Todd's 5.51 Dragster score

[Thomas Jentzsch]

My whole point wasn't about fame, but to illustrate the fact the concern over this score isn't limited to Twin Galaxies. It's a lot bigger than people think.

What do you mean? What is "it's"?

[Thomas Jentzsch]

An Apology

 

More than 10 years ago, I created a quick and dirty hack, which provided automatic shifting to Dragster. I had almost forgotten it until yesterday, when the topic about the 5.51 seconds came up and someone showed my hack. Then I experimented a bit with it myself and soon was able to reach an even better time. That was cool, because I thought I had found Todd's trick and the proof for Todd's time.

 

Today I went back and disassembled my hack. I wanted to make sure that there are no flaws and no cheating involved compared to the real game. In hindsight, with all the unexpected (for me) flurry this created, I should had done this yesterday. Because I found a flaw. Actually even two of them. And this changes everything!

 

Here is my old hacked code in question:

    cmp     #$20            ; A = tach
    bcc     .notBlown       ; 
    lda     #$1f            ; limit tach to maximum value; 1st flaw: the original game subtracts 1 if you release fire. and
                            ;  you have to do that *before* it blows the engine. So the tach would jitter between $1e and $1f
    ldy     countDown       ; 
    bne     .noShift        ; prevent shifting early before timer = 0
    ldy     gearLst,x       ; 
    cpy     #$04            ; 
    bcs     .noShift        ; skip shifting if already in 4th gear
    inc     gearLst,x       ; shift
.noShift
    ldy     gearLst,x       ; load current gear; 2nd and major flaw: bit 7 could be set, and the following code (not included) 
                            ;  loops way to often, this causes the screen rolls
    nop                     ; padding...
    nop                     ; 
.notBlown
    sta     tachLst,x       ; save new tach value

So, all results retrieved with my old hack are irrelevant. And I am VERY sorry for causing the trouble (here and especially at TG too).

 

I now have changed my hack to address the two flaws above.

    cmp     #$20            ; A = tach
    bcc     .notBlown       ;
    lda     countDown       ; prevent shifting early before timer = 0
    bne     .noShift        ; 
    cpy     #$04            ; Y contains the corrected gear (which checks bit 7) from some code above
    bcs     .noShift        ; skip shifting if already in 4th gear
    inc     gearLst,x       ; shift
.noShift
    lda     #$1e            ; this is an approximation of the maximum tach value after releasing the fire button 
    nop                     ; padding...
    nop                     ;
    nop                     ;
    nop                     ;
    nop                     ;
    nop                     ;
.notBlown                   ;
    sta     tachLst,x       ; save new tach value

The best time I can achieve with this hack so far is 5.71 seconds (clutch used in 3rd and 4th gear). To me it now seems that there is no big difference in physics to the original version anymore.

 

EDIT: V3 fixes the tach during countdown which must be a multiple of 3. I once reached 5.61 with this one.

Dragster Automatic Shifting V2 (Dragster Hack) (2017) (Thomas Jentzsch).bin

Dragster Automatic Shifting V3 (Dragster Hack) (2017) (Thomas Jentzsch).bin

Edited August 24, 2017 by Thomas Jentzsch

[homerwannabee]

Thanks for the clarification. I appreciate it.

[Tempest]

So Todd's score may be bogus after all then?

[The Evener]

At minimum, if the original exercise was to demonstrate that 5.51 or faster was humanly possible with an automatic shifting ROM - and thereby indirectly supporting Todd's time of 5.51 - then the question is back in play with the updated ROM (thus far).

[moycon]

At minimum, if the original exercise was to demonstrate that 5.51 or faster was humanly possible with an automatic shifting ROM - and thereby indirectly supporting Todd's time of 5.51 - then the question is back in play with the updated ROM (thus far).

So I'm confused. Is Todd's score feasible or not?

On the one hand that would suck, but on the other hand there's a freckle.

[homerwannabee]

The argument is this. Is that not only is Todd not able to do it, and not only is no one else able to do it, but it's not possible if you TAS frame by frame inputs. So either Todd found a trick that hasn't been thought of by anyone else or Todd's got really bad memory, and mistook a 5.57 for 5.51 those three times.

[Thomas Jentzsch]

Omnigamer posted a nice explanation video here. He pretty nicely explains the maths behind the game and the theoretical limits.

 

Note: I have NOT validated yet that everything is correct.

[Thomas Jentzsch]

I took the time and looked deeper into the Dragster code and compared it with Omnigamer's sheet. I even created my own sheet based on his one. And I did not find any mistake or missing parameter.

 

And I created (another) dragster hack, which uses the input from Omnigamer's sheet and feeds it into the input. And the result is exactly like his sheet.

 

Then I played with the sheet and found no combination which improves Omnigamer's result (5.61) from the sheet. However the distance at 5.57 (96.95) is very close to the target distance (97), so only a minor change would cause the time to get better (5.57).

 

Also Omnigamer states in his sheet:

 

Please contact me if you can get a time faster than 5.57, or a 5.57 with distance greater than 97.25.

 

This seems to indicate that he already found an improvement. Which means that my findings are not optimal.

 

Nevertheless, based on my current knowledge, I have strong doubts that 5.51 is possible with the dumped NTSC ROM.

Dragster (Autoplay 5.61) (2017) (Thomas Jentzsch).bin

Edited August 27, 2017 by Thomas Jentzsch

[homerwannabee]

This is sort of saddening to hear. The way you started out in the thread it's obvious that you didn't have a grudge against Todd. If anything, you were pro Todd Rogers.

Omni Gamer I don't know from Adam, but I have been around the community long enough to know you are an upstanding person.

So based on this evidence, it's not looking good for Todd.

He's been so entrenched in talking about this score, that I fear a coffee stain excuse won't work this time.

[Thomas Jentzsch]

Meanwhile I was able to improve a little (5.57), but with the current approach 5.51 seems impossible.

 

However, Todd stated that he was able to start the game in 2nd gear ("I told them how I engage the clutch until 0 and then pop the clutch, already in 2nd gear."). If this would be the case, then 5.51 might be possible. But I haven't found a way to make this happen yet.

 

Any ideas?

Edited August 27, 2017 by Thomas Jentzsch

[Omnigamer]

Hello Thomas! Just wanted to thank you for digging into the game yourself.

 

As for improving your time, it sounds like you enjoy the intellectual challenge of it, so I won't spoil it for you. I don't know what strategies you have been trying, but there are a couple of un-intuitive ways to get better distance. I have a guess where you can see some improvement though: closely examine your end-game and shift into 3rd. The best possible distance is 24930 (97.38) at 5.57 according to my model.

 

A fellow on YouTube used my spreadsheet model and exhaustively checked it via a dynamic programming algorithm. He confirmed that the input set I used for my TASVideos submission is one of the optimal sets, and also gave a generic listing of the optimal set with required inputs with don't-cares. I have not performed this type of analysis myself, though.

 

As for starting in 2nd gear, or even getting into 1st during the countdown, I also was not able to identify any means of doing so. Even assuming there was a way to do it, I couldn't identify any pattern where having your first shift go to 2nd would enable better than 5.54. The only way it would allow a 5.51 is if you were already in 2nd on the first frame without needing to shift into it. You can double-check for yourself though! I didn't spend much time testing those specific scenarios due to how unlikely they were.

[Troy_Whelan]

Meanwhile I was able to improve a little (5.57), but with the current approach 5.51 seems impossible.

 

However, Todd stated that he was able to start the game in 2nd gear ("I told them how I engage the clutch until 0 and then pop the clutch, already in 2nd gear."). If this would be the case, then 5.51 might be possible. But I haven't found a way to make this happen yet.

 

Any ideas?

 

The conversation I remember regarding starting in 2nd gear is this: Have the clutch depressed during the countdown, then when timer reaches zero quickly release-tap-release thereby starting in 2nd gear. The main point, other than spending little to no time in 1st gear, was to depress the clutch during the countdown.

[Thomas Jentzsch]

Hello Thomas! Just wanted to thank you for digging into the game yourself.

Just for fun, definitely not even remotely as serious as you guys at TG.

 

I used to have a few old records (Pitfall! and Starmaster, both PAL) for the 2600, all from the 80ties. Only based on Polaroids, so I suppose all have been deleted by now.

 

As for improving your time, it sounds like you enjoy the intellectual challenge of it, so I won't spoil it for you. I don't know what strategies you have been trying, but there are a couple of un-intuitive ways to get better distance. I have a guess where you can see some improvement though: closely examine your end-game and shift into 3rd. The best possible distance is 24930 (97.38) at 5.57 according to my model.

That's my best so far too.

 

A fellow on YouTube used my spreadsheet model and exhaustively checked it via a dynamic programming algorithm. He confirmed that the input set I used for my TASVideos submission is one of the optimal sets, and also gave a generic listing of the optimal set with required inputs with don't-cares. I have not performed this type of analysis myself, though.

I have created a simplified spreadsheet which doesn't care for the tach. Thus it allows accelerating by 2 every frame, except when there was a shift in the frame before. The maximum speed is 253.

• With no shifts at all, the best time is 5.41
• With the bare minimum shifts executed (4) at the latest possible speed (0, 32, 64, 127; without shifting the speed would not increase any further), the best time is 5.54.

Even though the latter ignores some necessary, time increasing steps (e.g. tach limits and extra shifts in 4th gear), the time is already above the record's time of 5.51. So unless I am missing something here, in theory it does not look good for the record.

 

As for starting in 2nd gear, or even getting into 1st during the countdown, I also was not able to identify any means of doing so. Even assuming there was a way to do it, I couldn't identify any pattern where having your first shift go to 2nd would enable better than 5.54. The only way it would allow a 5.51 is if you were already in 2nd on the first frame without needing to shift into it. You can double-check for yourself though! I didn't spend much time testing those specific scenarios due to how unlikely they were.

Yes, I think that's not possible. Most likely Todd meant that he is shifting to 2nd immediately. Which I cannot see resulting into improving the time.

Edited August 28, 2017 by Thomas Jentzsch

[The Evener]

Your record on Starmaster is there, but with entry date of 2002. Pitfall! doesn't show your record under PAL. The database has undergone many forms (notebook, file on TRS-80 I think, then a few other variations/migrations since 2000) - it's known that scores were lost during the migrations, and sometimes metadata for existing scores.

Edited August 28, 2017 by The Evener

[Thomas Jentzsch]

I think I found my old Polaroids for Starmaster back then. For Pitfall! I could not find it, but I later proved that my time (1:42) is possible with a modified emulator. IIRC shortly after that, even better records and validated were posted.

[Serguei2]

I'd really love to see this score broken using original game on original 2600, not on a hack version using an emulator.

 

What a shame.

[AlyoshaTAS]

I didn't realize a formal challenge was made on this, thanks for making this thread.

 

I think Omnigamer's work is pretty thorough, and with thomas Jentzsch reaching the same conclusion I don't think there are really any open ends here.

 

But now there will be some kind of live event? That's an interesting result.

 

I'll be interested to see what comes from all of this.

[Omnigamer]

I've been spending some time re-checking situations which might allow you to skip the memory wipe, but haven't found anything worthwhile. Some tomfoolery with the difficulty select switch puts it in an odd state, but you can't "start" a race without either pressing right or reset, which both trigger the memory wipe. Basically, the only way I see for getting a faster time is by being able to start a race with the memory in a "dirty" state. I tried some basic input fuzzing before, but that came up empty as well.

[enthusi]

Maybe I missed it but what is the recorded 2nd best?

Is Todd likely to be 'the best' still with whatever score? Or have many reached 5.57 or 5.54?

[D.Yancey]

May I suggest that history may be repeating itself here. Activision originally created a computer simulation of a perfect run in Dragster to be 5.54. Along came Todd and broke that time. Activision scrutinized his performance. They even required him to repeat his performance. Activision found Todd's 5.51 to be legitimate. Activision even contacted Guinness and had the score submitted and verified by Guinness. Now 35 years later, it's happening all over again. May I ask exactly what is it that makes the data of today's analysts more reliable than that of the original programmers of the game at Activision? Why should any of us believe that they somehow know more than David Crane and the Activision team?

 

May I suggest that an effort be made to look for ways to PROVE Todd's score rather than to discredit his score and his integrity. Maybe there could be more then one rom version of the game that was publically released. If there were, the differences would be so subtle as to be undetectable by the human eye. Maybe there could be a nuance at the start of the timer similar to the glitch in Night Driver that allows a perfect score of 81 vs a perfect score of 80. Maybe the distance required to finish the race could be ever so slightly less. Maybe the tach could allow an ever so slight bit of extra rpm's before the engine blows. There ae endless possibilities.

[Tempest]

They even required him to repeat his performance.

Is there evidence of this somewhere? This is the first time I've heard this.

 

The issue is that most of us are having a hard time believing Todd found some super secret method of getting 5.51 since all the code analysis and data seems to point that it's impossible. It's true that we can't prove that there isn't some super rare rom variation out there that allows it or that there might be some console glitch that makes it possible, but so far all the evidence seems to point to the score being bogus. It doesn't help that in many people's eyes he already has sort of a black mark as some of his other scores seem to be impossible as well.

 

In the end it's all just a game with the only prize being some bragging rights so who really cares? But I'm also a bit puzzled by the 'don't look into this too closely. Just let it go' crowd. I'd like to know the truth just because I like a good mystery and have absolutely no stake in this one way or the other.

[D.Yancey]

Is there evidence of this somewhere? This is the first time I've heard this.

As far as I know, only anecdotal evidence, which I have no reason to doubt.

[Tempest]

As far as I know, only anecdotal evidence,

The best kind of evidence!

[D.Yancey]

It's true that we can't prove that there isn't some super rare rom variation out there that allows it or that there might be some console glitch that makes it possible, but so far all the evidence seems to point to the score being bogus.

I'm not even suggesting that a "super rare" rom exists. I am suggesting that only ONE ROM has ever been analyzed. A few random cartridge roms could be dumped and the code checked to see if it matches exactly. Try an early release of the game, maybe one that contained the sponge dust cover from the really early days of Activision. Try a very late release too. There could be a difference. Maybe since Activision determined that the best possible score had been reached on the game, they created a subtly different rom to create more buzz about the title among the avid gamers in the community of the day. If you believe in Todd's integrity, then we have all the evidence that we need. For those that do not believe in Todd's integrity, will they ever have enough evidence?