Why does over-proofed dough collapse?
This looks like the same problem as why cakes collapse, but the mechanism is genuinely different -- yeast dough fails by over-stretching its own structure, not by outpacing it.
The kitchen answer
Under-proofed dough is dense because it hasn't made enough gas yet. Over-proofed dough collapses because it made too much for its own gluten network to keep holding -- the structure doesn't fail from lack of gas, it fails from too much of it stretching the walls until they tear.
What's actually happening
Yeast keeps producing CO₂ for as long as it's alive and fed (see Leavening), and each gas bubble stretches the surrounding gluten network thinner as it expands. Gluten can stretch a remarkable amount before it fails -- but it isn't infinite. Past a certain point, the bubble walls get so thin that they start rupturing, releasing their gas into neighboring pockets or out of the dough entirely.
Once enough of those walls fail at once, the dough loses its ability to hold shape structurally, not just its gas -- which is why over-proofed dough doesn't just "not rise further," it visibly deflates and often turns slack and gummy rather than springy.
What does that mean for your recipe
The classic poke test exists because it's directly testing gluten's remaining stretch, not just the dough's size: press a floured finger in gently -- if it springs back quickly, there's plenty of structural capacity left (under-proofed); if it springs back slowly and partially, it's ready; if it doesn't spring back at all, the network is already near its limit and losing elasticity, a warning sign that it's headed toward over-proofed rather than a confirmation it's done.
What can go wrong
- Dough that deflates when handled or scored: over-proofed -- the structure was already close to failing, and the disturbance finished the job.
- Gummy, dense crumb despite a long rise: the gluten network partially ruptured during proofing, so gas escaped before baking could lock the structure in place -- more rise time isn't the fix, less is.
- Warm kitchen speeding everything up unnoticed: yeast activity and gas production both accelerate with temperature, so a dough left "the usual amount of time" on a warm day can genuinely over-proof when it wouldn't have on a cooler one.
- Relying on time alone instead of the poke test: proof time is a starting estimate, not a guarantee -- temperature, hydration, and yeast quantity all shift the real timeline.
Try it yourself
Same dough, three proofing points:
- Make one batch of dough, divide into three pieces.
- Bake one under-proofed (short rise), one properly proofed (poke test passes), one deliberately over-proofed (well past the poke test, until it looks slack).
- Compare crumb structure, height, and texture across all three.
Watch for: the over-proofed piece rising the least in the oven despite having risen the most beforehand -- direct, edible proof that "more time proofing" and "more final height" aren't the same thing past a certain point.
Go deeper
This is structurally the mirror image of why a cake collapses: a cake collapses when its setting systems (egg protein, starch) are too slow relative to gas production, while yeast dough collapses when its holding system (gluten, which doesn't "set" with heat the same way, just stretches until it tears) simply runs out of stretch. Different failure mode, same underlying idea: gas production and structural capacity have to stay roughly matched.