Social systems require a latent plasticity of response to unpredictable new conditions on larger scales (climate change, invasions, new technologies) and this is embodied in the variability of units at lower scales. The normally maintained distribution of numbers of units favors the normal types which interact in predictable ways to produce the usual emergent higher-scale phenomena that characterize the system’s meta-stable persistent and self-reproducing dynamics. But it also includes a fluctuating percentage of outliers which are capable of introducing perturbations which the system at larger scales ‘resists’ by normal filtering and buffering mechanisms, except when the functionality of these is disrupted as part of a general disruption due to larger-scale unusual forces acting on the system (these may be internal unpredictable emergents as well as extenal changes). In such circumstances the net diversity of the system increases and the effects of former outliers can cascade upwards (amplification by critical point catastrophe). These effects may or may not generate sufficient novelty to allow the system to adapt to the new situation (law of requisite variety being necessary but not sufficient for adaptation and survival with change).
It is theoretically interesting to compare, as in evolutionary biology, the conditions under which perturbations of an ecosystem favor the survival of the most normative members vs. those that favor the survival of (some) outliers over more central regions of the distribution space. My guess is that when perturbations are random and not statistically very different from those which have shaped the previous evolution/adaptation of the system, then on average, such perturbations will favor the pre-adapted (central) members. It is only when the perturbation are unprecedented (as with internal emergents or first-meetings with external systems), and when all their instances push the average environment of members in some systematic direction away from the prior norm, that outliers are favored.
So, for example, a cyclic economic downturn will favor conservative, centripetal investors, and such a strategy will be optimal over many cycles up and down. But when a new technology appears, and makes large-scale and permanent shifts in the economy (distributed electric power, inexpensive computing capacity), then some outliers who have a non-conservative strategy may be favored, but only over the long-term. Over the short-term there will still be fluctuations that will favor the old-line conservatives.
There are two sorts of identity outliers: the counter-normative and the unprecedented. Counter-normative performances re-inscribe the polarities of the existing system, and at the same time prevent over-rigidification and uniformity which would prevent the system from taking advantage of any genuine latent diversity when it is needed for survival. Unprecedented outliers present the genuine novelty and variety of the system. They tend to be seen as not making sense, rather than as criminal or evil. They may be regarded as idiosyncatic, dysfunctional, disturbing, insane (but with no precedented diagnostic category available), and transgressive only as a side-effect. Not deliberately transgressive, but accidentally transgressive. Going their own way they do not necessarily notice or care that they are transgressing social norms. It must be pointed out to them, and only when the cultural police are too busy saving themselves from disaster to effectively limit the effects of these outliers are their effects able to propagate extensionally and up-level.
There is also a strategy of deliberate transgression designed to test boundaries and expose implict norms, or force social examination at a larger scale of the grounds of those norms. Unfortunately it is only when this happens on a sufficient extensional scale that the re-examination can be forced, and this condition obtains only when there is once again already some sort of serious higher-scale perturbation in the social system.