It is no secret that classifying puzzles can be a real nightmare, especially as we have cases
where the properties of different puzzles can intersect. So what should we do?

Creating large lists is not really the way to go. We need a set of properties which can really
describe the nature of a puzzle. The property list is flexible in its own right, in the sense that
new properties can easily be added, while addressing previous properties.

So how is this possible? Well, first we need to name those properties of a puzzle:



1. Movement type:
TW: Twisty (3x3x3)
SH: Shifting (Peter's Black hole)
RO: Rolling (Hungarian Rings)
GR: Gravity (Cubedron)
FO: Folding (Magics)

2. Skill type:
MA: Mathematical (3x3x3)
DE: Dexterity (Rubik's 360)
GE: Geometrical (most Hanayama Puzzles)

3. Solving Type:
SE: Sequential.
CO: Connect.
DI: Disconnect.


Note that, a twisty puzzle can or cannot be a dexterity puzzle.
Also, one puzzle can have more than one type of movement,
skill, or solving types. Therefore, this type of classification is
flexible enough, and helps in our best interest. Someone may
also add as a property, the materials a puzzle is made (but in
my opinion, this is not necessary).

For example some Hanayama Cast puzzles which are required to be
connected and disconnected, but are also sequential, may use
all three "solving type tags" tags SE, CO, and DI.

The Cast Radix should be: (TW) (GE) (SE+CO+DI)
Another example is Rubik's 360: (TW+GR) (DE) (SE)

As stated, this list is easily adjustable to cover and add many more puzzles.
In any case, I plan to use this classification for myself and all my puzzles,
because the next time I plan to use a nice database which will use all those
flexible properties.

Please let me know of your feedback.


Pantazis


PS. Shifting (SH) is about having a hole or a missing piece(s) so that other pieces can move in.
PS2. Take Apart, put together, (dis)entanglement, interlocking, are all covered by using the
CO and DI tags (together with the other tags).
PS3. I do not regard "Impossible Objects" as puzzles, because simply, they are not!
PS4. This is my first attempt to do such a classification, so be easy with me or I will get
temperamental again! (grrrr....).

Looks good, I would change shifting to sliding.

Nice way to keep things broad, allowing for multiple classifications. Perhaps I would add another category to the "solving type" because I recall reading that the 360 is not a sequential movement puzzle (one does not have to initiate a sequence of movements to solve it). It makes sense somehow, but then again I can't think of any other way to classify it... Other than that, though, looks fine.

I'm not sure there is any benefit to describing the Hungarian Rings style puzzles as "rolling". The fact that the pieces are spheres might be a mechanical convenience, but isn't really anything that defines the puzzle.

Most of Doug Engle's puzzles would perhaps fit in this category, but there is no rolling involved.

When I organize my puzzles I have a "marble" group but that is a visual concept, not so much a proper classification.

The Zauberkreuz, UriBlock, Tsukuda's Square and Elementals all fall into a similar category, but with no rolling and even no rotation. This to me is "sliding", but they have no gap.

Perhaps we can repurpose the term "sliding" to not involve the concept of a gap, and instead shove those puzzles into a category of "moving hole"?

Dave

To clarify, the 360 needs to be solved by using a number of steps (which is the definition of "sequential").
However, those steps are not based on mathematics, but dexterity. Hence the notation I gave.







True, the naming needs some better "meaning calibration". Take two:

1. Movement type:
TW: Twisty (3x3x3)
MH: Moving Hole(s) (Peter's Black hole)
SL: Sliding (Hungarian Rings)
GR: Gravity (Cubedron)
FO: Folding (Magics)

Any other objections?

To continue the clarification, the Cubedron includes both GR+MH.
This way, we can include different types of movement.

The next step (after ensuring this notation is strong and acceptable enough),
is to include more accurate definitions, e.g. the "fraction method" I proposed
for the twisty puzzles some time ago. Then we could do something similar for
the rest of the moving types.

For instance the moving hole(s) and sliding ones, can use the number of pieces of all
closed loops which are (or can be projected) on the same plane with respect to the
pieces being shared. In the case of MH, the number of holes (usually one) should also be
mentioned (The above definition is a bit vague, but I have already come up with a structure).

We are still a long way to a complete classification, but I feel that the "Sharing Property" +
"Fraction Method", can lead to something nice.




Pantazis


PS. The "fraction method" always gives different puzzles for when we have different numbers.
For if we had the assumption that same numbers exist for two different puzzles, or different
numbers exist for the same puzzle, there would be some serious contradictions. That said,
I have also thought of a nice solution for bandaged puzzles etc.

How about a gears category/ I'm sure Oskar and Bram would feel left out

Then it should be a TW+SL case. Easily covered.

But I am sure there may be cases that I cannot think of right now, that should also be covered.
Even so, adding new properties is no problem.




Pantazis

So, essentially what you're doing here is eliminating the single term concept of classification, right? In other words, you're taking the broad terms that we all use (I use Mechanical Puzzle) and breaking them down into sub categories?

That's a good idea. This should, somewhat, put an end to the arguments over which group a puzzle belongs. I'm sure there will still be discussions about how many different sub categories a single puzzle belongs to, but that would be a trivial argument compared to the current broad terms people use.

- Billy

This is exactly what the goal is. An efficient unification of all existing terminologies.




Pantazis

I disagree. The puzzle is to figure out how it's made. Even though no movement is required. Perhaps they could just be given the "IO" description. I have a few ideas for making some of these, but haven't gotten off my butt to actually make them.

-d

Where would secret puzzle boxes fit? I'm not talking about the Japanese puzzle boxes with external sliding panels (these would probably be geometrical) but the boxes with hidden internal mechanisms which you must try to manipulate if you can even figure out what the mechanism is. They may or may not have any external moving parts. Some examples would be the Isis puzzles, the Revomazes, and Carta Blanca (part of the Sacred Myths and Legends puzzle collection).

Fair enough, I can see your point now. In fact, this discussion is made to clarify such points.
And the "IO" tag is surely on.







Well, for the hidden ones, no notation can do much...
Unless the designer wants (if still alive LOL) to provide some details!

But we can always use some of the more generlalised ones which
have been stated, but this time combined with a hidden tag "HI".
It *is* a flexible notation, isn't it?




Pantazis

IMO, impossible objects are a sort of mental exercise, like a riddle, so they should be classified in the same way.

How about pencil puzzles, like crosswords, sudoku, slitherlink, etc.? Some are logical, and some are vocabulary-based.

Paper puzzles, can express myriad more concepts. But at this stage, it will be ideal to only focus on those
mechanical objects. It is true though that many of those paper puzzles would resemble some absolutely
brilliant mechanical puzzles (that is, *if* they are converted to mechanical puzzles). This is why I repeat
all the time that we should help our mind to break free and come up with new puzzle types and concepts.




Pantazis

I do like this system as it's very flexible and broad in scope. In particular, I like the fact that it is combinatorial, as it allows any combination of three different sets of terms, based on movement, skill type and solving type.

The only problem I see is that it does not (yet) go deep enough to distinguish different puzzles within the same class. For example, 90% of the puzzles on this forum would go under TW+MA+SE.

How would you then classify all the different twisty puzzles according to shape, symmetry, mechanism, bandaging, fused, void, inverted, pillowing, sticker mods, etc., etc., etc., etc.? The possibilities seem endless!

We really need to take one step at a time. As I mentioned above:





Hope that clarifies the question.




Pantazis

I have a feeling there's alot of memorization to come to remember all these...

Why? Most definitions are super easy to remember (much easier than a chemistry chart LOL),
while the "fraction method" directly defines a puzzle.

Remembering is what the current system requires (which has endless names for a non-expert).
The system I propose will require a slight reconstruction of the numbers.

It is like trying to remember all the formulas in mathematics. It is impossible, so I don't. I prefer
to reconstruct them even if that takes a little bit more time. And for those who don't want to remember
or reconstruct, they should be able to get some sort of "Atlas" describing existing or even not-yet existing puzzles!

Or I can put it in another way. In biology and medicine, they have special structured names (in Greek and/or Latin)
some of which look quite long and difficult. But once you know how the structure works, you can
immediately pinpoint the structure behind the name. Works extremely efficient, trust me on that!




Pantazis