Decibels are only ever a relative measurement on a logarithmic scale. In this case, we aren't even talking about sound (these are dB of power), but this is true independent of base unit. For sound, to represent silence, you actually need -∞ dB. The fact that this is a relative scale is also why your mixer/home theatre receiver represents full intensity as 0 dB and any volume adjustments as some negative number of dB (some mixers may allow you to go a little bit over 0 dB but of course this risks clipping when a full-strength signal comes in).
Fun fact though: the sound scale is capped on the upper extreme depending on your environment. You know how recordings of launches of rockets sound like the sound is clipping? It's not hardware limitations. The sound is actually clipping as the pressure hits vacuum. See https://en.wikipedia.org/wiki/Sound_pressure#Sound_pressure_... .
A 0dB tone is a tuning tone. Most amplifiers are tuned to a particular frequency at 0dB, and with a good phone connection, you can even do this over the wire. It's problematic if there are too many hops between you and the source. For example, calling a 0dB tone in a small CO in central Montana didn't give me a usable reference when I was calling from Huntsville, AL, in the 1970's. There were about four hops in there.
It was originally 1000Hz, but it turns out this could cause framing errors on D4 trunks because the framing recovery circuit could misidentify the bit pattern as valid framing bits. So they changed it to 1004Hz, enough to sidestep the problem, but still close enough that it would still represent "exactly" a milliwatt of energy, within the required precision of the instruments anyway.
When ESF framing was invented, precise tones no longer posed a problem, but a lot of D4 trunks hung around for decades, so last I checked most milliwatt tones were still 1004Hz.
Any phone people here?