I think those numbers ware measured in some kind of company.. where you have an idea of what you want to make, in difference to when you are making it yourself.

And another suggestion , not absolutely, for what he meant might be of "flawless code" if something that limits to it exists.

FWIW - I work in a very large company, with all the procedures, etc.

1. Managers love metrics. Not because they necessarily have meaning, but more because they provide a way to report on something (programming) that is difficult to report on to senior management. LOC is one of those metrics.

2. I wrote 2400 lines of ksh script last month - comments and blank lines discounted. There are months when I write less than 100 lines of script.

At the moment, I'm working on C - wrote and tested about 400 lines last week.
However, 10 lines of shell script can "stand for" several hundred lines of C. LOC is not a metric with meaning in this context.

3. Poor code. For example, C code that repeats itself a lot and violates the "Don't repeat yourself" dictum will have a very high LOC compared to properly constructed code. The cost to the company to maintain cruddy code like that is very high. A programmer may have to read through thousands of lines of code, making the same correction over and over - even if he decides to pull it out into one function. Here, LOC produces counterintuitive results.

That's a part big part of the point i was trying to make.... I think, though, that you could make it clearer :P