No single metric can be used to measure the internal complexity of software.

When several metrics are used they ideally should measure independent attributes and should thus not correlate.

Software length, measured as LOC (lines of code), is probably the most frequently used complexity metric.

Another popular metric is McCabe's cyclomatic complexity metric (MCCM), which for a program is 1+ the count of decision nodes. MCCM can therefore be expected to correlate with length.

It has therefore been suggested to instead measure "complexity density" (CD) for instance as MCCM/LOC, resulting in LOC and CD being independent metrics.

This gave us the idea to generally treat metrics related in this way.