates what is possible and what is not.”

Advertisement

“Activision validated Todd’s Dragster score using the accepted methods of the day,” Crane said. “The time to question any of those records has passed.”

As evidenced by films like King of Kong, there is still a thriving community of players attempting to set records on decades-old games. This is what makes it perplexing that no one has ever been able to equal Rogers’ score in the 35 years since. The leading skeptic is Eric “Omnigamer” Koziel, a speedrunner and creator of tool-assisted speedruns (in which a specialized computer program plays a game with superhuman timing).

Advertisement

As one of the first major speedruns, the Dragster world record caught Koziel’s eye as historically significant. But as Koziel began to investigate the inner workings of the Atari game, he started to question whether Rogers’ run was mathematically possible. His theoretically perfect tool-assisted run could only manage a 5.57.

Dragster’s simulation of drag car racing is surprisingly detailed for the time.

“The best [tool-assisted runners] could do wasn’t even close to what Rogers had claimed,” Koziel told Kotaku. “The more I dove into it, the harder and harder it became to identify ways to improve further.”

Koziel’s doubt regarding this record rests with how the game functions. The best recorded time with video evidence is a 5.61, which five people on Twin Galaxies’ chart currently hold. And while Koziel simulated a method that could realistically achieve a time of 5.57, there is little room for error: a mere 8 frames of leeway. At a speed of thirty frames per second, that leaves slightly over a quarter of a second for mistakes.

Supporting Rogers’ claim that 5.51 is achievable is the fact that, although his is the only record that was recognized by Twin Galaxies, Activision said that two other players had achieved scores of 5.51 in the Spring 1983 edition of its newsletter, which was called Activisions.

An excerpt from Activisions volume 6, Spring 1983 shows two other players tied with Rogers.

Activision encouraged players to photograph their scores send them to the company. If those scores were below six seconds, Activision would reward the players with a cloth Dragster patch and certificate.

Advertisement

Advertisement

David Crane said that Activision was stringent about its own high-score record-keeping. “Activision used photographic evidence, requiring a Polaroid photo of the TV screen,” he said. “This was pre-internet, and pre-Photoshop. It would have been very difficult to falsify a Polaroid photo.”

Moreover, says Crane, if Rogers were simply making up a high score, it would be highly unlikely that he would invent the number 5.51 out of whole cloth. While the game’s timer may appear to count each hundredth of a second, it actually advances 0.03 seconds at a time. Activision’s internal tests gave them a theoretical best time of 5.54, and if Rogers was able to best that by a single tick of the clock, it would be 5.51.

“If someone was trying to cheat in some way, he wouldn’t have known to claim that exact time,” Crane said. “I think you can rest assured that Todd’s time was, in fact, the time of his run as played on the actual game system.”

Advertisement

And, of course, the fact that three players individually claimed that time would back up Rogers’ story. But Rogers says that the other two scores were “disqualified.” There’s no printed evidence to back this claim, since no retraction was published in the final Activision newsletter from Fall 1983. Rogers says that his score alone was allowed to stand because he was able to convincingly explain to Activision how he did it, and says that the other two high scorers could not.

In a 2002 forum post signed “Todd Mr. Activision Rogers,” he explained his method: “The only way I beat their time was to engage the clutch at the count down and rev up my engine in the red just before the count of zero. Once the timer reaches zero, I pop my clutch and I’m already in second gear.”

Advertisement

Advertisement

Koziel, the speedrunner, believes that even this would not be enough to secure Rogers’ time, although he said it would theoretically be possible if Rogers had somehow been able to start the race one frame early.

“Technology in the 80's was open to a few problematic issues,” Koziel said. “It’s possible that his Atari happened to have a memory block stuck at some value that enabled certain things to happen that could have caused a 5.51.”

“Could the hardware affect the game scores? Possibly,” Crane said. “Does that explain Todd Rogers’ Dragster score of 5.51? No.”

Advertisement

Rogers has never publicly shown a photograph of his Dragster high score, and it was not witnessed in person by any Twin Galaxies official. Rogers says he submitted a “bevy of material” to Twin Galaxies, including original copies of the Activision newsletter, which its referees used to confirm his run to their satisfaction and add it to the leaderboards around the year 2000. [Update 4:30: The wording in this paragraph was updated after publication to better indicate the order of events. While Rogers said he did send a photograph of the score to Activision in 1982, no such photograph is available today.]

Some of Rogers’ other Twin Galaxies scores have been contested in the past, including an alleged 15 million-point score in 2600 Donkey Kong. (The current record is 1,472,100.) Many of Twin Galaxies’ officially recognized records credited to Rogers are hundreds of thousands or even millions of points higher than the second-place score. Some of them are perfect round numbers that are several orders of magnitude greater than any score with video evidence. Rogers is officially credited with a score of 65,000,000 points on Centipede for Atari 5200; the second-place score is 58,078.

Above: Eric Koziel’s tool-assisted speedrun, which still could not match Rogers’ time.

Advertisement

Advertisement

In a new tool-assisted run submitted in March, Koziel works frame by frame to ensure what he believes to be the best possible time. The run finishes with a 5.57, which is 0.03 seconds slower than the rumored Activision simulation (5.54) and 0.06 seconds slower than Rogers’ world record (5.51). Previous tool-assisted runs by other speedrunners were also unable to beat this time. So far, no one is faster than Todd Rogers, not even computers.

Rogers’ contention is that the videos are only simulating one way of playing a quite complex game. “There’s like nine ways to shift in Dragster—and I don’t share that with too many people—but [Koziel is] going on one specific pattern where you stay in first gear and second gear quite a bit of time,” he said.

“I could sit in front of a TV right now and play for an hour straight and get 650 different types of play and it would never be the same,” Rogers said. “If he’s basing his spreadsheets and his shifting on one particular pattern, then that’s pretty ignorant and closed-minded, because you’re not factoring in the human element of how the game would respond.”

Advertisement

In the opinion of Dragster’s creator, Rogers’ record needs no additional proof. “Those records were set at the time, by a player at the time, and given validation by the authority at the time. Thus those records should stand,” David Crane said. If there is a discrepancy between Activision’s record and a computer simulation, Crane said we should be questioning the simulation, not the record.

“We have credible empirical evidence that 5.51 was, in fact, possible,” he said. “The question should be, ‘Why does the mathematical analysis disagree with the empirical evidence?’”