weekly contest VGS NES Weekly Contest - Dr. Mario

[docile tapeworm | 4,198]

@Dr. Morbis @Krunch i think one constant rule for the contest should be you put the cartridge in the nes, power/reset, og controller, thats it.

ya theres holes in that but its kinda like a gentlemans agreement.

then there is wampson @bertsampson....he is a high score cyborg. dude can just "play games forever" did it dk jr too...

Edited July 13, 2021 by docile tapeworm

[mbd39 | 2,020]

The contest moderator Bea Iank is a programmer who used to analyze code of games when she played in the contests. I remember she did it with Klax. 

I don't know about exploiting known glitches though.

 

[docile tapeworm | 4,198]

24 minutes ago, mbd39 said:

The contest moderator Bea Iank is a programmer who used to analyze code of games when she played in the contests. I remember she did it with Klax. 

I don't know about exploiting known glitches though.

 

booooooooo

[BeaIank | 1,618]

The situation is being discussed internally.
Looking at the code to analyze it, figure patterns and stuff is good on my book.
The whole simulating TAS inputs and exploiting that glitch, though, falls into a VERY grey area.

[0xDEAFC0DE | 1,277]

9 hours ago, Dr. Morbis said:

I don't feel like copying TAS inputs frame for frame on a puzzle game really falls within the spirit of our VGS weekly NES competition, but whatever.  People are gonna do what they're gonna do.  My Dr. Mario skills are so laughably bad that I'm like third last, so it doesn't really affect me, but seeing a guy put up a really amazing score in the millions after hours of incredible play "lose" to a guy who followed a TAS for magical points is pretty disappointing...

To be clear I didn't follow any preexisting TAS. That TAS resets between every level to get the perfect RNG seed with doesn't follow the contest rule. The TAS I was referring to following was one that I made during the week. I could have done the same thing with a bunch of savestates and writing the route on paper, just this was easier for me.

5 hours ago, Krunch said:

Yeah I'm not gonna lie I was kinda like ??? with this one once I seen the video, when the contest was still live/just closed I figured he had simply found that way to get constant patterns from the start, and then used that knowledge to create consistent combo chains, which I also thought was quite a bonkers approach to such a frenetic game.  But - this is just some magic point swoop based on overloading the game, and copying the pill arrangement off TAS/WRs no less?  Weird sauce.  Anyway according to Youtube this is a known glitch so maybe it comes down to rule-making and explicitly banning known glitches when it's relevant.  "Don't hate the player, hate the game" might apply but the fact is the NES is a renaissance library resting on loose scoring systems and coding shortcuts, which happened to rescue the video game industry, but also makes these things real breakable.

I did some math on the vid and basically it's roughly clearing 50 viruses for 1,290,200.  That's 25,804 per virus, versus what the manual says on a chain of 6 viruses being worth 12,600 total (2,100 per virus, then.  I can't really remember if 6 is the cap and it just keeps feeding you 6,400s from 7 on, or if the cap is higher, but clearly it's impossible to hit a 25K virus average in normal play).  I wasn't the guy with the amazing score, but I just look at this and say ah shit, I was playin the combos too, but could have just put my effort into sledgehammering the game with an epic combo instead of shooting for consistent 3-4's and as many 6+'s along the way as you can get.

...not that I did well anyways.  My effort was hot garbage couched in a pretty good point average, which is nothing new.

Yeah, for future contests it probably will be better to ban the erase glitch. I'm actually surprised TG allows it considering their normal stance on glitches.

For the combo thing, the bonus per virus is supposed to cap at 3200 at a 6 combo. However, the game get's this bonus based off of table stored in memory and it doesn't check for the length of the combo. So, any combo greater than 11 will start reading garbage values past that table and multiplying by 100. To make matters worse zeros are interpreted as 256 and there are several zeros and values greater than 240 right after the table. I found a table for different combo lengths here.

1 hour ago, docile tapeworm said:

@Dr. Morbis @Krunch i think one constant rule for the contest should be you put the cartridge in the nes, power/reset, og controller, thats it.

ya theres holes in that but its kinda like a gentlemans agreement.

then there is wampson @bertsampson....he is a high score cyborg. dude can just "play games forever" did it dk jr too...

Beside mapping out the route on emulator, I actually followed all that for my submitted score. You can see my CRT which is hooked up to a real NES playing a real Dr. Mario cart with a real controller.

[mbd39 | 2,020]

I think I got 7th place. I'm happy with that considering that I'm pretty much garbage at Dr Mario.

 

[docile tapeworm | 4,198]

51 minutes ago, 0xDEAFC0DE said:

Beside mapping out the route on emulator, I actually followed all that for my submitted score. You can see my CRT which is hooked up to a real NES playing a real Dr. Mario cart with a real controller.

right on. i just dont like using emulation/save states as practice tools. but you still gotta execute. awesome job with the score btw.

Edited July 13, 2021 by docile tapeworm

[Bearcat-Doug | 5,104]

22 minutes ago, mbd39 said:

I think I got 7th place. I'm happy with that considering that I'm pretty much garbage at Dr Mario.

 

I was going to play again just to see if I could crack the top 8 and I totally forgot. I'm nowhere near great at Dr. Mario, but I did own it as a kid so I'm at least average.

[bertsampson | 345]

16 hours ago, 0xDEAFC0DE said:

If you are curious about how all the other viruses disappear at the end, it's the same method used in the TG WR, and also explained by the TAS. TL;DR essentially, if your combo points are too big it will take over a frame to count up. However, right when the new frame starts, the RAM location that stores the endpoint of the line that it's erasing is overwritten by the value of P2's controller. This allows you to manipulate the endpoint to clear all the viruses. (It's a bit more complicated than that, but that's the gist of it at least)

The "value of P2's controller" would be my only exception.  Do you have to have a button pressed on the second player's controller to get the glitch to work?

[Krunch | 303]

4 hours ago, 0xDEAFC0DE said:

To be clear I didn't follow any preexisting TAS. That TAS resets between every level to get the perfect RNG seed with doesn't follow the contest rule. The TAS I was referring to following was one that I made during the week. I could have done the same thing with a bunch of savestates and writing the route on paper, just this was easier for me.

Yeah, for future contests it probably will be better to ban the erase glitch. I'm actually surprised TG allows it considering their normal stance on glitches.

For the combo thing, the bonus per virus is supposed to cap at 3200 at a 6 combo. However, the game get's this bonus based off of table stored in memory and it doesn't check for the length of the combo. So, any combo greater than 11 will start reading garbage values past that table and multiplying by 100. To make matters worse zeros are interpreted as 256 and there are several zeros and values greater than 240 right after the table. I found a table for different combo lengths here.

Beside mapping out the route on emulator, I actually followed all that for my submitted score. You can see my CRT which is hooked up to a real NES playing a real Dr. Mario cart with a real controller.

 

So it does cap at 3200 eh, the math was indeed showing me that the numbers were likely pulled from somewhere else, I guess where this glitch gets dicey for me isn't so much that it's a "Hail Mary" or a code-based approach, but that you're far, far exceeding the average value of killing 1 virus.  That's where the glitch shouldn't be allowed, but I also think since the rules didn't disallow it that the ruling shouldn't really change anything for your specific score this time.

With a high virus average like that, it puts the onus on anyone not doing the glitch to then play level 24 forever, which has proven this week to be a very high skill bar cause only bert could do it.  Logically and hypothetically speaking, if 8 of us pulled off the glitch on level 16 and 17, that would have meant that anyone else who wants to get in the top 8 either has to follow suit OR reach level 24 and play bert-style.

I see now that you mapped your own route just using TAS as your tool, (and supporting it with your fixed pills start, right?) and that's not really the problem in my opinion even if you were pausing to assign pills in your planned order cause you still gotta plan that, pull that off and it could be flubbed.  Funny cause I wanted to do a similar thing except where I just wanted to focus on some of the more common shapes that emerge that could reliably get me pulling big combos throughout the whole time - but it proved too daunting so I just winged it - and so I know it's hard to bring that approach.

The problem, I don't think, is looking behind the curtain at the code.  (Cause that's like saying don't learn things about the game, just play, which is limiting in its own way.)  You're looking at the inputs first to understand the game, whereas the classic way is to watch how things output and then backtracking your player inputs from there.   BUT where it gets fishy is going back to the simple fact that it's a glitch granting a gargantuan amount of points that far exceeds the virus average value in a game where nothing but viruses score points.   This isn't like a World Games scenario where outside of Caber Toss you mostly were just determining the maximums.  Here it's blown wide open, and the irony is that bert's score probably has one of the *lowest* virus averages, since he would have spent much of his time clearing level 24 paths over and over.

I guess where I'm at, and I'm not trying to be offensive here, is thinking going forward if you find something like this that breaks the ceiling on an average value you should maybe be posting it earlier and clearing it with the moderators.  Basically just more transparency would be nice even though there's potential the mods might ban your approach?  There's not really a way to enforce this except just asking you straight-up.  Comparing to World Games again you were transparent to the point that you made us ALL play better AND you still won.   As I alluded to in an earlier post we're not gonna necessarily know in what way these games have broken code, and it's a bit of a sad pill (ha!) to swallow when it gets dumped on late in the contest's week like this.  (and again this only really applies when you're breaking the point ceiling).

I've been wondering Deafcode are you into N64 at all?  I'd love to see your deconstruction of that coding, because I find it very mysterious.  Mario Party 3 code deconstruction would be a project I'd be highly, highly, highly, highly, highly interested to see

 

[0xDEAFC0DE | 1,277]

7 minutes ago, bertsampson said:

The "value of P2's controller" would be my only exception.  Do you have to have a button pressed on the second player's controller to get the glitch to work?

It depends. The glitch does usually* work without it in which case the endpoint is set to the top left corner, but this means you can only clear viruses to the right and below the last virus before the big triggers. However, if you want to clear viruses above that virus, you can press certain buttons to change the endpoint to right before the start point (and therefore wraps around to get the viruses above you). For my route, level 16 required me to press B on P2, and level 17 worked without pressing anything.

*there are certain cases where you must use P2's input, but those cases aren't as useful for setting up a good combo

5 minutes ago, Krunch said:

I guess where I'm at, and I'm not trying to be offensive here, is thinking going forward if you find something like this that breaks the ceiling on an average value you should maybe be posting it earlier and clearing it with the moderators.  Basically just more transparency would be nice even though there's potential the mods might ban your approach?  There's not really a way to enforce this except just asking you straight-up.  Comparing to World Games again you were transparent to the point that you made us ALL play better AND you still won.   As I alluded to in an earlier post we're not gonna necessarily know in what way these games have broken code, and it's a bit of a sad pill (ha!) to swallow when it gets dumped on late in the contest's week like this.  (and again this only really applies when you're breaking the point ceiling).

I've been wondering Deafcode are you into N64 at all?  I'd love to see your deconstruction of that coding, because I find it very mysterious.  Mario Party 3 code deconstruction would be a project I'd be highly, highly, highly, highly, highly interested to see

Yeah, that's definitely fair. The reason why I didn't post about it until Sunday was because I usually wait until I get a score with it and in this case that wasn't until Sunday (whereas for World Games I was posting every time I broke an new category). But I agree in retrospect, that's not fair for the moderators who don't have time to ban it if they deem it unfair. I'll definitely post about questionable stuff as I find it for future weeks.

Also that reminds me, I never wrote up what I found for world games. I'll try to do that tonight.

I've not really explored N64 much. Gamecube was my first non-handheld console (only had a GBC for awhile), and I haven't got around to playing through it's library. As for code deconstruction, these weekly events I've done it are fun, because I don't have to go too deep. I've tried doing a full deconstruction on NES Tetris awhile back but only got a third of the way through before I got bored and stopped. I was going to pick it up again, but last time I looked, someone else had already done it in the meantime. From the speedrunning stuff I occasionally follow, I know N64 games (like OOT) are much more complicated than NES. So, a full deconstruction is too much work. But I could be persuaded to take a look at something specific (at least as long as the tools don't suck).

[Krunch | 303]

1 hour ago, 0xDEAFC0DE said:

It depends. The glitch does usually* work without it in which case the endpoint is set to the top left corner, but this means you can only clear viruses to the right and below the last virus before the big triggers. However, if you want to clear viruses above that virus, you can press certain buttons to change the endpoint to right before the start point (and therefore wraps around to get the viruses above you). For my route, level 16 required me to press B on P2, and level 17 worked without pressing anything.

*there are certain cases where you must use P2's input, but those cases aren't as useful for setting up a good combo

Yeah, that's definitely fair. The reason why I didn't post about it until Sunday was because I usually wait until I get a score with it and in this case that wasn't until Sunday (whereas for World Games I was posting every time I broke an new category). But I agree in retrospect, that's not fair for the moderators who don't have time to ban it if they deem it unfair. I'll definitely post about questionable stuff as I find it for future weeks.

Also that reminds me, I never wrote up what I found for world games. I'll try to do that tonight.

I've not really explored N64 much. Gamecube was my first non-handheld console (only had a GBC for awhile), and I haven't got around to playing through it's library. As for code deconstruction, these weekly events I've done it are fun, because I don't have to go too deep. I've tried doing a full deconstruction on NES Tetris awhile back but only got a third of the way through before I got bored and stopped. I was going to pick it up again, but last time I looked, someone else had already done it in the meantime. From the speedrunning stuff I occasionally follow, I know N64 games (like OOT) are much more complicated than NES. So, a full deconstruction is too much work. But I could be persuaded to take a look at something specific (at least as long as the tools don't suck).

 

Yeah I had read it originally as being the case that you didn’t achieve the run til Sunday, which is legit.  At this point I guess the mods can look at all the info.  As for me I am quite interested in collaborating on deconstructing Mario Party 3 and if it were possible would be willing to offer up games to pay for it.  I got no technical experience but I know everything about what value assessments and gambits are important to the game and to what degrees.  I got a good reason for wanting to know this I swear.   I’m not even really interested in deconstructing every aspect of the game, just a few things (example: Duel mode doesn’t matter to me, only interested in the Party Mode).  But we can take that up in PM sometime.

Edited July 13, 2021 by Krunch

[Dr. Morbis | 2,088]

13 hours ago, docile tapeworm said:

@Dr. Morbis @Krunch i think one constant rule for the contest should be you put the cartridge in the nes, power/reset, og controller, thats it.

That's the only way I play!  The only downside is that most people don't have a full NES set so it's kinda hard to get people involved if you have to own every game.  I'm missing seven NES games and a third of the Famicom library myself, so if they had put Power Blade 2, for example, in this years compo, I would have been sitting that week out...

My issue with this weeks code-breaking is not the exploring of the code, but the fact that the resulting score is a literal glitch that gives you a massive amount of magical points (and to a lesser degree, using controller 2 in a 1p game to alter a glitch).

Moving forward, part of the preamble that is copy-pasted to the first post of every weekly competition should be amended to include something to the effect of: "If a glitch is found during gameplay that significantly enhances your performance in the weekly contest, it must be brought to the attention of the staff immediately for a ruling."  Ba-da boom, ba-da bing!

[docile tapeworm | 4,198]

i really dont mind emulation here though

ive used my everdrive when i dont have a game. i try to make sure those weeks are participation points.

my issue is i dont know if i really understand looking at code as an advantage. i feel like that takes a special tool or program? and is it like you have a map of the game, point values, hidden items? i dont like the idea of it if that is the case.

save states i think have to be fair game because its a week and you will be challenging someone who has played that game thier whole life so...

idk

i only mean if your using emulation play it like you would with og simple setup. i dont understand why you would want to look at the code? 

remember saying "the computer cheats!"?

ya well look at us now...

Edited July 14, 2021 by docile tapeworm

[guitarzombie | 924]

2 hours ago, Dr. Morbis said:

That's the only way I play!  The only downside is that most people don't have a full NES set so it's kinda hard to get people involved if you have to own every game.  I'm missing seven NES games and a third of the Famicom library myself, so if they had put Power Blade 2, for example, in this years compo, I would have been sitting that week out...

Funny, im the opposite.  I don't have the cheap stuff, so I can't play this.  But Zombie Nation?  No problem!

[BeaIank | 1,618]

For this time, we will accept 0xDEAFC0DE's score, but from now on using coding bugs to increase one's score is explicit forbidden.
Diving into the code to figure how things work is acceptable. Exploiting a poorly coded function will no longer be acceptable.

[BeaIank | 1,618]

Final scoreboard posted.

[mbd39 | 2,020]

That's an impressive top 8, wow. Congrats to those who got over a million.

[Richardhead | 2,813]

Wow!

 

Edited July 18, 2021 by Richardhead

[0xDEAFC0DE | 1,277]

7 hours ago, BeaIank said:

For this time, we will accept 0xDEAFC0DE's score, but from now on using coding bugs to increase one's score is explicit forbidden.
Diving into the code to figure how things work is acceptable. Exploiting a poorly coded function will no longer be acceptable.

Gotcha. And I'll do a better job by bringing up ambiguous things earlier as I find them (i.e. the caber toss thing was not technically a bug just bad game design)

[BeaIank | 1,618]

8 minutes ago, 0xDEAFC0DE said:

Gotcha. And I'll do a better job by bringing up ambiguous things earlier as I find them (i.e. the caber toss thing was not technically a bug just bad game design)

I know poorly coded shit when I see it and that was some of the nastiest poorly coded shit out there.