A roof is a chain of water-control decisions.

Not the field of the covering, which is what most people watch. Roofs overwhelmingly fail where the covering stops: valleys, the flashing where a roof meets a wall, chimney and skylight surrounds, plumbing boots, and the edges. Those are the points that rely on detailing rather than on the material, which is why installation quality shows up there first and why an assessment spends most of its time on them.

Sometimes it is permitted, and it is rarely a good idea. A second layer means nobody has looked at the deck since the first roof went on, it adds weight the framing was not designed around, and it makes the next tear-off longer and more expensive. It also shortens the life of the new covering, because the layer underneath holds heat. Check what your city allows, then ask whether you want it anyway.

For the water the shingles do not shed. Wind drives rain up under a course, a valley backs up behind debris, a fastener sits proud, a detail is imperfect. Underlayment is the layer that deals with all of that, and it is also the temporary roof between the tear-off and the new covering going on. It is invisible for the entire life of the roof, which is exactly why it is worth naming by product in a scope.

It decides how hot the covering runs from underneath. A roof takes sun on top and, in an under-ventilated attic, heat from below as well, and asphalt ages faster under both together than under either alone. Intake at the eaves and exhaust higher up have to be balanced and unobstructed to do anything — adding exhaust while the intake is blocked achieves very little. It is a system question rather than a count of vents.