Resolving the Design

This week I started turning my drawings into what Looking Glass would look like with the designs, since this sometimes helps me figure out whether pieces of the design are realistic. I also started thinking about the implementation and what the technical process for each of the steps would be. I think some parts of this may be complex, such as saving the code modifications and rule revision histories. It's important that the structure of this system is carefully designed, since there are many interconnecting complex pieces. 

I also continued thinking about how to test the specificity of a rule. While it would be preferable to be able to remove constraints from a rule individually to check if a rule is too specific, I'm not sure that this is realistic for our implementation. However, if mentors have selected lines of code in the rule-writing process that should be removed in the suggestion, it is possible to create a basic rule for each suggestion that just checks whether a world has some number of matching lines of code to the ones that should be removed. This number may be variable or could also be changed by the the mentor to check if worlds with a high number of matching lines fit the rule. 

Reading Suggestbot: Using intelligent task routing to help people find work in wikipedia gave me more to think about in deciding which suggestions to show mentors. Cosley et al. found that certain suggestions for wikipedia articles to edit did increase wikipedia editing participation. It seems like the "intelligent task routing" succeeded most when the tasks suggested were closely related to previous tasks the users had completed. For the community mentoring, where mentors write rules for suggestions that already exist, there may be two groups: those who are also mentoring specific children, and those who do not have specific children they are mentoring. I think these two groups will probably have different desires. For example, if a mentor has just rated a set of suggestions for their child, they might be more motivated to edit the rules for those suggestions, especially if they downvoted a rule. However, for mentors not working with a specific child, it may be more important to show them suggestions similar to those they wrote rules about in the past, so that they can create communities around types of rules and become "experts" of sorts about a certain type of suggestion, like animations or custom procedures.