Hot Jupiters orbit perilously close to their stars — periods of days, temperatures of thousands of kelvin. They didn't form there. They migrated inward, likely through high-eccentricity pathways: gravitational kicks from other planets or companion stars flung them onto wildly elongated orbits, and tidal friction at perihelion circularized them. But circularization dissipates orbital energy. In many simulations, the dissipation is lethal — the planet spirals in and is destroyed before the orbit can circularize. This overproduction of dead planets has been a persistent problem: models predict too few survivors compared to what surveys observe (~0.5% of FGK stars host a hot Jupiter).
Vick, Lai, and Anderson (arXiv:2510.26882) find that mass loss — the very process that seems most destructive — is what saves them. As the planet plunges through perihelion on its elongated orbit, intense tidal heating strips mass. The stripped mass carries angular momentum. The angular momentum return pushes the orbit outward, slowing the inward spiral. The planet bleeds to survive.
The survival rate increases by a factor of 2-3 with mass loss included, closely matching the observed occurrence rate. The mechanism is self-limiting: a planet that loses too much mass loses its gaseous envelope entirely and becomes a bare core — a “hot Neptune” or smaller object that no longer registers as a hot Jupiter. The window is narrow. Enough mass loss to return angular momentum and arrest the spiral, not so much that the planet evaporates. The survivors are the ones that bled the right amount.
The structural insight is that a process can be simultaneously destructive and protective, depending on timescale. Mass loss on orbital timescales (perihelion passages) strips the planet. Mass loss on migration timescales (thousands of orbits) returns the angular momentum that prevents tidal destruction. The same mechanism, measured against different clocks, does opposite things. The planet is being killed and rescued by the same process — which one wins depends on how fast.