Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

"Super Mario World vs. the Many-Worlds Interpretation of Quantum Physics"

4 views
Skip to first unread message

BS

unread,
Feb 4, 2008, 6:12:31 PM2/4/08
to
http://msm.grumpybumpers.com/?p=20

First watch this video

http://www.youtube.com/watch?v=T2OytHzZ72Y

Okay, now what was that?

So a few months back some of my friends were passing around these
videos of something called “Kazio Mario World“, which I was told, at
first, translated to “Asshole Mario World”. This turned out to have
actually been a misunderstanding of something in the youtube posting
of the original creator’s videos:

[Asshole Mario] is not the real name for this series of videos, but it
is my personal name for it.
The literal translated name for 自作の改造マリオ(スーパーマリオワールド)を友人にプレイさせる is
“Making my friend play through my own Mario(Super Mario World) hack”,
hence Kaizo(hack) Mario to the USA.

…but, the name is pretty appropriate. Kazio Mario World is one of a
series of rom hacks people create in special level editors that let
you take Super Mario World and rearrange all the blocks; the point of
Kazio appears to have been to create the most evil Super Mario World
hack ever.

I started watching these videos, but after seeing how the player got
past the first gap stopped, went wait, this actually doesn’t look so
bad, and started playing it instead. It’s actually not that bad! I was
expecting it to be like Super Mario Frustration, Kazio Mario World’s
equivalent in Super Mario Bros. 1 hacks– all ridiculous jumps that
require pixel-perfect timing, memorizing the location of a bunch of
hidden blocks that exist only to foil a jump and, occasionally,
actually exploiting glitches in the game engine.

Kazio Mario World actually turns out though really to be kind of more
like a puzzle game– giving you a series of seemingly impossible
situations and then leaving you to figure out how to get past them. It
only uses the sadistic-invisible-block trick sparingly (and, hey, even
SMB2JP did that a couple times). And it actually turns out to be kind
of fun.

It’s still sadistically hard, though, so if you want to play it you
have to use what are called “save states”. Most emulators let you do
this kind of save-and-rewind thing, where if you screw up you can back
up just a few seconds to the last time you were standing in a safe
place. So if you’re playing Kazio Mario world you find yourself
playing the same four-second section over and over and over until you
get the jump just right, listening to the same two seconds of the
soundtrack looping Steve Reich style…

Anyway, the idea for the video up top was inspired by an offhanded
comment in the “original” Kazio Mario World youtube post I linked
above:

The original videos were in god awful codecs that were a bitch to
convert, so unfortunately the Tool Assisted Speedruns came first to
most youtube watchers.
This is rather unfortunate, as I feel you lose a lot of the “appeal”
by watching those.

This refers to the way that most emulators, if you are recording a
video of yourself playing a game and you do the save-state rewind
thing, they’ll rewind the video too, such that the video shows only
your final attempt, not any of your messups. People use this to make
“speedruns” showing a best-of-all-possible-worlds recording of them
playing through some game or other, with all the errors scrubbed out.
The guy’s point was that watching Kazio Mario World this way kind of
ruins it, since most of what makes Kazio great is watching someone
fail over and over and over again until they finally get it right.

On the other hand, Kazio Mario World involves SO much failing that
this means all the “real” videos are, like, twenty minutes long just
to get through what in a tool-assisted run would have been a two-
minute level. So I was thinking, what if you had a special tool that
instead of erasing all the screwups, it saved all of them and made a
video of all the screwups plus the one successful path superimposed? I
kept thinking about this and eventually I just sat down and hacked
SNES9X to work exactly like that. The result was the video up top,
showing the 134 attempts it took me to successfully get through level
1 of Kazio Mario World.

I think I’m going to make some more videos in this style of different
Kazio Mario World levels and post them back here, but in the
meanwhile, if you want to make your own many-worlds speedrun videos,
here’s my custom version of SNES9X 1.43 with the multi-record
function:

For the Mac OS X version, click here.

For a Windows version, click here. (Many thanks to Syndalis of
360Arcadians for compiling this for me.)
If you want a Linux version, you’ll have to compile that yourself, but
you can do this by finding a copy of the 1.43 source and replacing
movie.cpp with this.
And for the full Mac OS X source, click here.
Note that this is a quickly-tossed-together hack all done to make a
single video, and I make NO promises as to the quality, ease-of-use,
correctness, or safety of these links. Also, I think the video feature
should work with any SNES game, but I’ve only tested it with Kazio. If
anyone attempts to try this yourself, I’d be curious to hear about
your results.

To make a video: First, use SNES9X’s “record movie” function to record
yourself playing some game; while the game is running, use the save
and restore feature at least once. When you’re done, you’ll find that
SNES9X has created a yournamehere.smv file and also a series of files
with names like yournamehere.smv.1, yournamehere.smv.2, etc.
These .number files are all the different “mistake” playthroughs, so
keep all these files together in one directory.

To turn this into an actual movie you can watch, you will need to use
the OS X version of the emulator. Unfortunately, the Windows and Linux
versions can only record multiple-run SMVs– they can’t do the export-
to-quicktime thing. The quicktime-export code is based on alterations
to the mac-specific parts of 1.43 (although considering that I hear
the Quicktime API is mostly identical between Mac and Windows, it
might be pretty easy to port that code to Windows at least…).

Anyway, in the OS X version, open up the appropriate ROM and choose
“Export to Quicktime Movie” from the Option menu. Before leaving the
export dialogue, make sure to click the “Compression…” button. You
*MUST* choose either the “None” or “Planar RGB” codecs, and under the
“Compressor” pane you *MUST* choose a depth of “Millions of Colors+”.
The “+” is important. Once you’ve saved the movie location, go to
“Play Movie” in the Option menu and choose the .smv you want to play.
The emulator will play through each of the playbacks one by one; when
it’s done (you’ll know because the background turns back on) your
movie will appear in the location you chose. Note that there’s one
more step! You won’t be able to actually play this movie, at least not
very well, because the export feature works by creating a different
movie track for each playthrough and the file will be huge and
bloated. Open your video in Quicktime Player, then choose “export” and
export to some video codec with actual compression (like H.264). This
will flatten all the different layers of the movie into one. Okay, NOW
you’re done.

…So what’s this about quantum physics? Oh, right. Well, I kind of
identify the branching-paths effect in the video with the Everett-
Wheeler “Many Worlds Interpretation” of quantum physics. Quantum
physics does this weird thing where instead of things being in one
knowable place or one knowable state, something that is quantum (like,
say, an electron) exists in sort of this cloud of potentials, where
there’s this mathematical object called a wavefunction that describes
the probabilities of the places the electron might be at a given
moment. Quantum physics is really all about the way this wavefunction
behaves. There’s this thing that happens though where when a quantum
thing interacts with something else, the wavefunction “collapses” to a
single state vector and the (say) electron suddenly goes from being
this potential cloud to being one single thing in a single place, with
that one single thing randomly selected from the different
probabilities in the wavefunction. Then the wavefunction takes back
over and the cloud of potentials starts spreading out again from that
randomly selected point.

A lot of scientists really don’t like this “collapse” thing, because
they’re uncomfortable with the idea of nature doing something “at
random”. Physics was used to dealing with randomness before quantum
physics came along– the physics of gases are all about the statistics
of randomly moving gas particles, for example– but those kinds of
randomness aren’t assumed to be actually random, just “effectively
random” because the interactions of air molecules are so chaotic and
complicated that they’re too unpredictable for humans to track. Think
about what happens when you roll a die: the number that comes up when
the die lands isn’t strictly speaking “random”, it’s absolutely
determined by the physics of motion and the velocity at which you let
go of the die and so forth. The “randomness” of a die roll isn’t about
actual indeterminacy, but rather just a way of talking about your
ignorance of how the deterministic processes that control the die
operate. Quantum physics, on the other hand, has things that as far as
anyone can tell are really, objectively random, with no mechanism
producing that randomness and nowhere apparent to stick one.

Since this makes some physicists uncomfortable, they came up with a
sort of a philosophical trick: they interpret quantum physics in such
a way that they say when there’s more than one possible random outcome
of some quantum process, then the different possibilities all happen,
in alternate universes. They can’t prove or disprove that this idea is
true– from the perspective of someone inside one of these universes,
everything behaves exactly the same as if the “wavefunction collapse”
really was just picking a random option. But it’s one way of looking
at the equations of quantum mechanics, and as far as the mathematics
cares it’s as valid as any other. Looking at things this way, if
there’s a 3/4 chance of a quantum process doing one thing and a 1/4
chance of it doing the other, then we get three universes where the
one thing happens and one universe where the other one does. This does
mean that there’s some universe where two seconds ago all of the atoms
in your heart spontaneously decided to quantum-tunnel two feet to the
left, but in almost every universe this doesn’t happen so we don’t
worry about that.

Science fiction authors love this. There’s a bunch of stories out
there exploring this idea of a multiverse of infinite possibilities
all occurring side by side (the best of these I’ve ever read being
Robert Anton Wilson’s Schrödinger’s Cat). Most of these stories get
things totally wrong. Science fiction authors like to look at many-
worlds like, this morning you could either take the bus to work or
walk, so the universe splits in two and there’s one universe where you
decided to walk and one universe where you decided to take the bus.
This is great for purposes of telling a story, but it doesn’t really
work like that. The many-worlds interpretation is all about the
behavior of quantum things– like, when does this atom decay, or what
angle is this photon emitted at. Whereas human brains are big wet
sloppy macroscopic things whose behavior is mostly governed by lots of
non-quantum processes like neurotransmitters releasing chemicals.

This said, tiny quantum events can create ripples that have big
effects on non-quantum systems. One good example of this is the
Quantum Suicide “experiment” that some proponents of the Many-Worlds
Interpretation claim (I think jokingly) could actually be used to test
the MWI. The way it works is, you basically run the Schrödinger’s Cat
thought experiment on yourself– you set up an apparatus whereby an
atom has a 50% chance of decaying each second, and there’s a detector
which waits for the atom to decay. When the detector goes off, it
triggers a gun, which shoots you in the head and kills you. So all you
have to do is set up this experiment, and sit in front of it for
awhile. If after sixty seconds you find you are still alive, then the
many-worlds interpretation is true, because there is only about a one
in 1018 chance of surviving in front of the Quantum Suicide machine
for a full minute, so the only plausible explanation for your survival
is that the MWI is true and you just happen to be the one universe
where the atom’s 50% chance of decay turned up “no” sixty times in a
row. Now, given, in order to do this, you had to create about 1018
universes where the Quantum Suicide machine did kill you, or copies of
you, and your one surviving consciousness doesn’t have any way of
telling the people in the other 1018 universes that you survived and
MWI is true. This is, of course, roughly as silly as the thing about
there being a universe where all the atoms in your heart randomly
decided to tunnel out of your body.

But, we can kind of think of the multi-playthrough Kazio Mario World
video as a silly, sci-fi style demonstration of the Quantum Suicide
experiment. At each moment of the playthrough there’s a lot of
different things Mario could have done, and almost all of them lead to
horrible death. The anthropic principle, in the form of the emulator’s
save/restore feature, postselects for the possibilities where Mario
actually survives and ensures that although a lot of possible paths
have to get discarded, the camera remains fixed on the one path where
after one minute and fifty-six seconds some observer still exists.

Note: Please do not use the comments section of this post to discuss
ROMs or where to get them. IPSes are okay. Thanks.

tcbev...@yahoo.com

unread,
Feb 4, 2008, 6:56:52 PM2/4/08
to
Brilliantly done! I, for one, (sometimes) think quantum suicide is
true, and also that it explains away the grandfather paradox. You go
back in a time machine, kill Gramps, and yet you do not vanish or
otherwise experience any paradox at all. Why? Same reason: you are not
the copy whose grandfather died! In your history, he didn't get
whacked before your father was conceived. You, in traveling back in
time and killing your grandfather, have hopped in a new timeline, and
there you are. A particle has as many possible past paths as future
ones, after all. Of course, one objection to quantum suicide is that
if it is true, you will survive many maiming near-misses, becoming
more and more miserable as your endless subjective immortal life
continues. This we do not seem to perceive: or do we? If we live an
infinitely long time, it is odd that we should be young now: I am not
yet fifty, according to my memories and evidence; yet also I live at a
time when immortality through future science is not a silly notion.
This points to a variant of Bostrom's doomsday conjecture, which
mathematically may place us close to the last generation of humans,
owing to the principle of mediocrity. http://en.wikipedia.org/wiki/Doomsday_argument
However, there are several competing interpretations, with results
that are all over the place, statistically.

I also have the notionof a sort of "puzzle-fit" principle, such that
your consciousness is a pattern in an information matrix of which many
worlds are also part; so there are a subset of worlds where a
conssciousness like yours will inevitably be found. See:
http://en.wikipedia.org/wiki/Boltzmann_brain

If you can never tell the difference, is there a difference?

Tom Buckner

0 new messages