SDA logo

Marble Madness was originally released as an arcade game in 1984 and was ported to NES in February 1989. Take control of a marble in a set of six maze-like levels. Traps abound to delay your progress, including other marbles, acid pools, hammers, and vacuums.


Best time: 0:02:50 by Steve 'Elipsis' Barrios on 2014-03-21.

Get Flash to see this player.

Author's comments:

SDA speedrun History:

July 31 2005, 3:13, Elliott Feiertag
October 27 2008, 2:54.2, Andrew Gardikis
March 18 2014, 2:52.1, Steve Barrios
March 21 2014, 2:50.9, Steve Barrios

This speedrun of Marble Madness times out to 2:50.9 with 0 deaths. It was done live on and retimed afterwards. I really didn't have a high expectation of successfully shaving any more time from my previous record anytime soon, but I am much more pleased with how this run looks compared to its predecessor.

My previous WR of 2:52.1 was rightly criticized for imperfections on the first three races, and managed to take the record only because of the Aerial and Silly race improvements. This new run is faster than AndrewG's prior world record on four out of the six races (missing a fifth by 6 frames), rather than merely those two, and still manages to utilize the time saves of my 2:52.1 run.

Level commentary:

Practice Race:

If you do this level right, you're playing the last half of it blind. I always shoot to finish this level with 55 seconds remaining on the in-game timer, which translates to 7.0s in real time. Every 0.1s saved beyond that becomes exponentially more rare and challenging to execute. In this run I managed to do it in 6.9s, surpassing both my and Andrew's previous runs by a narrow margin.

Believe it or not, almost half a second can theoretically be gained over a standard 55 by dumping the marble downward as soon as possible and playing nearly the _entire_ level blind. I have only done this perfectly once in over a thousand attempts.

Beginner Race:

I call this the level of three bridges: the draw bridge, the wavy bridge, and the weird bridge. Trying to be proactive, rather than reactive, in regards to the draw bridge movement is one of the very few ways to save time on this level. The draw bridge dictates that any amount of time saved during the first 1/3rd of the level goes into waiting more frames for it to go down and let you pass. After sitting here hundreds of times, I have developed a few methodologies to attempt to execute the optimum strategy which I call "frame perfect draw bridge". The best thing you can do for yourself is to be moving at maximum velocity onto the bridge at the same frame it lowers. I use the bottom green slinky as a visual cue for when to begin accelerating. I wasn't aggressive enough to get it frame perfect this time around, but was still accelerating towards it as it went to the down position.

The wavy bridge is the number one leading cause of death in the beginner race. Navigating it correctly requires very quick and precise directional movement coming out of the pipe. Optimally you want to stay as close as you can to the center without falling into the hole. Cutting the corners on the ramp between pipes is also another minor time save.

The weird bridge has bizarre physics, and I still have no idea what it's supposed to represent visually. What I do know, however, is that I want to hit it somewhere between the fourth and fifth tile to be ejected straight downwards towards the goal.

Even the TAS finishes with 48 seconds remaining on the timer, there is little more than a few tenths of a second to be gained here.

Intermediate Race:

The two biggest time saves in this race are relatively high risk for low reward, as they greatly reduce the amount of screen real-estate available to the player. The strategy, which I call a ramp-drop, requires jumping off the downhill ramps in order to take the most direct route to the goal. I usually go for the first ramp-drop after the initial high-wall section, and did so in this run. The acid is deterministic, and can be dealt with consistently as long as you have enough screen left to know where the marble is.

After the pipe, I took the final section of the race much more cleanly than last time, cutting no less than four corners close enough to dizzy the marble and getting well ahead of the first wave of the green rug.

Although this run contains a much faster intermediate race than its predecessor, theoretically there is even more time to be saved here. If you are a complete lunatic, you could go for the second ramp-drop and, assuming you also did the first, execute the final portion of the Intermediate race completely blind.

Aerial Race:

I am very proud of my time on this level, and somehow managed to save 0.1s over my previous run which I felt was already very strong. When I was looking at AndrewG's run, I identified the Aerial Race as one of two major places where I could improve the time. First of all, not getting wanded was a huge help... but after studying the run further I noticed that Andrew's WR had some significant hesitation getting through the hammers safely. My goal was to clear this level without any of that time loss. For the longest time, I used a strategy which I called "there's no such thing as hammers" to literally pretend that they aren't part of the level and proceed as quickly as possible. This turned out to be more or less a coin flip between a time save and a dead run. I later adapted the strategy to traverse the gauntlet as close as I could manage to the upper-right side of the path while still completely ignoring their cadence, which seemed to be more consistent way to do no-stop hammers . (Hats off to NESKamikaze at SRL for inspiring the adjustment by pointing out that they will _completely_ miss you if you manage to line it up on the very far edge of the corridor.)

The only time that I think could conceivably be saved here would be to take the second jump on a slight diagonal vector and (assuming the marble doesn't crack) use that momentum to then perform a ramp-drop off of the last downwards ramp. This isn't something you _go for_ so much as _barely survive_, but the very few times I have been able to recover it, it was slightly faster.

Silly Race:

This is the other race where I knew significant gains could be made on Andrew's record. By far the most significant gain in the Silly race comes from hitting The Big Silly Shortcut. After a ton of practice, I managed to get this shortcut from a 5% success rate to a roughly 75% success rate... although with varying degrees of efficiency.

If you do the beginning of this race well, you go up the pipe coming into the hardest corner of the game with very little screen real-estate to work with. Slowing down at all means that the level's physics play hell with the marble, and worse yet a screen-bounce is pretty much always fatal. At this point, it has to be all muscle memory. Fortunately, this time I executed the quickest pattern I know of without error.

There are two ways around the next rectangular hole in the floor. I have read a bit of debate about whether it is faster to proceed up-right (risking another screen bounce into the hole), or reverse course and go around the other way (likely a shorter physical distance, but fighting inertia). I feel confident that my route is faster because I found a small time-save built into it. I intentionally "fall" into the visually easy to miss hole after rounding the corner, and then "roll uphill" to increase the marble's speed beyond that which can be achieved on the flat surface. This time around a bird slowed me down by about half a second by being in the way of the optimum route, so I had to path around it.

Next it was time to confront the aforementioned Big Shortcut. The TAS does this shortcut without touching any other walls, which as far as I can tell is utter nonsense in a human run. My ideal strategy here is to bounce off the front wall, then the back wall, and take that horizontal speed with me to go diagonally along the third grid-line towards the goal. There are a number of caveats to be aware of here. 1. As near as I can tell, the grid-line positioning has to be pixel perfect. 2. Sometimes the first bounce actually sends you back out towards the birds for whatever reason. 3. If you overshoot the positioning even slightly, the physics of the level do not allowed you to head backwards.

While I slowed down a little more than I would have preferred in order to hit this shortcut, it still represents a huge time save over the standard route. Overall, this race still turns in faster than AndrewG's, but is slower than the one in my prior record. Over a second could be saved on this run with an even more aggressive use of the shortcut and better luck(?) with the birds.

Ultimate Race:

Although this race is the last one, I feel like it is about tied for third (with Intermediate race) as far as executional difficulty. That said, approaching it knowing that you are on world record pace can make it extremely jarring. This is the only race of the six that is perceptibly slower than AndrewG's WR thanks to the executional slip-up mentioned below.

Some time can be saved off of the very first jump by approaching it on an angle instead of head-on. (See Elliot's previous WR for the safe-strat and a comparison to how Andrew and I approach it.) After this, I take an unintentional screen-bounce on the bumpy surface section. This loss of about 0.7 seconds was unfortunate, but since that mistake usually kills the marble, I was happy just to have reacted quickly enough to save it. This is my biggest mistake in the run, and is still far less costly than some of the errors in the previous record.

You cannot change direction or speed whatsoever once on the blue surface, so it is important to hit it straight and fast. In this run, I was perilously close to bouncing off of the upper-left pyramid... which would have cost multiple seconds and buried the run.

There is no point in trying to hurry through the first few disappearing floor sections because your progress is limited by the rate at which the next tile appears. The optimal strategy is to hang back on the last straight and hit the very last tile on the same frame that it appears moving at full speed towards the goal. Had I been more aggressive about this, I might have saved .1 second, but like every previous record, I opted for a relatively safe finish.

Final time: 2:50.9

I'm much happier with this run than my previous entry, and am inclined to leave it alone for awhile this time. Andrew's amazing run stood for 6 years before finally being cracked. It was only through pursuing a number of very aggressive time-saving strategies, and thousands of attempts, that I was able to finally shave just 3.3 seconds off of it.

Is there room for improvement? Yes... but the margin of error is starting to get very thin, and this is going to be an extremely tough run to improve on without resorting to some truly insane risks. It think it will be inordinately challenging to put it all together between resets without succumbing to Marble Madness.

I would like to thank AndrewG for being supportive and encouraging me to pursue this record, NESKamikaze for the enjoyable Marble Madness races on SRL, TheThrillness for helping me figure out my capture setup as well as everyone on #sda who spent time listening to me stress out about interlacing, my mother for actually keeping all my NES / SNES stuff at her house for 20 years until the day I wanted it back, and SeamusOdrunky for the NES that actually still worked.

It's really exciting to be a part of the history of one of my favorite games and see my name up on the same website as some of the most talented gamers out there. I feel like I'm still just getting into speedrunning, and would love to have more of you drop by my stream sometime.

Return to the Game List, the FAQ, or the Home Page.