This partly happened in tango in the 90's, when Gustavo Naveira, Chicho Frumboli and Fabian Salas came together to create a structural and kinesiologica base for something that previously existed more as "an intuition".
For each basic configuration of leader and follower and their bodies they looked at all permutations.
Then they constrained it first to the subset of those that are possible to do at all, kinesiologically.
Then further to the subset of those that could be danced with reasonable comfort.
And finally the subset of those that are easy enough to be taught to students and would work on a crowded social dancefloor.
None of this was done with the help of computers.
The most systemic documentation of this is possibly Mauricio Castro's book "Tango -- the structure of the dance".
But lately a lot of new books were published on tango technique; I may be out of the loop.
A friend of mine who's also a tango professional is currently looking into the feasibility of doing a PhD thesis on this topic.
He wants to use ML to spit spit out the full motion tree of tango.
Both to be able to document it automatically, i.e. using generated 3D animations, as well as to discover new combinations the manual approach used by Naveira, Frumboli and Salas, over a quarter century ago, will have missed.
As a one-time avid Tango dancer, while I believe all the motions/steps can (and have been) be sketched out, named, and notated and put to paper, that which can be written down is necessarily insufficient to make it easy for people who understand the syntax but not the dance to replicate the dance easily. Let alone beginners. There's a great deal of unspoken communication in the embrace, in the way the partners' center of gravity is shifted about, in the music and song, in the timing, in their understanding of each other's abilities, and in their personal connection. Those details would have to be rediscovered by those reading the notation.
Some things are just hard to write down with fidelity. Think of tastes: we can have cookbooks, but its hard to reproduce the exact ingredients of grandma's cooking, or that one restaurant. Smells are even harder to express in words. Dance is kinda like that.
None of which is to say that one shouldn't do it, or try. Quite the contrary. It's a good challenge! Even if you could write it all down, leaving a bit of mystery/mystique would be a good idea, but I don't fear that mystique will be lost.
> As a one-time avid Tango dancer, while I believe all the motions/steps can (and have been) be sketched out, named, and notated and put to paper, [...]
This is obviously untrue. For simple math reasons.
If you believe that, I suggest reading an intro to combinatorics.
Or just doing the math in your head for a combination of two steps. Leader does forward step towards follower's forward cross.
Just in one system (cross or parallel) and considering how deep the step is (outside, sacada, deep sacada/behind/in front) and where the leader steps (behind free foot, next to outside, next to inside, in the middle between legs, next to to standing foot inside, next to outside, in front), we can reasonably say we have 21 possibilities. Now multiply by system and that's 42. Now multiply by combinations of systems on one side and that's 84.
This still makes a lot of simplifications but we're talking one step.
Furthermore, if we mirror, we can't just assume the same observations apply, as tango has an open and closed side, etc. etc.
The endeavor is also not about capturing the essence of a dance but about exploring a motion space; structurally and kinesiologically.
Tango, specifically, will elude documenting on so many levels, otherwise, it's not even worth talking about.
I have had the same idea about martial arts. Not sure it is possible. In dancing you have at least some symmetric patterns and rhythm, in martial arts you have even more stances, rotation around more axes and arrhythmic moves.
good dancing like good fighting isn‘t mechanical either, it is good because it relies on little imperfections. Just like MIDI files cannot replace a concert pianist.
In the end I think motion capture of extremities along with some easing of paths and compression of point-clouds (think bezier curves) might be more worthwhile than a notation.
> in martial arts you have even more stances, rotation around more axes and arrhythmic moves.
I wonder how true this is. Specifically I think you could simplify the moves into steps with "who hits who" states. You could normalise out the rhythm, following the "everything's in 4/4 if you don't count like a nerd" idea.
So you'd end up with just the key frames of either contacts/impact or change of the movement direction. Would martial arts really have even more variety here?
That's super helpful thank you for writing it all out! I bought Castro's book, looking forward to reading it more. If you have any other resources/recommendations like this I would love to see them. I had tried TANGOBOOK: Dance Notation of The Classic Tango by Hsieh, Felipe and found it to not be all that helpful to me as a beginner Tango dancer. I feel like the footwork diagrams left out a lot of where I was supposed to be pointing to with my chest, but I'm not very good at Tango, so it's hard for me to say that that actually matters.
I enjoy the attempt to analyze the dancing (really choreography). But I'm surprised that in this new world of LLM hype not one word in there was dedicated to what dance really is: a language.
Dance is a physical, kinesthetic language.
And to riff on some of the posters, it's this expression and communication between the dancers that makes dancing worth dancing. There's also the social communication between the group, cultural "dialects" and "accents" (on1 vs on2, stylings, etc.)
I began dancing when I realized there was an entire spectrum of the senses closed off to me from overthinking. Besides the specifics of "body language," good dancers can convey intense amounts of emotion. I met someone who could "talk" to me for hours and she never needed to say a word. While the poster attempts to represent dancing in terms of a symbolic language, I'd recommend anyone interested in dance to also try the opposite: realize the physicality is the language and medium of expression. It's more fun, and I promise you your style of thinking will enlarge and grow.
Agree. I think the approach in TFA is like trying to analyze, say, improv comedy via a context-free grammar for English. There might be something there, in the syntax, but it misses a huge amount of semantics and paralanguage.
Yeah. While I'm sympathetic to the attempt to formalize the representation, there's a psychology to this that a lot of people in tech seem to miss when I see them dancing.
The reasoning goes: "if I understand and represent this choreography properly, I will be better at the activity of dance." Dancing, for them, is the end result of understanding a representation and then doing it.
But the psychological shift is "if I understand my emotions and body properly, I will be able to use the representation of dance to convey that." While dance obviously needs a backing vocabulary of basics and moves, great dancers accidentally come up with better dance moves because their body mechanics combined with the dance's structure, musicality, and emotion constrain what they do. They—and I'm going to use this word intentionally—literally think and communicate through dancing. To use your phrasing, the "semantic" meaning of the dance is their emotions: the actual activity of dance is simply the representational outlet for that.
When you're with a good dancer, your brain shuts off but you're still stuck within the aforementioned set of constraints. And that's when the magic happens. That's when you actually start speaking. That's when your improvised move is the only way you could have possibly said what you wanted to say.
I didn't include it in this version of the post, but I do view dancing as a language and I think it's useful to do so. Every dance is a conversation between two people, with cues and signals following back and forth the whole time. I explicitly don't think about this choreographed dancing, I am primarily interested in analyzing improvised social dancing between two people. I've found a lot of stuff that I'll try to write about soon via way of metaphor by just applying the tools of linguistics and semiotics to social dancing. For example, the idea of a phonetic inventory composed of the ways that people move during the various social dances and styles makes a lot of sense. The small ways that people move their bodies in Cuban Salsa are different are different from LA Salsa, but they are more similar to each other than the way that people move when dancing Zouk or Tango. The tilted head in Zouk and the torsion in Tango feel like the rolled R of some Spanish dialects. Once I have something more concrete I'll try and post about it.
I don't know. This is ok if it's helpful, but reminds me of why I quit playing chess: the mystery was lost.
As I got more into chess and with the advancement of chess computers and championships, I felt there was little creativity left to have fun with the game.
So for dancing, it's literally meant to be "danced".
Of course you can figure out all the combinations and so on, but being overly analytical about it removes -- for me -- the entire reason to do it : to escape the intellectual world and do something kinesthetic.
I understand I am probably an outlier here on HN. :)
When I was a dance beginner I wanted to be more analytic about it. Surely one could write it all down and visualize and think about it with rigor! But, no. There are things that just have to be learned the hard way on the floor rather than by memorizing notation/images from a book. Dancing cannot be learned only from analysis. Professional dancers might be able to learn new dances quickly from notation, but let's not kid ourselves: professionals would have to spend some time recreating a dance on the floor, they would bring a great deal of experience that makes that task easier, and they wouldn't end up recreating the original quite the way its long-ago practitioners would see as equal. Amateurs would need quite a bit more time and guidance to recreate a dance well from notation.
I also play and enjoy chess as well as dancing, and I do enjoy reading about and using chess engines and how championships are impacted by them. Different strokes for different folks though, obviously nobody needs notation or printed materials or anything to learn how to dance. My supposition is that more abstracted dance resources would help some people though who are being underserved by the current dance education market presently, but we will see.
Well, the problem is that with all the online resources, you quickly learn that chess is more about blunder avoidance than any brilliant, dynamic scheme.
Having danced for around a decade, I would say the hand positioning example is a bit reductive. The most obvious example is that there's only one variant of "both hands crossed over" - while it's true that the hands might only fit together one way, the things you can do with the hands depending on what arm is on top varies significantly.
Having said that I'm pretty excited about the topic and what the author can do with it.
It is, but also it's a helpful simplification, I believe. It's kind of like a cup and a ring being topologically equivalent. Helps you talk about some generic thing without overloading with details.
Then there are many more layers on top: palms facing up/down/side, cupped or pressed against, which part of the hand/arm/body are you holding, which one's on top, compression or stretch, at what height, etc.
To me, it's explicitly about being topologically equivalent. I brushed up on my topology and knot theory when working on this to make sure I had it right. The post is considering two social dancers as connected components of two topological objects and the 15 different ways are all the different ways specific points of the surface (hands reduced to points that is) are able to connect.
There's more topology as well when you start bringing in other characteristics of a position, like relative distance and orientation of the dances, you can find more equivalence classes for the other types of equivalences that fall out of that.
the supposed 15 hand holds is a little misleading as it doesn't account for holds where a hand is placed on the body, which is important to many dancing traditions, such as the mirror image polka hold, with the man's left arm stretched to his left holding his girl's right hand, and his right on her body, or the rotationally symmetric ceilidh hold, where both partners hold left hands, and pass their right hand over and through the hold to grasp their partner's left side, inside of their left arm. the statement needs a big caveat so it is clear that it does not accurately describe all real dance holds.
Yes all of that was purposefully left out. It's just about which hand is touching which other hand, not about how the hands are actually in contact. I was purposefully skipping those details to try and find something simple to build on.
Wow, that's pretty neat. I've not seen modern western square dancing and this calling business before. Certainly there is a language of moves/steps here that the dancers know well.
I see this as an almost scientific/scholarly exercise in modeling. While the author seems not to start with any intent to "do science" he or she tries to capture a concrete phenomenon abstractly in order to study it more closely, which is arguably at the heart of science or at least some of the sciences.
The post is "almost" scholarly because it does not try to look into what others did before or currently, eg. prior attempts at notation of dance. I would be very surprised if there weren't any. Modeling cultural practices is also somewhat of a hot topic in the so called "Digital humanities" for what that is worth. A quick search for example brought up this article which talks about Dance Studies and attempts at digitization:
As to whether this all will be useful for learning how to dance, it remains to be seen. I didn't learn how to dance from diagrams and notation, but I also don't think there's that many good diagrams and notations to learn from, so it's hard to say whether it would have been effective or not. My answer at this point is to try and produce stuff that helps people and see whether it's possible or not.
Choreology is an established discipline. There's no reason to
re-invent it without improving on it. Some people might do it
professionally. A big budget musical production typically will employ
a choreologist to make notes during the rehearsals. Members of the
company can ask for reminders about their parts, which the
choreologist not only will have written down but will be a good enough
dancer to demonstrate.
I've studied Labanotation and Benesh movement notation, among others, some on my own and tried my hand at annotating social dancing with it and didn't find it to be particularly useful for social danced partnerwork in particular. For Choreographed dances, I am sure having a choreologist on staff would be a god send. However, for improvised social dances, I don't feel like the dance notations that I have seen capture the cues and signals the lead sends to the follower in a way that is adequate to be learned from later on. If you have any resources about this though I would love to read them!
Hello! Zack here, author of the post, thrilled to see people excited about this stuff. A friend submitted this to HN and I just realized it, sorry for not swinging by sooner. Always happy to discuss this further, the blog post wasn't quite done yet and there will be more to add later. Will respond to comments when I have time.
Hah same - especially in the beginning - all the moves looked like programming functions - inputs and outputs, state mutations etc.
Thing is, the more I learned the more I started to use internally consistent vocabulary, so I no longer needed analogs, but would construct new moves from older simpler ones.
I don’t think though that this article captures the complexity of of social dance, maybe that’s why its so universally fun? Its not just arms, legs and body positions - there is weight transfer, musicality, disparity between partner’s skill levels, let alone all other dancers around you that you need to track so as not to have an accident. And thats before you start adding shines, styling and body movement.
There _is_ an underlying system of course - every move has a finite possible exits, if you dance “by the rules” but dance evolution is all about breaking those rules. And if you’re at that level, you can start mixing other dances into your move set … Its all incredibly hard to reason about.
Maybe thats why the gold standard is just to record your lessons and get back to them when you want to.
Thank you for pointing this out! I had seen this page previously and had tried halfheartedly to find a copy of the original pdf at the time and given up. Just now I tried again though and was able to find one. Seems like a great guide, thanks a ton!
Could someone explain what a hammerlock is and why it is significant? Searching just said it's a type of joint lock in wrestling which doesn't make sense to me in this context.
It's interesting, because you have to do something active about it. If you move into hammerlock, you can't continue moving the same way anymore, so you have to decide on some combination of releasing hands, rotating in the other direction, staying and playing in that position, etc.
I also enjoy juggling and have spent a lot of time try to reconcile siteswap notation into being somewhat useful for dancing. I haven't been able to make that much progress as compared to what I was able to show above, but in my heart I hope somebody figures it out someday. Maybe something to do with Cuban style arm knots, but who knows.
For each basic configuration of leader and follower and their bodies they looked at all permutations.
Then they constrained it first to the subset of those that are possible to do at all, kinesiologically.
Then further to the subset of those that could be danced with reasonable comfort.
And finally the subset of those that are easy enough to be taught to students and would work on a crowded social dancefloor.
None of this was done with the help of computers.
The most systemic documentation of this is possibly Mauricio Castro's book "Tango -- the structure of the dance".
But lately a lot of new books were published on tango technique; I may be out of the loop.
A friend of mine who's also a tango professional is currently looking into the feasibility of doing a PhD thesis on this topic.
He wants to use ML to spit spit out the full motion tree of tango.
Both to be able to document it automatically, i.e. using generated 3D animations, as well as to discover new combinations the manual approach used by Naveira, Frumboli and Salas, over a quarter century ago, will have missed.
reply