Recent research results have suggested that double percolation processes play a significant role in the formation of intermediate phases (IPs) in non-crystalline thin films. One class of IP windows, involving competitive double percolation, occurs in binary As_xSe_1−x and Ge_xSe_1−x alloys and in the pseudo-binary As_xGe_xSe_1−2x alloy. This IP window occurs over an 8–12% composition range. Transitions that define the IP window in the Ge–Se alloys involve a competition between the elimination of compliant local bonding dimer groups, e.g. Ge–Se–Se–Ge, at the expense of an increasing fraction of rigid local bonding monomer groups, e.g. Ge–Se–Ge. Compliant monomer group bonding defines the first window transition for an average number of bonding constraints/atom, n_c = 3; the second transition from rigid to stressed rigid occurs when the compliant monomer concentration drops below a concentration for percolation. A second class of IPs with significantly narrower composition windows, ~1 to at most 3%, is proposed to explain experimentally determined IPs in chalcogenide alloys with halogen dopants; e.g. a-Ge_0.25Se_0.77−xI_x, where the IP window width is ~1%. We suggest that this narrow window is determined by a confluent coherent double percolation process that includes (i) broken bond-bending constraints that minimize local bond strain, and (ii) a percolation pathway based on a second and complementary local bonding group. However, this second class of IPs is not supported by theory and modeling as yet, and as such our designation of this class of IPs must be regarded as more speculative. On the other had, it is significant that at least two other alloy systems, Ge₂Se₂Te₅ and pseudo-ternary Hf, Zr and Ti Si oxynitrides, display narrow regimes where bond constraint counting indicates local strain suppression, and where a second and larger bonding arrangement is present at the percolation limit.