wiki:org/meetings/2015WinterDevMeeting/Notes/SupportingResearchCommunity

Incentives are misaligned for researchers to make their code usable and see their work through to deployment, because there is little "academic credit" beyond publishing the work in an academic conference/workshop. With this is mind, Tor should try to set up the situation so that when researchers help themselves by working on Tor research projects, they are also helping Tor.

Some things that Tor can do immediately:

  • Keep a wiki page with important research problems ranked in some priority order, so that researchers that may be unfamiliar with the broader Tor research world have a better chance to choose the projects that Tor actually cares about. (Currently this info is on the web and very stale.) Perhaps have a way for groups to indicate they are working on some problems, to prevent duplicate effort?
  • Create a best-practices / FAQ page that explains to researchers what they should do to maximize the usefulness of their code. For example, what Tor version should they fork, they should submit a git url and not a random tarball of code, etc.

Longer term strategy:

  • Each Tor dev team (approximately) should have a point-of-contact that will answer questions by researchers working on research related to that project. The POC should be a go-to person / mentor to make sure the researcher is working on the right problem in a way that Tor would best be able to make use of the results.
  • Hire a new "Director of Research" (or multiple people for the research team) to fulfill the above role for many projects.
  • Figure out a process for triaging incoming Tor patches and research results: this involves (1) figuring out if the research result is something that Tor should actually do (it should be if the POC/mentor was involved from the beginning); and (2) get the patch or contribution in a state that would be acceptable for merge (identify what needs to be done to get to an acceptable state).
  • Someone from the research team should travel around to universities and research labs and explain Tor and all of the hard open research problems. This will increase awareness of the important problems Tor wants to have solved, and will hopefully help attract new students to work on Tor-related research.

Other Good Ideas to better align the incentives for researchers so their work is more useful:

  • Tor should create a "Tor Best Research Award" or awards (best research, best student research, best research resulting in deployment, etc.) and give it to, e.g., one paper a year. This would be a way for Tor to signal researchers which papers are good examples of doing research the way that Tor likes, and would hopefully get researchers to do care more about getting their work in a state that is useful to Tor if they can win an award for doing so.
  • The PETS research community might try to create a new workshop that specifically invites submissions for papers about issues that arise during deployment of research. Submissions would be like case-studies in research deployment and include lessons learned, etc. The idea is to enable researchers to get another publication from the extra work that goes into deployment, so that they can get some "academic credit" for this extra work.
Last modified 3 years ago Last modified on Mar 3, 2015, 9:29:47 AM