Skip to content
 

An R programmer’s job is never done

Russell writes:

Have you ever taken a look at the version numbers of the packages in CRAN. The median and probably the third quartile are below 1.0. What does that imply about statisticians? R programmers?

I say, let’s just start with version 2.0 and go from there; that will relieve some of the pressure to get version 1 working ok.

5 Comments

  1. Kaiser says:

    That's what many startup software companies do… to give the impression that the software is at a more advanced stage, even though most use their first customers as testbeds.

    What you are pointing out is the Achilles heel of the open source paradigm. People who write software for free are doing it for reasons other than to satisfy the most needy, nitpicking "customer". They/we are much more interested in the process of creation than that of maintenance and incremental improvement.

  2. An optimist would interpret this as simply meaning that R is great for rapidly whipping up utilitarian scripts, functions and packages that do one thing and do it well enough.

  3. Martin Theus says:

    I would rather conclude that much of the stuff out there on CRAN is mostly quarter baken and probably not very well tested.

    Many of the packages would benefit from being tested by users who know the risk they take before being released to "the public".

  4. b says:

    I read it as modesty. Some developers are willing to confess that their work has room for improvement, and some feel it more important to pretend their work is flawless.

    Anyway, version numbers are what the economists call `cheap talk', and most of us know to take it as an empty signal.

  5. Chris says:

    People who write commercial, off-the-shelf software for profit are doing it for reasons other than to satisfy the most needy, nitpicking customer as well, aren't they?