The general thinking of many people (including myself when I first started out), is that they would want to record tracks with the maximum number of bits/ resolution without clipping. In reality, the Analog-to-Digital Converters (ADC) are mixed-signal components that have optimum range of operation. In the hotter ranges towards digital zero 0db, there is actually a slight bit more degradation introduced to the sound than in the lower level ranges. Provided that audio is recorded with 24-bits, you would actually have a 144db of dynamic range. That’s enough to capture the difference of volume of a sleeping baby vs a roaring jet engine.
With hot levels, you could always risk of momentarily going above 0 (digital clipping). While the meters of the DAW or hardware metering may not show, the reality is that the meter itself may not be fast enough to catch the quick transients (for things like drums, for example). It is rather better to record with conservative levels than having the chance of ruining that perfect take. If you’re going to record all hot, then find the need to reduce the levels of every channel in your DAW (which mathematically subtracts from each recorded track) that feeds the master bus, where it all sums back together, then what’s really the point of having the hot levels in the first place?
While I’m not here to say one should aim for the opposite extreme of recording with the lowest levels as possible, because of noise floors, but having a peak of -12db in the DAW meters (when the volume fader is unaltered and set to 0) should be a good target to aim for.