Message #701

From: Melinda Green <melinda@superliminal.com>
Subject: Re: [MC4D] Re: 3^4 parity problems
Date: Thu, 15 Oct 2009 17:44:22 -0700

To answer Klaus’ question, it currently *is* possible with the current
puzzle to supply a scrambled log file that does not include the scramble
history. I’m not sure that makes much difference however because I
suspect it wouldn’t be too hard for someone to tell whether a solution
essentially backed out the scrambling, and it certainly wouldn’t
resemble any sort of human solution. There are definitely interesting
questions regarding which methods should be considered fair and how to
disallow or detect cheating. I already caught one attempt, so these are
not an abstract questions.

Early on we had to drill deeply into the question of macro use before
deciding that they are fair to use but that solutions using them just
shouldn’t be compared with ones that didn’t. Even still there are
unanswered questions such as whether they should be allowed during speed
solving competitions and if so, should they have to be created during
the actual timed runs or whether use of previously created macro files
should be allowed.

For the "shortest" category, my feeling is that just about anything
short of backing out the scramble should be allowed. If you can write or
find software to help, then power to you though I hope that you’d
declare any aids you used. OTOH, there’s still this issue of searching
for random scrambles that let you side-step parity problems or even
whole steps. As an extreme and completely impractical thought
experiment, imagine writing a program that can examine an ungodly number
of full scrambles in search of one that just so happens to be one twist
away from being solved. If you then solve it, can you really claim to
have performed the shortest solution to a full scramble with a single
twist? And even if we disallowed such cherry picking, and assuming we
could enforce that rule, then would it fair even then since some people
will simply get lucky sometimes. I certainly wouldn’t want to tell
someone that they have to finish each full solve before starting another
one! OTOH (I get at least 3 hands in 4D, right?) this is the idea behind
averaging several solves commonly used in speed solving.

For myself, I don’t worry about people cherry picking very much in most
cases largely because most puzzles are really 3 puzzles in one (2C + 3C

• 4C), and so cherry picking is probably only going to be able to help
  you with whichever element you like to solve first. It could however
  help you to *decide* which element to start with since problems with one
  element can be much harder than problems in another, but overall it
  feels fair enough. My main concern is only with the 2^4 which has
  nothing but corner pieces. These puzzles have always bothered me because
  they’re oddballs in other ways too. In retrospect I suppose this was a
  rather lengthy way of saying that I have no opinion. Sorry about that! :-)

To make it up to all of you who have read this far, I’ll let you in on a
bit of very juicy news: Roice and I have been hard at work generating a
new version of MC4D using a new puzzle generation engine from Don. The
new version will allow you to solve a whole bunch of beautiful new 4D
shapes other than the cube. It is also much improved in many other small
ways including sound effects, interactive arbitrary 4D rotations and a
much needed GUI facelift. Best of all, you get to be the first people to
try it out when we create our first beta testing version soon. So watch
this space for news of MC4D 4.0!

Happy puzzling!
-Melinda

matthewsheerin wrote:
>
>
> […]
>
> I think you may have point about providing a scramble which cannot be
> reverse engineered. I agree, and I can’t think of a way to police
> against the trial and error approach with different scrambles either.
>
> I suppose looking up algorithms for smaller steps would be acceptable,
> since methods for 3D cubes rely on learning algorithms too, which are
> generally found on the internet these days.
>
> I second the request for an upper (and lower) bound for 4D, though I
> will stop short of asking for a God’s number, since that hasn’t been
> found for the 3x3x3 yet!
>
> happy hypercubing
> Matthew
>
> — In 4D_Cubing@yahoogroups.com <mailto:4D_Cubing%40yahoogroups.com>,
> "Klaus" <klaus.weidinger@…> wrote:
> >
> > Hi everyone,
> >
> > I also thought about this but for my system this doesn’t really make
> sense because it just takes to long to get to a position from where
> you can decide if this problem/parity occurs (Well it was only 50
> turns but to optimize them took me about 3 days). I will however try
> to find a way to predict it earlier and to work around this awkward
> situation.
> >
> > But even if you decided that trial-and-error is unfair, I can’t come
> up with a way how to deal with that topic. Is it even possible [with
> the current programme] to supply a cube scrambled by hand without the
> possibility that someone can derive the fewest-move solution from the
> log-file?
> >
> > @ matthewsheerin: I have to solve some 2^3 cubes in my solution,
> too, and I’m using the Guimond method (if there is any faster method,
> please tell me). I tried to find some PLL algorithms with the
> computer, but I don’t think this is cheating, because if you look them
> up on the internet where some other people have found them by
> computer, or if you do the work yourself makes no difference. If you,
> however, compute a fewest move solution for the whole 2^3 or 3^3, I
> would call this cheating.
> >
> > btw: has anyone ever made an attempt to prove an upper bound for
> fewest move solutions on the 3^4?
> >
> > Have a nice twist,
> > Klaus
> >
>
>