Todd Rogers banned from Twin Galaxies and records removed

[Supergun]

Mindy Jongo:

"...the optimal time is 5.57..."

 

Thomas Jentzsch:

"...And even then 5.54 was the best result..."

 

I have nothing further to say today.

[Thomas Jentzsch]

Me too.

[CapitanClassic]

Mindy Jongo:

"...the optimal time is 5.57..."

 

Thomas Jentzsch:

"...And even then 5.54 was the best result..."

 

I have nothing further to say today.

One is a simplified model

 

The last tab of the sheet contains a much simplified model, which completely ignores gas and tach management. It assumes that the tach is always at maximum, which results in an always maximal acceleration. Only when shifting, the acceleration is zero for one frame. Therefore the results are equal or better than those of the complex model. This should help understanding where the limits of the game are.

[BillyHW]

Has anyone tried frying the Atari to get a better score?

[Bryan]

Just making sure my math is right here, but the reason all possible cases cannit be tested is that the possible permutations are in the range of 18 trillion.

 

6 possible joystick states.

Button UP/Down * Joystick Left/Neutral/Right

 

30 frames per second when JOY status is checked * 5.5ish seconds.

 

165 P 6 = 18 trillionish

 

I dont think it is necessary to test all test cases, because Todd already proved himself a liar by stating he started in 2nd gear, which the game code shows is impossible.

 

It really isn't that bad because you can easily throw out cases with inputs that the game doesn't care about and sub-optimal progress is pretty easy to cull out (like not accelerating for a frame). Once you're down to only the permutations of meaningful actions, you could write a program to run through them all.

 

The only other possibility becomes if someone finds a bug that makes a poor run suddenly return a good time.

[Osgeld]

One of the programs that came with our apple ][ was a paint program (not from apple but from the dealer running a promo) 1 keystroke and boom full screen

 

And people could use tape as storage that was common I have to wonder what mental loopholes one has to jump though to think computers didn't have any form of storage outside of an expensive floppy

[neotokeo2001]

I am not reading all 14 pages but has anyone gone and did a re-dump on a 1980 Dragster cartridge to test that BIN? Or is everyone just taking the same BIN that is everywhere on the net and using that one. Maybe there is a difference. BIN's can be re-named and shared over the years so it may say 1980 but not be. What bothers me are the other players who claimed to get 5.51 and then the 5.51 also being repeated with Activision employees witnessing.

 

Also what systems were available at that time. Heavy sixer, light sixer? The game needs to be tested on each system as there may be differences.

 

This may have already been done, if so then carry on...

[Thomas Jentzsch]

I have wondered about this too. There is no sign of a different version existing, but maybe we should ask around for dumping more Dragster carts (the older the better). Just to be sure.

[MaximRecoil]

And people could use tape as storage that was common I have to wonder what mental loopholes one has to jump though to think computers didn't have any form of storage outside of an expensive floppy

 

I have to wonder what mental loopholes one has to jump though to think that anyone in this thread claimed, suggested, or even hinted that computers didn't have any form of storage outside of an expensive floppy.

[accousticguitar]

Dave Crane thinks 5.51 is possible, and that is good enough for me since he programmed the game of Dragster.

[Mingy Jongo]

Mindy Jongo:

"...the optimal time is 5.57..."

 

Thomas Jentzsch:

"...And even then 5.54 was the best result..."

 

I have nothing further to say today.

You left out the important part of Thomas's post:

"So I created a simplified scenario which ignores almost all complications. It only does the shifting at the latest possible time (since any shift slows down the progress for all coming frames) before the engine blows. All other parameters are either ignored or set to most optimistic values.

And even then 5.54 was the best result, definitely not 5.51."

Re-read the first three sentences again. 5.54 was the maximum achieved by ignoring fiddly parameters that would only further negatively affect the time. WITH those extra parameters included to match the real code, the best possible time is even slower. This is analogous to your own Barnstorming example where something that could only make you slower was removed and a perfect run still wasn't fast enough. The scenario Thomas made changes certain parameters to be more favorable than they could possibly be during normal operation in return for a model that could be analyzed completely, and it STILL wasn't fast enough.

 

I will admit that I overreached and retract my claim that 5.57 was proven to be the fastest -- apparently there is still potential wiggle room in certain areas, but after months of people and computers running through different strategies, it still has yet to be beaten. THAT SAID, given that the lowest time that can be reached in the simplified model is 5.54, and adding/adjusting additional parameters can only hurt your time, we can say that on an unmodified game, 5.51 is impossible, 5.54 might be possible (though people that know both know way more than I do on the specific mechanics of this game and have spent dozens of hours analysing different scenarios tend to think it's impossible), and 5.57 is the lowest score that has been reproducible. In short, the fastest possible time is either 5.54 or 5.57, with statistical evidence and expert opinion heavily favoring the latter.

[Bryan]

Also remember that Todd has given different explanations of how he got his time, including shifting into 2nd before the race starts.

[Osgeld]

 

I have to wonder what mental loopholes one has to jump though to think that anyone in this thread claimed, suggested, or even hinted that computers didn't have any form of storage outside of an expensive floppy.

 

 

And to save you would need to have a separate storage device (e.g., a floppy drive), because the Atari 8-bit computers didn't include one by default, and neither did other affordable home computers of the time, such as the Commodore 64, TI-99, VIC-20, and so on (not that I know of anyway).

 

 

yea there was no storage device available period just cause a computer didnt come with one, and yea every single paint program of the day I ever used had a full screen option

 

jeezus christ its not that damn hard to do man, what do you think people did with their storageless computers, print everything out (which cost more than the stupid computer did FYI) then type it all back in the next morning

 

you are grasping at some really crappy straws

[Mingy Jongo]

Dave Crane thinks 5.51 is possible, and that is good enough for me since he programmed the game of Dragster.

There is a big difference between thinking 5.51 is possible in the affirmative sense, and between not being sure if it is possible or impossible. It is clear from this interview that he falls into the latter camp, being extremely careful to not say anything for certain: http://www.twingalaxies.com/feed_details.php/87/dragster-designer-without-a-shadow-of-a-doubt-about-todd-rogers-record

 

That said, programs can be complicated, and if you think what the programmer says about what can happen in their program should be held to such a high standard of authority, then you should be extremely surprised at how plentiful software bugs are.

[neotokeo2001]

We have been assuming that everyone has been using car #1. What if they are using car #2 at the bottom of the screen? The screen updates from top to bottom, sometimes the times do not match up. Check out this emulator screenshot:

 

 

 

Thomas Jentzsch:
"...And even then 5.54 was the best result..."

 

The top car can show 5:54 and the bottom can show 5:51.

 

That would explain how multiple people claimed the 5:51. What do you think Thomas? Worth looking into or not?

 

EDITED: It looks like Todd does use the second car sometimes.

[+Random Terrain]

A great programmer doesn't have bugs in his games. He has undocumented features.

[CapitanClassic]

We have been assuming that everyone has been using car #1. What if they are using car #2 at the bottom of the screen? The screen updates from top to bottom, sometimes the times do not match up. Check out this emulator screenshot:

 

Dragster (1980) (Activision)_12.png

 

 

The top car can show 5:54 and the bottom can show 5:51.

 

That would explain how multiple people claimed the 5:51. What do you think Thomas? Worth looking into or not?

 

EDITED: It looks like Todd does use the second car sometimes.

My guess is that is due to the fact that P1 is updated on even frames, and P2 is updated on odd frames. I wonder if Ben Heck could run his simulator using player 2 and Omnigamers spreadsheet and get the same 5.57 result.

[neotokeo2001]

My guess is that is due to the fact that P1 is updated on even frames, and P2 is updated on odd frames. I wonder if Ben Heck could run his simulator using player 2 and Omnigamers spreadsheet and get the same 5.57 result.

 

I think at the very least it should be checked.

[MaximRecoil]

 

yea there was no storage device available period just cause a computer didnt come with one,

 

That's an utter non sequitur, given that exactly no one in this thread claimed, suggested, nor even hinted that there was "no storages device available"; quite the opposite in fact. And you've still yet to explain what "mental loopholes you had to jump through" to think that anyone in this thread claimed, suggested, or even hinted that computers didn't have any form of storage outside of an expensive floppy. I suspect you don't know what the term "e.g." means. It is an abbreviation of exempli gratia, which is a Latin term that translates to, "for example", and by definition the example(s) given is/are not exhaustive. So when I said...

 

"And to save you would need to have a separate storage device (e.g., a floppy drive), because the Atari 8-bit computers didn't include one by default, and neither did other affordable home computers of the time, such as the Commodore 64, TI-99, VIC-20, and so on (not that I know of anyway)."

 

... that means that a floppy drive is one of multiple examples of a separate storage device.

 

and yea every single paint program of the day I ever used had a full screen option

 

What of it?

 

jeezus christ its not that damn hard to do man, what do you think people did with their storageless computers, print everything out (which cost more than the stupid computer did FYI) then type it all back in the next morning

 

Many people (especially kids) who owned those affordable home computers in the early 1980s used them for video games. The e.g., TI-99/4A, Commodore 64, Atari 400/800, VIC-20, had cartridge ports.

 

The Apple IIe was quite expensive; I didn't know anyone who owned one when I was a kid. They were common in schools, and that's the only place I ever used one. They also did not have a built-in or otherwise included RF modulator, which complicates the process of connecting one to a typical TV in 1982.

 

you are grasping at some really crappy straws

 

My list of uncommon things which would have had to come together for the "generated on a home computer and output to a TV" theory remains unfazed by your efforts.

Edited February 11, 2018 by MaximRecoil

[neotokeo2001]

The edited photo theory is a dead end. The 5:51 was supposedly accomplished in person at a trade show. I think the car #2 thing may be on the right track to explaining it.

[Mingy Jongo]

 

I think at the very least it should be checked.

The P2 car theory is debunked fairly early in the TG thread, though I can't remember exactly where -- a search should bring it up. I read a couple hundred pages of it a few days ago and had to wade through a loooooot of crap to find the nuggets of explanation and genuine questions/answers in it.

[Andromeda Stardust]

Of course not.

And IMO the evidence here is definitely close enough.

While I generally agree with everything you have stated thus far, especially a simplified theoretical model prooving that 5.54 is the absolute minimum theoretucally possible time, and a more complex model showing that 5.54 is improbable, and so far no series of inputs has yielded less than a time of 5.57, I strongly caution the use of the word "close". A wise man once said:

 

Close only counts in horseshoes and thermonuclear warfare

A few years back Google prooved beyond a shadow of a doubt that no possible position on a Rubick's Cube requires more than 20 moves to solve. A bunch of "hard" cases were proven to require 20 moves, one such imfamous position being the superflip, but at the time it was unknown if any position required 21 or more moves to solve. The number of unique positions on the cube totalled so high ( {8! * 3^7 * (12!/2) * 2^11} = 43,252,003,274,489,856,000) the world's current computing resources since the dawn of the universe would not be enough to brute force solve the Rubick's Cube problem. Excluding rotations and symmetries can cut this number down a significant amount. Furthermore someone drafted a proof that any "hard" position must exhibit a certain degree of symmetry because from an optimally hard position, literally any manipulation to an asymmetric position would result in decrimenting one from the remain8ng number of moves required to solve. In short, this "proof" reduced the number of possible positions to a few million unique cases, which could then be brute forced by distributed computing. And that's exactly what Google did. By running the task on idle computing cycles across Google's campus, it was prooven in a short couple of months, that no existing position on the Rubick's Cube requires more than 20 moves to solve.

 

Why did I bring this up? The Dragster problem in general seems to be simpler that the Rubick's Cube, and a fairly easy case compared to many other video games. If one analyzed all possible inputs on a frame per frame basis, most inputs would either: A have no effect, or B become a detriment to the final outcome. The remaining combinations of inputs, C, only serve to improve the final score. Taken on a frame by frame basis, software has shown that no known combination of inputs can result in a time of less than 5.57, and it is theoretically impossible to score under 5.54. A brute force attack on the scale of the Rubick's Cube project could likely proove or disproove once and for all if a 5.54 score is possible, especially if weeding out the A,B inputs specifically via training the AI.

 

But all of this means absolutely nothing with regards to Todd Rogers bogus 5.51 score. It is theoretically impossible to score less than 5.54 and verified 5.57 scores exist. So whether or not an unlikely sequence of inputs exists that can achieve 5.54, a time below that is prooven impossible so Twin Galaxies was right to invalidate Todd's score of 5.51. And I don't believe Dragster is relevant enough in modern pop culture to facilitate the kind of brute force computing effort necessary to disproove the possibility of a 5.54 time score. Someproblems only God knows the answer to. I for one would like to see a rigorous proof for Collatz conjecture https://en.m.wikipedia.org/wiki/Collatz_conjecture (now that Fermat's last theorm is solved).

[Thomas Jentzsch]

 

I think at the very least it should be checked.

It has been checked. Both cars are running on alternative frames and have their own timing.

[Thomas Jentzsch]

There is a big difference between thinking 5.51 is possible in the affirmative sense, and between not being sure if it is possible or impossible. It is clear from this interview that he falls into the latter camp, being extremely careful to not say anything for certain: http://www.twingalaxies.com/feed_details.php/87/dragster-designer-without-a-shadow-of-a-doubt-about-todd-rogers-record

 

That said, programs can be complicated, and if you think what the programmer says about what can happen in their program should be held to such a high standard of authority, then you should be extremely surprised at how plentiful software bugs are.

So true. When I programmed Thrust, I was convinced that no one could roll the fuel. And, guess what, a few years later, I had to add a bug fix.

 

And that's only one single episode from a developer's life.

[Sheddy]

 

I once knocked up a quick script where I assumed I could brute force best possible arrangement of what was normally 6 or 7 items, but could occasionally be up to 15. The time wasted trying to debug was a bit embarrassing after I went back to basics of permutations...

Edited February 11, 2018 by Sheddy