Indeed: Pair programming does not work for all programmers or all programming shops. It is possible to do it wrong, especially by making it mandatory for people who do not want to do it. Depending on its implementation it may have other negative properties that render it unsuitable for any given purpose. Its suitability for any particular purpose is a broad question which cannot be adequately addressed in the context of this little conversation!
But.
In my experience, when I was pair programming, and the pair programming was effective to begin with, then: when there were interruptions, the interruption-recovery process was significantly streamlined. I infer that this is one of the ways through which teams that actually already do pair programming manage to improve their productivity enough (relative to singleton-programming) to justify the arrangement.
But. In my experience, when I was pair programming, and the pair programming was effective to begin with, then: when there were interruptions, the interruption-recovery process was significantly streamlined. I infer that this is one of the ways through which teams that actually already do pair programming manage to improve their productivity enough (relative to singleton-programming) to justify the arrangement.