Below, you'll find a large collection of animated GIFs. PLEASE take the time to read this ENTIRE post before jumping to them, and please continue reading what is posted AFTER them.
In summary, these GIFs show that each of the Donkey Kong world record direct feed recordings presented by Billy Mitchell and verified by TG were generated in MAME and not by original Donkey Kong hardware. These scores are: 1,047,200 (the King of Kong "tape"), 1,050,200 (the Mortgage Brokers score), and 1,062,800 (the Boomers score). In order to demonstrate this, it's necessary to understand how DK images are generated by MAME versus an original PCB.
Video emulation has changed across MAME versions, but there are basically 4 eras: 0.115 and prior, 0.116 - 0.121, 0.122 - 0.126, 0.127 and newer. Each of these groups or eras show different methods of video emulation as a result of MAMEdev making various improvements. As it currently stands, video emulation is not perfect, but it's closer than it used to be.
The important bit: prior to version 0.127, MAME generated video in essentially a "snapshot" fashion. The Z80 would run for 1 frame, then the video emulation takes the "snapshot" of that point time and displays the image. Then the Z80 runs for 1 frame, the next snapshot is taken, displayed on the monitor, etc. etc. Real DK hardware, on the other hand, generates video in a "rolling shutter" fashion: the video generator scans left-to-right/bottom-to-top, while the Z80 builds the screen in memory from the opposite direction right-to-left, top-to-bottom. Both video generation and the Z80 are running concurrently with no "snapshots". The result is that it becomes fairly easy to determine WHAT generated the image based on HOW the image has been generated.
A simple analogy would be this:
-Real DK hardware generates the image in the same way you would open or close vertical window blinds...from side to side.
-Older versions of MAME (pre-0.122) generate the image in the same way you would put together a puzzle...piece by piece.
Since Billy has claimed that his footage is all direct feed, it's also CRUCIAL that we take a look at an extremely well-documented process of creating a direct feed setup for DK. Chris Gleed went through the entire painful process, documenting each step in his journey. You can read about it in the links below. He discusses his hardware choices and provides photo updates throughout. It is very likely that he has the only true DK direct feed setup in existence and his scores have been accepted at DKF and TG. Footage of his direct feed gameplay is included in the GIF collection for comparison...if we are looking at direct feed footage, THIS IS WHAT WE SHOULD SEE.
Idea: video/audio capture from edge connector
Donkey Kong Direct Feed How-ToFor further comparison, I've included footage from my own DK cabinet. This was recorded using a Samsung 8 phone filming at 60 fps. This gives us a high-framerate comparison of true DK hardware that is NOT direct feed. It also captures the idiosyncracies in filming a monitor with a camera (brief ghost images or double frames). Since existing footage of Billy's 1.047M and 1.062M are cameras-pointing-at-monitor-playback, those sources will exhibit some of the same effects. Note that this is VERY different from the concepts discussed earlier regarding how video is generated...these are artifacts caused by the external re-recording of the original sources, not artifacts present in the sources themselves.
Another MAME behavior that is worth pointing out is rotation. Although we see DK in the correct orientation when we play on a real cab, the video is actually generated on its side, with the top of the screen being on the right-hand side. MAME also emulates this incorrectly. If you've read through Chris Gleed's threads, you can see posted photos of correctly rotated DK video. Billy's 1.047M and 1.05M games show the incorrect rotation. The 1.062M recording, however, appears to be the correct orientation. Since MAME versions haven't changed the rotation, it's likely this discrepancy was noticed at some point and corrected for future recordings (simple as using the -ror command).
A final note before the GIFs: all of this can be independently confirmed by anyone. I am not in sole possession of any critical, secret piece of information here. Knowledgeable hardware experts like Sock Master, who played a crucial role in dissecting all of this, can verify the hardware aspects. Any person can inspect any and all versions of MAME, the provided video sources, and their own hardware if they have it.
So, with all that being said, let's get to the GIFs. These are small segments from each source, showing the transition from the break screen into the next. Each GIF is showing every frame in the transition; there is no cherry picking here. Each frame in the GIF lasts approximately 750 ms to make it easier to see. As for framerates, the MAME sources, Chris Gleed's direct feed, and my cab are recorded at ~60 fps. Each Billy source is ~30 fps. I apologize in advance for the formatting/placement.
They have been grouped by screen type (barrel, pie, spring, rivet) and show a variety of sources:
MAME 0.115 (relevant for versions 115 and below)
MAME 0.117 (relevant for versions 117 - 121)
MAME 0.122 (relevant for versions 122 - 126)
MAME 0.194 (relevant for versions 127 - 194)
Billy 1.047M
Billy 1.05M
Billy 1.062M (pie transition only...this is the only transition footage available for this game)
Chris Gleed 970k Direct FeedTest footage from my personal cab, recorded at 60 fps with a Samsung 8 phone
BARREL TRANSITIONSMAME 0.115 and below MAME 0.116 to 0.121 MAME 0.122 to 0.126 MAME 0.127 to present
Billy 1.047M Billy 1.05M
Chris Gleed 970k Direct Feed Jeremy Young Test Footage - 60 fps camera phone
PIE TRANSITIONSMAME 0.115 and below MAME 0.116 to 0.121 MAME 0.122 to 0.126 MAME 0.127 to present
Billy 1.047M Billy 1.05M Billy 1.062M
Chris Gleed 970k Direct Feed Jeremy Young Test Footage - 60 fps camera phone
SPRING TRANSITIONSMAME 0.115 and below MAME 0.116 to 0.121 MAME 0.122 to 0.126 MAME 0.127 to present
Billy 1.047M Billy 1.05M
Chris Gleed 970k Direct Feed Jeremy Young Test Footage - 60 fps camera phone
RIVET TRANSITIONSMAME 0.115 and below MAME 0.116 to 0.121 MAME 0.122 to 0.126 MAME 0.127 to present
Billy 1.047M Billy 1.05M
Chris Gleed 970k Direct Feed Jeremy Young Test Footage - 60 fps camera phone
Beyond the evidence above, it's worth pointing out some other questions and issues:
1) While clearly MAME, there is no way to tell exactly how these games were performed. It's possible they were recorded in one shot. Given the play style in Billy's videos, it's more likely that vanilla MAME's INP recording feature was abused, in which a person can re-record an INP as many times as they wish to craft their desired performance.
2) While many people have seen Billy play in public, there are no known independent, impartial, objective witnesses to any of the The Big 3 WR games. He has never scored over 1,000,000 in a live venue. Billy claimed the 1.047M was done in front of scores of people, but that he had no access to the inside of the machine...so how did he set up the direct feed? The 1.05 was supposedly done at an actual convention, but Billy was conveniently playing in another room. The 1.062 was done in arcade in Florida, but the only live footage from that day was staged (
the Boomer board swap) and shows no evidence of a direct feed setup. Todd Rogers, of Dragster infamy, was a supposed witness to the 1.05M and 1.062M games.
3) It's unclear if ANYONE ever reviewed the 1.062M recording. According to Patrick Scott Patterson, the TG refs present at the IVGHOF event in Ottumwa in 2010 simply voted on whether to accept the score, without reviewing any footage. While PSP's credibility in anything DK-related is questionable, his version of those events is still worth noting.
4) For Billy, who has taken special care to point out how little he knows about DK hardware, it would have been essentially impossible to create a direct feed setup in the years he was submitting tapes. So, who would have done it for him and why has that person never been a part of any verification discussions? Creating a dual-display version for the Mortage Convention (sending quality video to another room, not just a VCR) would have required even more technical know-how.
So...
PLEASE KEEP READINGIn light of the evidence presented here, I am removing Billy Mitchell's current score of 1,062,800 from the Donkey Kong Forum High Score List.This score will be replaced with his live score of 933,900, performed at the Midwest Gaming Classic on May 7, 2004. This score was performed in public and was
witnessed by at least one TG member as well as Dwayne Richard.
As DKF's (currently) sole high score moderator, I've taken my responsibilities very seriously. I've always tried to apply what I felt were the
community standards to every performance. When Corey Chambers was first crafting the original version of the HSL rules, an important section of the "Auxiliary Rules" was that a score may be accepted without following the main rules as long as it was convincingly supported in multiple other ways. This idea was tested early on with MAME games by Robbie Lakeman and Christian Van Meter...performances that were accepted with no INPs. In my view, this door swings both ways. Scores already on the board are ALWAYS subject to review, and if the preponderance of evidence is against it than the score should be removed, even if no single bit of evidence is a "smoking gun." In my view, we have reached that point with Billy Mitchell.
The idea to challenge Billy's scores has been floating around for quite some time now. I lodged a
formal dispute on TG months ago. A
poll on DKF showed an almost even split (10-13) on whether Billy's scores should be removed from our HSL. Top players and former TG officials, some quietly and others not so quietly, have been questioning his gameplay and the circumstances behind his recordings for over a decade. I don't know Billy personally. We've never met and the only time we've been in the same place was KO3. I'm well aware of his importance in gaming history and culture and especially to the arcade and DKF community. I understand this will not be a popular decision. It is not my wish to fracture a community or affect anyone's personal life. But ultimately, I have to do what I think is right.
Short of live, time-stamped, complete footage (including full views of cab hardware) of the games in question, I will not be reinstating any of Billy's scores in question. Should the community as a whole wish for reinstatement, then I will step down as DKF high score moderator. It's worth noting that there are still other forum mods (and an admin) that can update the scoreboard if they so choose.
If this community (and others like it) are built on the idea of friendship through competition, camaraderie through our shared pains in pushing ourselves, our friends, and these games to their limits, then we must strive for honesty and legitimacy. A house built on lies is not worth living in.
EDIT: The original transition GIF for the rivets for MAME version 0.116-0.121 was a duplicate of the version 0.115 GIF. I've updated the post with the correct GIF.
EDIT #2: ersatz_cats did some great work, tracking down video of an MTV interview from 2006 with RTM showcasing the 1.047M and further demonstrating all of the stuff from above:
A lost interview with Mruczek has revealed more evidence against Billy Mitchell
