URSSI and Research Software Engineers


#1

Many universities (and national labs) have people in staff positions who are focused primarily on software development rather than leading research. They go by different names, but Research Software Engineers (RSEs) is gaining acceptance in many circles.

One question that I think has to be answered for URSSI is what is their relationship with the RSE community. I can imagine several possibilities:

  1. URSSI employs RSEs or in some way provides such services to the community.
  2. URSSI works to promote and support the establishment of RSE capabilities at universities around the US. Development “evidence banks” to help make the case, to understaind different funding and organizational models, etc. Advice on how to talk to university administrators (and who) about this, etc.

I don’t think these have to be mutually exclusive, but there are some tensions here. Some universities already have RSE teams, others don’t. If URSSI provides RSE services, how are requests prioritized? Do projects at institutions without local RSEs get priority? What if the administrators there have repeatedly refused to establish their own (after all, URSSI is offering these services for “free”)? Can URSSI partner with existing RSE capabilities to provide broader coverage? How does that interact with funding models that RSE teams are currently using?

How much can URSSI afford to invest in the RSE community, and how can it get the biggest bang for that investment?


#2

Thanks David, this is an interesting topic.

Your item 2 fits into the policy category that an institute might have, and is also what the SSI has done. I personally would very much like an institute to support this and similar policy activities.

I’m much less sure about item 1. In part, this is what XSEDE ECSS does for XSEDE-related projects, and so we might learn from how they make selections and deal with the tensions you mention. On the other hand, it’s very unclear to me how much an institute could put into “development” activities like this, if this could be enough to make a significant difference to the community, and what alternatives might be more scalable.

Dan


#3

I agree that (2) is more scalable. However many people at the Berkeley workshop seemed to envision a fair amount of (1). So it would be good for some of them to add their thoughts to the discussion.


#4

Hi All,

I believe that a differentiator between software engineers and research software engineers (RSE) is that RSE’s need to know the research process and its work products (papers, presentations, proposals). In my experience, RSE’s have to tailor their development activities to meet the larger goals of the research projects they serve. In general, RSE’s start out as discipline-specific researchers who specialize in software or as software engineers who are brought onto a research project. The majority of the RSE’s I know are in the first category and I think their familiarity with the research process may be a factor.

Thanks,
Dave


#5

And, to follow on, distilling information about the research process necessary for an RSE may fit into the second category of activities for an URSSI.
-Dave