Randomization in Science, Society and Nature

Randomization, characterized abstractly, involves an agent deliberately “flipping a coin”, or using a randomizing device, in order to choose between alternatives. The agent could be a person, a non-human organism, or “nature” (i.e. evolution). A simple Bayesian argument suggests that randomizing, rather than choosing a certain alternative, can never be strictly advantageous, and is only permissible when the options are of equal value. Yet randomization is widely used in a range of contexts, for various purposes. The aim of this workshop is to reflect on the use of randomization in five such contexts:

1. decision and game theory (why do agents sometimes prefer to randomize?)

2. clinical trials / experiments (is randomization necessary for causal inference? what is its justification?)

3. politics and society (when is allocation by lottery/sortition a good idea and why?)

4. evolutionary biology (when are bet-hedging strategies advantageous?)

5. genetics (why did sex and recombination, i.e. random shuffling of genes, evolve? why did meiosis evolve to be fair?) 

The unifying thread is the question of when randomization (by agents or by nature) is "valuable", where this can mean rational, scientifically useful, socially useful, fair, or evolutionarily advantageous, depending on the context.