Mapping the Research Software Ecosytem

Karthik Ram

James Howison

Use keyboard arrows or touch gestures to navigate slide deck

UC Berkeley

UT Austin

Research software isn't a creditable research activity


Howison & Bullard 2016

Formal citations: 31% - 43% 

Informal mentions are the norm, even in high impact journals

Software is frequently inaccessible (15 - 29%)

 Lack of visibility means that incentives to produce high-quality, widely shared, and collaboratively developed software are lacking

Prior work

Software Heritage

World of Code

Scientific Software Network Map

Transitive Credit


Heather Piwowar & Jason Priem

Andrew Nesbitt

Roberto di Cosmo

Chris Bogart

Jim Herbsleb

Daniel S. Katz & Arfon Smith


Git history
Informal mention in text
Informal mention in text

Paper text & metadata yield data on topic, field, software use

Package managers yield data on dependencies

Git repositories yield data on history of development


Mapping the research software ecosystem

Which packages are increasingly used together in scientific workflows?

How might the map assist with how you know and interact with your project’s upstream and downstream dependencies?

Can funding help make software more compatible?


What groups of interdependent software are increasingly important for scientific fields?

How visible is their importance?

Can directed funding ensure the stability and maturity of critical dependencies and tool networks?

Does indirect usage do the work needed to demonstrate impact (with funders, with evaluators?)


Which software components are seeing use outside their areas of original development?

Can funded interventions shore up interdisciplinary opportunities?

Are there “leading” and “lagging” fields? Can funded interventions bring lessons in achieving change within fields?


How can we assess the weaknesses and opportunities in the ecosystem? Can project health data (like the CHAOSS project) be integrated to highlight strengths and weaknesses?

Which fields appear to be lagging? Can funded interventions bring lessons in achieving change?


Can visibility of interdependencies motivate industry to provide pro-bono support to those building software crucial to science?


What else do you want to do at an ecosystem level that this wouldn’t help with?


E.g., Might we learn about how different structures of dependencies (e.g., proper hierarchies, hour-glass structures) affect the efficient flow of limited labor in science for bug reports, fixes, and improvements?