Exploration & Conquest
The Feedback Canon — Installment #11b
This is the follow-up to On The Shoulders Of Big Data and part of an ongoing series delving into feedback. While repetition is a major concept in feedback, these articles are more iterative by nature. If you are having trouble following, you may need to start at the beginning.
In this article, we promised to use examples (games specifically) to explore and conquer the largest differentiating attributes of feedback. Enter Marco Polo and Rochambeau — the Italian explorer and the French conqueror. And two guys that have been blessed with their own games.
Rochambeau
Rochambeau was a French General during the American Revolutionary War. The game that bears his name is more commonly known as Rock-Paper-Scissors. It actually has other names in other cultures as well.
For our purposes, Rochambeau is convenient for tying our story together but more so because it offers a simple and effective feedback system. Two participants, three options, easy rules, it makes a great model.
Marco Polo
The Italian explorer is more famous than his French counterpart. His game, while not necessarily more popular, bares only one name.
There can be any number of participants. Only one is ‘it’. Feedback here is quite binary, other than an occasional ‘Fish Out Of Water’ call. The latter is a rule that is not universal and we will be ignoring it here to simplify the story.
Tag — Your It.
In an earlier article, I made the following formal analogy. Feedback is to data as a vector is to a scalar. Against the wide array of feedback, this is a vast oversimplification. In the game of Marco Polo, it is spot on.
Assuming I am ‘it’, my role in the game is to find the other player while keeping my eyes closed. I do this by demanding audible feedback from the other players. I announce — Marco. By the rules, upon hearing this, they must return — Polo. A simple feedback loop at it’s best. That ‘Polo’ contains both magnitude and direction. It is, in fact, a vector.
Using this vector, I can locate the other players. I will be most effective at this if I include a little context. Since Marco Polo is typically played in the pool, it helps if I have a sense for the size and shape of the pool as well as my own relative position in it. As I approach another player I hope to tag, I am likely to increase the frequency of my — Marcos. This forces them to provide more frequent feedback as well — allowing me to better find them.
Of course, if they swim underwater, they are not required to answer. And so the game is played with one player hoping for the clearest and most precise feedback and everyone else trying to obscure it while not breaking the rules (or at least getting caught).
But we don’t have time for all that…
Rochambeau is a much quicker game, especially if we set aside “best 3 out of 5” components. The game can be over with a single signal from each side, although often requires a few. If each side picks the same symbol (rock, paper, or scissors), we get a tie and need to re-shoot. Otherwise, someone wins and someone loses.
Feedback here is not a vector. There is no magnitude or direction. There is simply a choice and pre-defined calculation to determine the outcome. It is highly debatable how much value can even be gained from frequency or context in a sudden-death match.
This can be challenged of course, the infamous and fictional Vizzini of Princess Bride fame would clearly have an intellectual strategy based on real (or imagined) feedback he drew from simply observing his competitor. And perhaps he would have been right, had it been a fair fight. Honestly, I doubt it. But, if we are playing “best 2 out of 3” or “3 out of 5”, things begin to get more complicated.
A “best of” match starts allowing each player to employ game theory, not necessarily to any real advantage. But the addition of a serial format, allows the feedback stream to grow in complexity as well. Competitors begin attempting to predict each other. The value of history and context grows. The magnitude of that value is very debatable, so I wouldn’t go in against a sicilian or anyone else for that matter under the belief that you have some great skill in “reading” you opponent here. Regardless, it is not inconceivable that some advantage can be gained.
Two different men, two different but simple games, two vastly different feedback models
Hopefully, you are beginning to sense the complexity that is inherit in any feedback system. If you take the perspective of a developer, game, or AI — you start to realize where we must go next. These two games have many similarities and many differences. Each touches on many of the aspects of feedback we covered in prior articles — frequency, magnitude, and purpose to chose three.
We need to compare and contrast. But for that, we will need another addition to this article.
Compare & Contrast — Coming soon…
For more from this series consider:
