Loading ...
Sorry, an error occurred while loading the content.

Re: [XP] pairing

Expand Messages
  • jlangr
    ... Hi Ron-- Interesting. I always thought it more valuable to keep a consistent person ( owner ?) on a given task. So either of Ron and Chet would stick with
    Message 1 of 68 , Aug 26, 2009
    • 0 Attachment
      --- In extremeprogramming@yahoogroups.com, Ron Jeffries <ronjeffries@...> wrote:
      > Arlo's team's finding was that it was most productive to switch
      > every 90 minutes, by the clock, whether done or not ... leaving the
      > problem to the person who had been at the machine for the //least//
      > time. That is:
      >
      > 1: Ron and Chet;
      > 2. Chet and Bill;
      > 3. Bill and Sam;
      > 4. Sam and Sue ...

      Hi Ron--

      Interesting. I always thought it more valuable to keep a consistent person ("owner"?) on a given task. So either of Ron and Chet would stick with it.

      Wouldn't this "least-time-stays" mode tend to produce more interruptions--where, say, Sam has to stop and go ask questions of Bill (who by then might have forgotten what they were thinking, and then go back and ask Chet, and ...)?

      As someone suggested, certainly the environment comes into play. Working with a legacy C++ system, I've found that many times, an hour in, we're just barely getting started on something, and really a good half-day is probably warranted. But in other (not so painful) environments, 90 minutes to two hours works just fine.

      I look at the cost of context switching as a significant part of the "review" in pairing. Often I've noticed that when it's high, it's an indicator that the code quality is lower--lots of SRP violations, code too convoluted, tests that don't document well--basically, code not refactored sufficiently (and in some cases, because it was put off too long, i.e. not done incrementally every few minutes). Other times the task was simply too large, and in some cases, yes, switching was premature.

      As someone else suggested, the goal has to be getting the job done, not switching just for the sake of switching. But I think you do want to push having at least 3 sets of eyes on most solutions, thus requiring at least one switch. It's easy for two people to produce a solution that only they understood, because they were the ones actively building it. The question is, does someone else uninvolved understand it?

      Jeff
    • Adam Sroka
      Hi Chris: I, for one, am more for trying it and seeing what results I get. Analyzing someone else s data is considerably less interesting to me. It is possible
      Message 68 of 68 , Sep 6, 2009
      • 0 Attachment
        Hi Chris:

        I, for one, am more for trying it and seeing what results I get.
        Analyzing someone else's data is considerably less interesting to me.
        It is possible that mine is more representative of the views here, but
        I'm not sure that's true. In any case, I'm fairly sure you're not
        alone, you're just the most vocal representative of that view.

        On the other hand, if you can't get access to Arlo's data in a way
        that would make it possible for you to reliably verify it, you could
        always design your own experiment to prove the part of the problem
        that is interesting to you.

        On Sun, Sep 6, 2009 at 4:36 PM, Chris
        Wheeler<christopher.wheeler@...> wrote:
        >
        >
        > On Sun, Sep 6, 2009 at 10:20 AM, jeffgrigg63132
        > <jeffgrigg@...>wrote:
        >
        >>
        >> I think that the problem is that we are dealing with social issues: We're
        >> not trying to determine the trajectory of a missile through space, or the
        >> effect of a drug on (typical) patients. We're really trying to determine
        >> what environmental factors result in the most creative and effective work
        >> products produced by groups of human beings.
        >>
        >> I would certainly like there to be more rigorous scientific research. And
        >> I would like it if more people would measure and report what they're
        >> experiencing in real-world projects. But until someone magically makes
        >> those miracles happen, I find that I must continue to make decisions based
        >> only on the best information available at the time -- which, for this
        >> industry, has been mostly experience reports and personal experience.
        >>
        >
        > <sigh>
        >
        > I think this will be my last post on this matter - I doubt we are getting
        > further down any road. Here's what I was asking for:
        > 1) a more detailed description of the method that was used
        > 2) access to the data so that I could verify it.
        >
        > Previous posts describe why I wanted this information. I'm not looking for
        > double blind studies or formulation of a theoretical model. I am looking for
        > more information that would lead to the characterization of factors that are
        > significant in the experiment that Arlo conducted.
        >
        > It may be the case that I am the only person in this discussion group that
        > cares about this, and that is fine by me: I'm not going to attempt to
        > convince you that there is more that you could learn from Arlo's experiment.
        > On the other hand, I'm not going to budge from my own position just because
        > many of you eschew such an approach.
        >
        > To each his own, I suppose.
        >
        > Chris.
        >
        > [Non-text portions of this message have been removed]
        >
        >
      Your message has been successfully submitted and would be delivered to recipients shortly.