This is a long discussion and you'll find most of the data here, at cdfreaks and at cdr-info, if you search the forums (you can search with my username halc / halcyon, although not all data comes from me, but I've frequented many of the threads in question, often learning from those who know more than me).
It really boils down to this:
There are no simple 'metric' standard for what a dvd reading / writing should be. I.e. there is no absolute benchmark
That is, one drive optimizes for low DC jitter, another one for other disc low level characteristics.
These burns will read back differently on different drives, which in turn optimize the reading process differently.
The error often quoted are just what the drive sees. They are not 'absolute' errors low level on the disc.
They are errors that the reading drive made in the process of reading the disc in question. Hence, they vary from drive to drive and even from scan to scan (the process is stochastic, not deterministic).
Granted, both the reading AND writing have improved a lot in recent years.
Still, it is possible to get a burn that scan very well in drive 1, but quite bad in drive 2 or does not read back at all in drive 3.
As a physicist in training you understand what this means.
So, the process of measuring 'write quality' becomes an exercise in eliminating random and systemic error. First in instrumentation, then in statistical number crunching.
By measuring with various different drives (testing for the drive scan-to-scan variation first) one gains a better understanding of how the burn behaves in a larger population of drives.
That is, what is the likelihood that it actually reads back ok in most of the drives people have out there, i.e. it's a 'good burn'.*
* Actually, for this purpose one would need to have a lot of older drives as well, some of which are not so good readers by modern standards, but which represent a sizeable portion of the market of drives. However, this is proving to be increasingly more difficult, as the number of test capable 'older' drives available has become small.
That is the overall picture. Then there are the details about optical pick up heads (their variation), chipsets (how they read, filter and report back the data), programs used for scanning and what conclusions can be drawn. Additionally longevity (how the discs will age and retain readable data) is a completely different thing altogether.
If you want to read more, I can suggest a couple of articles in the Finnish MikroPC-lehti. The other published stuff is either from NIST/OSTA or in the German c't magazine.
The MikroPC articles in question (in Finnish):
http://mikropc.net/rml/arkisto/mikropc/ ... 200522.pdf
http://mikropc.net/rml/arkisto/mikropc/ ... Q4g09.html
http://mikropc.net/rml/arkisto/mikropc/ ... 200736.pdf
Unfortunately even the above is a scratching of the surface.
The best source? Your own research, including the forums already mentioned and a lot of hands on testing and statistical pruning :)
With that said, if you just want to get an overall feel of how things are with your burns, using your drives, then just getting a recent Lite-On (like 20A4P) might not be such a bad idea. It may not be always accurate in all cases, but for most non-research uses it's 'good enough'.
Hope I didn't bore you to death :)