On pair programming

[standage]

This is a place to mention a specific case where the "everybody makes coffee" assumption is important. The more specialized the problem domain, the more likely it is that the contributors will not have the same level of expertise on the subject. I brought up khmer in my last talk. We have several students and postdocs in Titus' lab contributing to this DNA sequence analysis software, but some are using it for sequence assembly, some are using it to study genome variation, some are using it to study diversity of environmental samples, etc. Pairing two of us up would work great for on-boarding and getting up to speed on (for example) the API, but not so much for hacking away at the most scientifically interesting bits of the codebase.

[standage]

The current speaker notes do a great job of describing the circumstances in which pair programming is or is not a good idea.