Please create an account to participate in the Slashdot moderation system

 



Forgot your password?
typodupeerror
×
Games

Longest-standing Video Game Record Declared 'Impossible,' Thrown Out After 35 Years (polygon.com) 234

Twin Galaxies, the video game record keeper and official source for Guinness World Records, has declared one of the oldest gaming world records invalid after 35 years. From a report: Player Todd Rogers has been stripped of his world record for finishing the simple Atari 2600 racing game Dragster, after months of debate over his completion time. "Based on the complete body of evidence presented in this official dispute thread, Twin Galaxies administrative staff has unanimously decided to remove all of Todd Rogers' scores as well as ban him from participating in our competitive leaderboards," reads a post on the Twin Galaxies forum from the organization's staff. That's a major blow to a prolific record holder, whose career stretches back to the earliest days of console gaming. Rogers courted controversy with his oldest record, however -- and it directly caused his ban. In 1982, Rogers submitted to Activision's official fan newsletter a time of 5.51 seconds, which the company recognized in print, awarding Rogers a patch Twin Galaxies later added Rogers to its own leaderboards in 2001, and Guinness World Records awarded the player with the honor of holding the world's longest-standing gaming record in April 2017.
This discussion has been archived. No new comments can be posted.

Longest-standing Video Game Record Declared 'Impossible,' Thrown Out After 35 Years

Comments Filter:
  • by Anonymous Coward on Monday January 29, 2018 @04:33PM (#56028903)

    Thanks to this, my record of 5.52 seconds is now on top.

    That's Arthur Sullivan Smith. Just the initials are fine.

    • Re:Yes, finally (Score:4, Interesting)

      by The MAZZTer ( 911996 ) <(megazzt) (at) (gmail.com)> on Monday January 29, 2018 @04:35PM (#56028921) Homepage
      Not sure if you care, but the best possible score is 5.57 seconds. That's how he got found out.
      • How do you figure that?
        And how did he file for the record?
        I mean if you write it by hand, my hand written 5.51 would read 5.57 for an american.

        • Re:Yes, finally (Score:5, Informative)

          by The MAZZTer ( 911996 ) <(megazzt) (at) (gmail.com)> on Monday January 29, 2018 @04:43PM (#56028979) Homepage
          Well, all games run on a fixed set of rules. Think of Monopoly etc. Video games are no exception. And the rules put in place by Dragster do not allow getting any faster than 5.57. One could argue maybe the analysis was incomplete or flawed but there's a whole lot more sketchy stuff about Rogers that makes it probable there's been no mistake (imo). I'll post some videos in a top level comment that go into detail about this stuff.
        • How do you figure that?

          They built a rig that can play the game optimally and that's the best it can do.
          https://www.youtube.com/watch?... [youtube.com]
          https://www.youtube.com/watch?... [youtube.com]

          • How do they know there's not some input edge case that allows for a better time? Seems like an audit of the machine code and the hardware would be the only way to really know. I didn't spend time watching the video.
      • Could his Computer be a hair faster, or the timer a bit slower?
        I mean it is a 2600, not quite a Real Time system. If the CPU was clocked up (even by accident) 1/20th of a second, difference is possible.

        • Re:Yes, finally (Score:5, Informative)

          by Anonymous Coward on Monday January 29, 2018 @04:49PM (#56029037)

          If the 2600 was clocked up the video signal would also be clocked up and the timer would also be clocked up.

          The 2600 does not have a real-time clock. The game only tracks time according to how many video frames were generated.

          • by mysidia ( 191772 )

            The game only tracks time according to how many video frames were generated.

            Is it possible that some frames were lost, or failed to be generated?

            0.06 seconds is a pretty tight margin.... it's not like the fastest possible was 5 seconds, and he was claiming 1 second with no explanation.

            Do the rules preclude using an external timer? Could also be a transcription error or other honest mistake. Was there no proof of the score included with the submission?

            • Re:Yes, finally (Score:5, Informative)

              by JMJimmy ( 2036122 ) on Monday January 29, 2018 @05:15PM (#56029217)

              The guy was a prolific cheater who use a friend (who's now in jail for fraud) to verify his scores or submitted them himself (he worked for Twin Galaxies). If you watch the video in the article there's a lengthy and thorough pile of evidence.

            • Possible there may had some problem with the Memory (something not-unheard of) where the memory segment that is taking the counter for the clock wasn't getting updated all the time.
              print c
              0
              let c = c + 1
              print c
              0
              let c = c + 1
              print c
              1

              Or when it counted the frames it had a floating point error.

              • The Atari 2600 had no floating point unit.

                Memory errors are highly unlikely - since generally these things rely on the hardware timer unit (which is synced to the system clock), without which nothing else works. As others have stated, rogers had a huge list of dodgy "records" that have either been proven to be impossible, or are strongly suspected of being impossible. He's even got records like scoring 6000 on a game that counts its score in an 8 bit register.

                There's basically no evidence that any of his

              • by gl4ss ( 559668 )

                sure if he didn't bullshit on a number of other records as well and said he got it multiple times and had people to testify.. ..but nobody recalls actually seeing it.

            • Re: (Score:2, Informative)

              by Anonymous Coward

              The STELLA chip in the 2600 only had enough RAM to hole ONE SCAN LINE of video.
              If your program didn't keep up, very bad things happened.

            • by cb88 ( 1410145 )
              The Atari 2600 doesn't drop frames as it essentially draws the frame directly onto the screen as the beam scans across the screen... so the only way to drop a frame is to blink.

              It also only has 128 Bytes of ram... yes that's right an eighth of a Byte. With a special expansion card you could add something like 6k and many cards, and banked roms so it could go above it's normal 4k of rom limit.
              • It also only has 128 Bytes of ram... yes that's right an eighth of a Byte.

                Ummm... Your math seems a bit dodgy there...

          • Re:Yes, finally (Score:5, Interesting)

            by lord_mike ( 567148 ) on Monday January 29, 2018 @05:26PM (#56029301)

            Perhaps he was playing the PAL version on an NTSC TV. It is possible to get a pal signal on an NTSC TV. It is black and white and you have to really mess with the vertical hold, but it works, especially if you have an old black and white TV. That's how I manage to get a PAL ZXSpectrum running here.

            It's unlikely, of course, but certainly not impossible. What's more likely is he may have turned the power switch on and off a few times quickly (called "frying" the cartridge), causing s blip in the ram which happened to give him a nice score that time around. Noticing that it was a great score, he took a picture and sent it in. I will have to find my old dragster cartridge and see if it's possible to "fry" it into a good score.

        • by mykepredko ( 40154 ) on Monday January 29, 2018 @04:58PM (#56029101) Homepage

          Interesting thought - I just took a quick look at the schematics of the CX2600 & CX2600A gaming systems at: http://www.atariage.com/2600/a... [atariage.com]

          and saw that there is only one main system clock which is roughly 3.58MHz - that means that this clock is not only used for the processor but for the video signal's NTSC colour burst (3.579545MHz).. I can't find a reference to the exact colour burst frequency tolerance (I thought it was around 20ppm or around 70hz) that is required for a proper TV signal output.

          Having a colour burst outside of the tolerance would mean, at a minimum, messed up colours and maybe the inability for a TV set to be able to display an output at all. No way could a variation of 5% (1/20 of a second) be tolerated by a TV Set.

          I guess all my NTSC knowledge/Skills/Experience are now worthless - except for trivia in cases like this.

          • by Anonymous Coward

            Not only that but speeding up the CPU would speed up the game timer so you're back to the same number.

            The game counts the number of frames generated and that is translated into the timer display ( 0.016 sec per frame).
            The game also runs in lock-step with the display: One game tick per frame. The game does not advances until a frame is generated and it advances by a fixed value.

            No matter how fast or slow you clock the system the game will display the same number.

            The system has no independent wall clock.

          • by Blymie ( 231220 )

            Well... we're talking late 1970s here. TVs surely could have lots of variance. Most even had vhold and lots of other such knob adjustments, and I think the set I played the 2600 on first, even had tubes!

            I agree it doesn't seem likely -- but, I don't think it's even remotely impossible for a set to get a sync on an off signal here...

          • by mykepredko ( 40154 ) on Monday January 29, 2018 @05:43PM (#56029441) Homepage

            Interesting seeing people's incorrect perceptions on 1970s/1980s TV technology.

            Sorry to disappoint you, but there were very strong standards for signal timing precision - a bit of Googling found: https://antiqueradio.org/art/N... [antiqueradio.org]

            Colour Burst frequency tolerance is +/-0.0003% which works out to roughly 10hz (I guess I mis-remembered or was thinking in terms of practical values).

            It wasn't all capacitors back then - lots of silicon, although they were fairly discrete functions at the time. You can get an idea of what a Sony Trinitron TV had inside it here: https://www.manualslib.com/pro... [manualslib.com]

          • I've been playing with generating NTSC signals from 8 bit microcontrollers lately, and when you mess up the color burst (you'd have to switch to PAL to spell it colour, sorry) it just degrades to black and white but still works well. There is a big area of failing to B&W in between the cases of messed up colors and total failure.

            • by dryeo ( 100693 )

              I'm sorry, but we have colour NTSC TV's here in Canada (some remote areas still have analog repeaters) and they used to be quite common.

        • by tlhIngan ( 30335 )

          Could his Computer be a hair faster, or the timer a bit slower?
          I mean it is a 2600, not quite a Real Time system. If the CPU was clocked up (even by accident) 1/20th of a second, difference is possible.

          No, it was physically impossible. To determine if it was real, they went so far as to disassemble the cartridge ROM and calculate what are the possible numbers that are valid for the time. That was how it was determined it was not possible - there was no way to score lower than 5.57 using the code.

          And no, PAL

        • Comment removed based on user account deletion
      • Not sure if you care, but the best possible score is 5.57 seconds. That's how he got found out.

        So his crappy early 1980s dot matrix printer and/or used up ink ribbon lost a few dots and the 7 looked like a 1? ;-)

      • Engadget is reporting his time was 5.51 MINUTES. I sure hope that's a speak-o.
      • Activision said it was 5.54 back in the day, and initially rejected the 5.51 claim. They later accepted the 5.51 claim.
        5.57 is based on some other competitor's analysis of the code.

        • by lgw ( 121541 )

          Not just some other competitor, but several people in the speed running scene reviewing that work. The code is simple enough to understand fully (the ROMs were quite small, and this game was simple even by 2600 standards), and everything that can happen frame-by-frame has been reduced to a spreadsheet.

          • Yes, but Activision at the time had a different "best possible score". No one knows why.

            • by lgw ( 121541 )

              They likely didn't ask the guy who wrote the code.

              It always amazes me, as a dev, the bizarre idea in gaming that "the devs" actually have a 100% correct understanding of how the code works. Very rare, in my experience.

    • by Ksevio ( 865461 )

      Thanks to this, my record of 5.52 seconds is now on top.

      Each frame of the game is roughly 0.03 seconds so it's not actually possible to get 5.52 - you should claim your record is 5.54 instead

      • Each frame of the game is roughly 0.03 seconds so it's not actually possible to get 5.52 - you should claim your record is 5.54 instead

        well todd didn't bother with such finesse so why should him? (the guy had some other games, where you couldn't even get a score ending in a 5 yet he put in a score with that into the tg db..

  • not the same machine (Score:2, Informative)

    by KiloByte ( 825081 )

    Tool-assisted means in an emulator. The vast majority of emulators are at most cycle-accurate, which in some cases changes observable behaviour. Also, it's possible it was a different version of the game -- a lot of game rips are not bit-to-bit identical; versions for different markets notoriously have slight or not-so-slight alterations beyond just translated messages. Likewise, PAL vs SECAM vs NTSC have different timings that often alter the game.

    • by Anonymous Coward on Monday January 29, 2018 @04:47PM (#56029005)

      All of that was investigated and discussed in the 271 page Twin Galaxies forum thread linked from the article, as well as in other places like TASVideos (where the tool assisted run was published.) Every known version of the game has been disassembled and analyzed, including looking for things like regional differences. 2600 emulation is very well understood at this point in time.

      It's possible that all of this analysis had an error in it, of course; but you'd need to do better than some vague "what if"s against a mountain of facts.

    • Tool-assisted means in an emulator. The vast majority of emulators are at most cycle-accurate, which in some cases changes observable behaviour. Also, it's possible it was a different version of the game -- a lot of game rips are not bit-to-bit identical; versions for different markets notoriously have slight or not-so-slight alterations beyond just translated messages. Likewise, PAL vs SECAM vs NTSC have different timings that often alter the game.

      Possibly, but were a lot of discrepancies and impossible scores entered, I think one record had a value of like 15,000,000 exactly where the other top scores were in the thousands. And most of the records were recorded under very dubious circumstances (with his friend as the referee and only witness).

      I suspect Twin Galaxies knew a lot of the records were bogus, but a celebrity is better PR than a cheating scandal.

    • by Eloking ( 877834 )

      Tool-assisted means in an emulator. The vast majority of emulators are at most cycle-accurate, which in some cases changes observable behaviour. Also, it's possible it was a different version of the game -- a lot of game rips are not bit-to-bit identical; versions for different markets notoriously have slight or not-so-slight alterations beyond just translated messages. Likewise, PAL vs SECAM vs NTSC have different timings that often alter the game.

      Yeah, I was wondering too if they were miss something.

      I'm not saying the guy is legit, but I still wonder if Todd Rogers really cheated and if all those people analysing the record are forgetting a little detail like that.

      Still, since the only proof he have is a picture of the score, it's pretty poor since he could have hacked the memory to achieve it.

    • by bugnuts ( 94678 )

      What you said about emulators might be true, but is immaterial.

      He was found out because they looked at the code and found it mathematically impossible for the code to generate that time.

      Many other techniques are used to find cheaters. Some of them double-down and have everything removed from the leaderboards as not only being cheats, but haven't learned to stop lying. Others have confessed, and recreated some of their legitimate scores.

      Here's a pretty good video about speed-run cheaters. [youtube.com]

      • Mathematically impossible on correct hardware. Not necessarily on the piece he had. 2600 hardware wasn't anywhere close to reliable, and timing errors were prevalent.

        And, analyzing just the source code is not always enough. For example, given the following:
        size_t v = n + offset + page_address(page) - page_address(head);

        if (likely(n <= v && v <= (PAGE_SIZE << compound_order

  • It really took 35 years... 35 YEARS to figure out that it was physically impossible, and deemed invalid?
  • by mykepredko ( 40154 ) on Monday January 29, 2018 @04:43PM (#56028983) Homepage

    For having the longest standing video game record being declared impossible.

  • I could see his machine being a bit overworked and perhaps there was something not 100.01% right in the hardware anymore. Maybe it clocked slightly different or something.

    But his statement about starting in second gear when a code review says that's not possible makes it really suspect.

    Occam's razor in me says - crappy TV in 1980 and a 7 looked like a 1 :)

  • If you want more info these videos are great.

    Some more info [youtube.com] about the other sketchy high score stuff this guy has been up to. Dragster is just the tip of the iceberg.

    Ben Heck builds some TAS hardware to attempt to verify the 5.51 Dragster record, using feedback from Todd Rogers himself. The attempt ultimately fails, with Todd's help only getting a 5.6-5.7 while plugging data in from deniers of Rogers' record worked first try for a 5.57 (not counting a data entry mistake).Part 1- Building the hardware [youtube.com] Part 2 - Trying to reproduce the record [youtube.com] Interestingly, nobody comments on camera about the failure.

    • by Ecuador ( 740021 )

      Thanks for the videos. I got more curious and read a bit of the dispute thread. It is crazy. Apparently there is this guy called Jace Hall who is "Head of TG" and he uses (well, used to for months at least) the most retarded arguments to defend that cheating player. My favorite one was where he is explaining that a model that only simulates the gameplay cannot be comprehensive and he proposes an example comprehensive method: read at your own peril! [twingalaxies.com]. Because he claimed in another post that he is talking abou

  • Wow! This guy's the real life version of Peter Dinklage's character, Eddie Plant, in the movie Pixels!
  • by Hussman32 ( 751772 ) on Monday January 29, 2018 @04:59PM (#56029119)

    Watch "The King of Kong" if you want to get a flavor for what the competitive video game community is like. The people who make up the players and judges are, oh, how to say it politely, different.

  • by OmnigamerSDA ( 5246009 ) on Monday January 29, 2018 @05:38PM (#56029401)

    Hi, I'm Omnigamer and I initially investigated this score back in April/May 2017. I performed the reverse engineering on the game code, and developed the spreadsheet model. You can find more information in my initial post on reddit, which also includes links to the Dragster simulator spreadsheet: https://www.reddit.com/r/speed... [reddit.com]

    Just to answer a few other technical questions being brought up in the comments:
    -Accuracy of emulators isn't part of the equation here, since the models were drawn up from machine code. You can argue that there may be some other anomalies in the system, but so far none have been discovered or observed in the wild. That said, the game lives almost entirely within the MOS 6507 in the Atari, which is among the most studied processors on the planet.
    -Changing the system clock would have no effect on the end time; the displayed timer increases by a fixed .0334 every gameplay frame per player. A faster system clock would also impact video output, as other commenters have noted.
    -The currently available "optimal" solution for in-game parameter of distance is known, and cannot reasonably be performed by human hands. This time is a 5.57, and is about 150 distance units from being a 5.54. The best available human strategy is about 220 distance units from a 5.54. Covering that remaining distance would require a breakdown of multiple game mechanics.

    I'm happy to answer other technical questions as well, either here or on my Twitter ( @TheOmnigamer ). Thanks!

  • This is all well and good, but the real question is what was his score on Desert Bus.

E = MC ** 2 +- 3db

Working...