What are the consideration at play to choose the length of a cycle ? As any other economic parameter in Tezos it is subject to amendment however it would be great to collect here the things to think about on this.
Fundamentally a cycle is a set of block heights for which the baking en endorsement rights are determined by a single random seed.
The reason to have many blocks in a cycle is that it lowers the incentive to try and manipulate the random seed (since most seeds will produce similar distributions on average). The reason not to have too many blocks in a cycle is that, the longer the cycle, the less representative of the stake distribution the rights might be. However, that is more affected by slashing conditions than by cycle length.
Cycle lengths are also used to measure different periods, but that is mostly out of convenience, not necessity. For instance, commitment to random nonces happen in one cycle, and the reveal in the next, but it would also be possible to have the former happen in the first half of a cycle and the latter in a second.
This is an empirical parameter but at high level it needs to be long enough to get enough to get statistical sample (for the randomization for the baking rights snapshot) and short enough to not know the expected rewards too much in advance. And again like every other parameter in Tezos it is subject to amendment if the community believes there is a strong case for it.