The life cycle explains the pressure. Larvae are ingested by an arthropod, grow inside the body cavity until they nearly fill it, and then must exit into fresh water to mate. A terrestrial insect that dies on dry ground is a dead end for the parasite.
Field and laboratory observations show that infected crickets enter water at far higher rates than uninfected ones, and the worm emerges within seconds to minutes of contact. Proteomic work on infected hosts has found parasite-derived proteins in the insect's central nervous system, including molecules resembling those involved in host neural signalling, which suggests direct interference rather than simple debilitation.
Two caveats belong here. The manipulation is often described as making the insect suicidal, which overstates a bias in behaviour as a compulsion; some infected insects do not enter water. And horsehair worms do not infect humans, pets or livestock. Reports of worms found in a toilet or a drinking trough are almost always adults that emerged from an insect nearby. The molecular mechanism of the manipulation is still not fully identified.