How to enable and sustain thriving Open Source Ecosystems (OSE)
Karthik Ram
Use arrows or touch gestures to navigate slide deck
1. The Research software landscape has come a long way...
Software impacts all modern research
Compared to a decade ago
it is much easier to develop software, find support & community, and not explain what an RSE means.
But things are far from ideal
i
It's still very hard to demonstrate the impact of software
JOSS is great but it is a stop gap
An SBOM might be more valuable
Especially when it allows us to assign transitive credit
ii
There is a lot of support for creating software, but very little to sustain them after reaching product market fit
Analysis code
Prototype SW
Research SWI
Ones that find product-market fit
A few with generalities make it over to a prototype
Software projects that need to think about sustainability
Analysis code
Prototype SW
Research SWI
2. ecosystem versus Ecosystem
e is the flow of
information through
dependencies
Local ecosystem
Dependency based
Data/Information flow based
An opportunity to
coordinate roadmaps,
resources, form a
collection
Can distinguish between a paper using the software versus announcing it
Counts mentions + context
We can construct a knowledge graph out of these data
Extracted software entities + dependency graph
the various actors (people),
and institutions (such as host
organizations, funders, and
infrastructure providers) that
enable you to function
Big ecosystem
E are the people and institutions that support your technical work
Bottleneck could be bad governance
e & E
need to be in harmony to achieve sustainability
3. There is a lot of momentum around sustainability
What does sustainability mean in the context of software?
Software you use today will be available - and continue to be improved and supported - in the future”
UK SSI
Sustainability refers to “qualities of availability, extensibility, and the maintainability of the software”
Venters et al 2014
Software is sustainable as long as the people behind it have the resources to continue fulfilling its mission
My definition
So what are the various dimensions of software sustainability?
5 areas
Org structure,
Governance,
Business Perspectives,
Community,
Project health
4. You need to understand your community
Yes, it is an overextended term
Think about who you support
Users, contributors, or both
High user growth
Low user growth
High contributor growth
Low contributor growth
FEDERATION
CLUB
STADIUM
TOY
In the early days, our core team built software for researchers
Early days of engaging with our community
Then we began accepting contributions. But 1:1 took a lot of time & effort
Inbound Contributions
The rOpenSci Dev Guide
Best practices for software development documented in a living book
Created a system to peer-review software
Creating a process for standardized peer-review
Became inspiration for The Journal of Open Source Software & PyOpenSci
Automation
+
human review
Ignores conversations between humans
Jump in when called
Perform <actions>
return <output>
Enabling others to set up & host their own communities
Built r-universe, a system that allows any community to easily set up their own rOpenSci
r-universe allows other COPs to have their own infrastructure
r-universe is like a
combination of a
journal + preprint service
CSCCE Collaboration Model
Convey, Consume
Contribute
Collaborate
Co-Create
Theory of Change
Open source was hard
UX became better
COPs emerged
Credit e.g. JOSS
OSPOs & Open Source Policy
Metrics from trace data can help COPs
CHAOSS Metrics
CHAOSS Metrics
Success
Sustainability
Risk
Health
# Contributors, Users, Developers, Diversity, Growth and social diversity, openness index…
Quality model, Development base, size of niche, financial resources, resilience, relationships among people
Bus factor, Truck factor, Elephant factor…
Social health, Gender bias, Positive experience, Robustness, Productivity
To date, there aren't any canonical set of metrics that are characteristic of open source project health
Users
Contributors
Maintainers
5. Security considerations are important
Can be technical or social threats
Most technical risks can be managed
The rest are social like governance
When all else fails
sunset a project safely
and responsibly
Coordinate efforts
If you're a COP
Support your community in identifying bottlenecks that get in the way of growth and sustainability
Find your bottlenecks
Ecosystem level support
bit.ly/csv2023
How to enable and sustain thriving Open Source Ecosystems (OSE) Karthik Ram