Meta
The concept “meta” can mean various things. Grouping them together can lead to accidental conflation, so let’s try to unpack them somewhat. What follows is a list of distinct things which people may mean by “meta”. To some extent, there may be a coherent cluster (an “organizing principle” if you will) sitting behind all these diverse notions. However, conflation seems more dangerous than splitting, so let’s try to tease things apart.
Anti-Sphexishness
The term sphexishness was coined by Douglas Hofstadter to refer to static behaviors which might appear intelligent in their original context, but become transparently unthoughtful when the context changes and the behavior reveals itself to be fixed stimulus-response with no ability to adapt. This idea is sometimes used to illustrate how intelligence can be mechanical. The original sphex-wasp anecdote illustrating the idea is highly questionable (ungated copy), but the concept is nonetheless useful. “Anti-sphexishness” refers to the opposite trait: an ability to notice a context switch, and change behaviors accordingly. Hofstadter suggested that humans were anti-spexish by comparison to insects, with the ability to notice more sophisticated patterns and thus jump out of malfunctioning habits, but that humans are sphexish at a higher level (stuck in mechanical patterns that we can’t yet see).
Going Off-Script
“Going off-script” is a closely related concept, meaning breaking out of typical social patterns. Predictable social roles are useful, so you wouldn’t want to go off-script all the time. However, it can become a game of breaking expectations. Going off-script can also be useful, since the script often constrains motion. The ability to go off-script well requires a mastery over the script, and an understanding of how it can be manipulated.
Playing the Biggest Game You Can
A winning move in a narrow situation is usually at least a decent move in the broader situation, but not always. The ones who are playing the bigger game often have an advantage, in that they can afford to lose some smaller games.
Playing One Level Above Everyone Else
There is a concept of “levels of play” based on predicting the other player’s moves. A 0th-level player can’t predict other players. A 1st-level player predicts the 0th-level player, and can account for them in strategies. A 2nd-level player can predict 1st-level players. And so on. You typically want to play at one level above others.
This concept has very little game-theoretic backing. Douglas Hofstadter once wrote about his disappointment when an AI game-playing competition designed to bring out these levels of play (in a game all about predicting the other player and trying to choose a number exactly one higher than the other player choose) was won by a randomized strategy.
However, the concept may be related to how humans operate. We predict what others may do, and try to respond appropriately.
Metagaming
Metagaming has various meanings itself, but I won’t try to enumerate them. The most popular meaning (I believe) is the use of factors outside of a game to try to win the game. Notice that this is almost the opposite of “playing the biggest game you can”: in that case, you’re taking a broader view of your goals. In metagaming, your goal remains fixed, but you take a broader view of the playing field.
Object Language vs Meta Language
Logicians study logic. For the most part, you could say they study logic with logic; they prove theorems about the abilities, limitations, and properties of their own tools. However, due to Goedel’s incompleteness theorems and Tarski’s undefinability theorem, there are fundamental difficulties with reasoning about a logic using the very same logic. Instead, to study one logic (the “object language”), logicians use what is called a “meta-language” which discusses the target of study.
For this reason, “object level” is generally used to indicate the opposite of “meta level”: an object-level discussion is about the things/actions in the world, while a meta-level discussion is about the object-level discussion. This can, of course, be iterated; there are far more than two levels.
Self-Reference
Despite the previous point, there is also a rich study of what a formal system can say about itself (in fact, Goedel and Tarski’s theorems spring from this study). Whereas the previous notion of “meta” created a separate meta-level and meta-language which talk about lower levels, we have something which loops back on itself. This can happen if we try to iterate meta-levels “too much”, and create a meta-level that wraps back on itself rather than referencing strictly lower levels.
The popular conception of “meta” coming from Hofstadter places a heavy emphasis on self-reference.
Moving up Levels of Abstraction
People can also use “meta” to indicate any kind of moving up hierarchies of abstraction; not only subgoal-supergoal hierarchies (“playing the biggest game you can”) or language levels (“meta language”), but also ontological levels (from spruce to pine to tree to plant to life) or levels of complexity (from particle to atom to molecule to polymer to cell to tissue to organism).
Moving up Levels of Contrarianism
Contrarians signal their intellect by disagreeing with a common view. Meta-contrarians signal further intellect by dissenting from the contrarians. This can iterate arbitrarily. Other forms of meta are typically involved; playing at a higher level (finding the arguments of those at a lower level predictable), playing a bigger game (bringing in more diverse considerations which are being ignored by the simpler view). It’s not even necessarily bad: it provides a social mechanism encouraging forming more and more nuanced views. On the other hand, contrarianism for the sake of contrarianism tends to result in distorted views.
Of course, understanding meta-contrarianism, you’ve now entered into an intellectual elite of post-contrarians.
