Home > Productivity > Halstead metrics for software complexity and an extrapolation to agreement complexity
Published on:

Halstead metrics for software complexity and an extrapolation to agreement complexity

By

A computer scientist named Maurice Halstead devised a measure of software complexity and the mental effort required to create it. As explained in IEEE Spectrum, Oct. 2010 at 34, the quantitative measurements, later called Halstead metrics, “counted the number of unique operators and operands as well as the number of operator and operand occurrences in source code.”

Translate that function into the context of legal agreements. Software could count the number of defined terms, the number of times they are used, the number of paragraphs (or pages or words), the number of nested lists, and other characteristics of agreements that reflect complexity. With that software and a common measure of legal complexity, it would be easier to assess the productivity of legal departments, the level of client demand for agreements, the value delivered by law firms, and improvements in the structure and content of agreements. Maybe this blogger will be remembered as the progenitor of “Morrison metrics”?

By
Posted in: Productivity
Published on:
Updated:

One response to “Halstead metrics for software complexity and an extrapolation to agreement complexity”

  1. Steven B. Levy, Author of Legal Project Management says:
    October 31, 2010 at 10:56 am

    The problem is that Halstead metrics measure code complexity (sort of) but do not measure either problem complexity, solution elegance (maintainability and resilience to changes and defects), or level of fit between problem and solution. IT departments that use Halstead metrics or equivalents to measure programmers reward inefficiency and quotidian semi-competence instead of actual problem solving and value.
    The legal industry already has problems mistaking quantity for quality. (We’re not alone in this regard, of course.) I love both the alliterative and quantitative aspects of Morrison metrics, but remember the First Law of Metrics: You get what you measure. To that end, do we really want to focus on measures of complexity?