Peer to Peer
One of the things we think didn’t work well in this term’s online run of the course was peer-to-peer discussion among students. Such discussion is one sign of a vibrant open source project or educational community, but our mailing list was almost silent except for announcements we posted ourselves, and other than a few bug reports, there were very few comments from students on the lecture pages. We think students will get more out of the course if they talk more amongst themselves for several reasons:
- Learning: you don’t really know something well until you teach it yourself.
- Scalability: there aren’t enough of us to give everyone personal help (particularly not with numbers tripling next time ’round).
- Relevance: grad students in geology are more likely to know what other geology grad students need and will understand than we are.
- Sustainability: right now, Software Carpentry has a bus factor of 1. The more course participants help each other, the more likely it is that this project will survive its founder being hit by a bus abducted by aliens having to get a real job.
What could we have done this time around to encourage more peer-to-peer discussion? What could we do next time? Please add your thoughts as comments…
Categories: Uncategorized

I would suggest looking into a forum of some kind, either as a ‘one place for everyone’ (partitioned with sections and categories, of course) or walled off somehow so that each session of the class (and possibly divided smaller) has its own area (that would likely need to be wiped at some point after the conclusion of the course). Forums have their own issues in terms of needs for moderating and upkeep, but I think that model meshes better with a course than a mailing list. It seems to me that it is hard to separate a mailing list from the idea of emailing the instructor, which isn’t the idea you are trying to promote. My wife teaches some online courses (not technical ones) and this is the model they use, but there participation is required, and part of the grade.
This is a harder challenge than it looks. Larry Wall describes open source as “learning in public” which I love. But being seen to be unclueful in public is really hard and is, if I may speculate, actively discouraged in the sciences.
What makes people comfortable with asking questions in public. A long time ago a teacher told me something like, “if you have a question then most likely others have that question too, so do everyone a favor and ask it”. Can we teach this as a value?
Sometimes the experience is better than the apprehension. Could we require that participants ask and answer questions? (This is kind of pump-priming, but risks crowding out the desired motivation, as does participation for grades in undergrad courses).
Can we treat this as a topic on which they need to build self-efficacy? ie teach asking questions as a way to demonstrate competence, rather than incompetence. I have some teaching materials based on ESR’s “How to ask questions the smart way” that I’m happy to share. I call it “Asking questions people want to answer”.
Teaching and presenting information are generally very different. The piece ‘teaching’ is what I see coming up in the SC after action report as problematic: for example, prerequisites, ‘silent running’ and this collaboration issue.
Because I see it as whole cloth, let me comment in a single place (here) succinctly.
Teaching involves taking a person from one state (in skills and abilities) to another via instruction, drills, exercises, training. This only works well if the teacher knows where the student is starting and where they should end up and can measure progress along the way. Thus, good teachers almost always issue some sort of ‘initial entry testing’ or ‘diagnostic.’ This prevents too much guesswork about where the student is starting. Everything else is just guessing.
Most testing doesn’t test the skills the teacher really wants to impart. Diagnostic testing and quizzes need to continually assess the ability of the students and reinforce the lessons being taught. If the skills being taught include collaboration, then collaboration should be required in the testing.
Peer assistance is a difficult tool to manage. If it is not cultivated, encouraged and rewarded, it will not occur. The best way to create peer interaction is to assign different realms to different specialists and require them to interact to achieve a joint mission. Specialization is what drives communication in the economy. It also drives communication in the classroom.