Where is the confining string in random percolation

The percolating phase of whatever random percolation process resembles the confining vacuum of a gauge theory in most respects, with a string tension having a well-behaved continuum limit, a non trivial glueball spectrum and a deconfinement transition at a well determined temperature T_c. Simple numerical experiments reveal an underlying, strongly fluctuating, confining string, with an internal vortex structure formed by a core trapping inside a Coulomb-like phase composed by the vacuum at the percolation threshold. The width of the core almost coincides with 1/T_c and it turns out to be separated form the confining vacuum by a domain wall of definite thickness.