Re: [Swarm-modeling] american football vs other handball sports

[Scott Christley]

American football is a hybrid (discrete/continuous) simulation with continuous mechanics that take it from one discrete state to the next. With an interesting point being that the continuous mechanics are an adaptive (online) adversarial system with the resultant discrete state determined by the adversarial outcome. The team "with the ball" gets to (partially) decide the initial state for the next adversarial encounter. Many other team sports, soccer, handball, hockey, field hockey, water polo, basketball and so on, are almost purely continuous simulations. They have no states per se but are more statistical mechanical in that they exhibit "energy configurations", some with higher probabilities than others. Transitions occur when the team "with the ball" enters a low probability configuration. I'll note that when these continuous simulations fail to resolve themselves, they often default to a discrete system so that some solution can be found.

I would tend to agree with your friend that American football is the "most poetic". Poetry (and I'm just regurgitating wikipedia...) may allow ambiguity in semantics but is more precisely defined by order and structure; poems have a rhythm, meter, patterns and forms. The alphabet of poetic states are language words that are arranged according to rules (grammar) that take you from one state to the next, i.e. one line to the next, one stanza to the next. The alphabet of American football are specialized players that are arranged according to rules (allowable configurations) that also proceed from one state to the next. A distant second might be baseball?

As for "amenable to high order _expression_", I'm less sure. I feel those continuous sports might have the potential to display high-order configurations, but in actuality they are too short-lived to present themselves. Mostly the first-order "energy configurations" present them, and there might be a simple statistical hypothesis that the team which reaches the most high-probability configurations during a game is the most likely to win. Furthermore, there aren't enough agents in the system to create high-order complexity through scale. I also feel that American football has high-order potential though it's based upon the size of its state space (infinite? enumerable?) and I don't think there are particularly simple or compact expressions to describe the transition from one state to the next.

cheers

Scott

On August 18, 2021 at 3:51 PM, glen ep ropellaa <gepr@agent-based-modeling.com> wrote:

Update! We had a chance to argue a bit about this. So, by "poetic", my friend actually means "amenable to high order _expression_" ... my words, not his. The idea being that zero intelligence agents with relatively high diversity compose in ways prevented by high intelligence agents with low diversity. That, of course, should have been followed by a discussion of whether stacking orders of _expression_ leads to more or fewer degrees of freedom. But it didn't. Instead we pivoted to the idea of a game theoretic assessment of various team sports, their objectives (both explicit - e.g. scoring, and implicit - e.g. athleticism), and complexity.

Off-list, a swarm-modeling subscriber suggested that the progressive nature of sports like American football is the primary appeal. But I think this, too, can be reduced to amenability to higher order operators. The 0th order might lie in the exercises and physiology of the players (kickers train differently from those on the line, etc.) and their bodies will exhibit different repertoires based on that training. 1st order might be the brains of the players and how they use their bodies. 2nd order might be positions and players' fit-to-purpose for those positions. 3rd order would be tactical assembly of the positions to suit given ends (not all reducible to yards gained or scoring). Etc.

Anyway, I kindasorta buy his argument at this point. By analogy with, say, chess, one could add another row and column to the board, invent a new piece with a new move set, and calculate the number of games that addition produces. Then compare that number to the number the original game produces. It might be less or more ... but probably not the same.

But I completely reject the use of the word "poetic" because poetry relies on ambiguity, the ability of any given symbol to hold more than 1 meaning. Another person in the discussion suggested "elegance" instead, which I think is slightly better than "poetic" because elegant theories are more compressed/compressible than inelegant theories. And compressibility can be correlated with amenability to higher order operators.

As always, swats with the clue stick are welcome.

On 8/13/21 9:22 AM, glen ep ropellaa wrote:

So a friend of mine made the statement that American football is "the most poetic sport". I objected strongly because my sense is that it's too explosive, not only in the individual players, but in the plays ... too discretized ... "punctuated".

I think his argument is that because players are so specialized (defense, offense, qb, kicker, backs, etc.), game strategy is more orchestrated than sports like rugby where individuals are more general. But he made the argument by talking about the athletic prowess. And because players are specialized, my sense is they *should* be able to set records and do any small set of things better than anyone else on the planet. The olympics are similar. We wouldn't expect a shot putter to be good at the balance beam. This raises the old problem of where the "logic" lies, in the individual or the collective, and orchestration by a composer/conductor/coach or via some sort of emergent structure. Which is the better *team athlete*? The one who tightly specializes and knows her place? Or the one who smears her capabilities across a large repertoire of tasks?

A brief duckduckgo search shows some hints that research on tactic complexity and such happens in these sports. But, so far, it seems to lack the research bloom we see in games like chess, go, or poker. Rather than argue in this qualitative way, I'd like to discover some quantitative results, if they're out there. There's bound to be such things, especially with all the simulations we build for various video games.

Any ideas would be appreciated.

--
glen ep ropella 971-599-3737