# 2i2c Roadmapping Q4

Below are the notes and major takeaways from the 2i2c team's Q4 roadmapping and planning meeting. This meeting is meant to help us zoom out a bit, center ourselves on our mission, and decide where to focus energy for the coming months. Because this was our first meeting of this nature, it is a bit more high-level than we expect it to be in the future.

These notes are broken down into three major sections. For each we provide a short summary along with raw notes from the conversation.

## 2-3 key communities that we are serving

### Summary

There were a few major themes that came out of this conversation, described below.

While we discussed many kinds of communities in research and education, there was a thread that seemingly tied them all together. 2i2c's team wishes to support communities that **create and share knowledge for the public good**. This feels like a good parent category for all of the communities that we serve, though we discussed that there is a lot of diversity within this category as well.

In addition, we wish to support **communities that cross boundaries** by providing them infrastructure that facilitates connection, collaboration, and sharing. For example, research communities are cross-institution and cross-nation. Many of their institutions provide infrastructure that creates barriers to easily sharing work outside of their local groups. 2i2c can help provide institution-agnostic spaces for collaboration and research via the cloud.

Finally, there was much enthusiasm around **supporting communities that are historically marginalized**. Many are interested in participating in global networks for knowledge creation/sharing, but have barriers to doing this on an international level. For example, enterprise offerings tend to offer products geared towards larger, wealthier institutions. This leaves out smaller institutions like community colleges, as well as most organizations outside of countries that have a lot of wealth.

### Raw notes

- Communities in the educational sector
- We know that they lack capabilities that JupyterHub can provide
- Companies (in the data science space) do not tend to offer things that are matched to what universities want
- Different types of educational groups
- Universities - have more resources in general
- In smaller / marginalized educational institutions, they have fewer resources, and are thus not represented as much in the broader conversation.
- Communities that focus on sharing knowledge for the public good
- Educators and researchers in universities is a big overlap
- Educators and researchers may be the same person - e.g., using a hub for a class and also use it for research
- International communities that span institutional or national boundaries
- Infrastructure in neutral spaces can help break down barriers
- Diverse communities that collaborate across research field
- Big problems are multi-scale, multi-sector, etc
- Marginalized organizations in general
- They have more ground to make up to gain the same power, influence, etc as the larger universities etc
- 2i2c can accelerate these organizations to help balance resources across these institutions, provide a boost to organizations that might be under-resourced historically.

## 2i2c's goals for the Managed JupyterHub Service

After centering the conversation around key communities that we wish to serve, we discussed how our Managed JupyterHubs service, and our organization more generally, can best-serve these groups. Below are a few key takeaways.

### Summary

First, it is clear that the **Right to Replicate** resonates strongly with everybody, and we should lean into more. This means that we should center the R2R in our technology development and design so that it's clear that it is a major influence on our design, not just a "nice to have".

We also discussed
- Transparency at all levels (not just open code, but also open discussions, public strategy and plans, etc)
- Creating with others (modularity, upstreaming, governance, etc)
- Balancing platforms vs. services

### Raw notes

- Co-production of information, with an invitation for communities to move from user -> participant -> co-leader
- We want to prevent the loss or corporatization of knowledge
- Break up knowledge silos, make it easier to discover, share knowledge
- Keeping track of the communities we serve
- Most organizations are used to keeping track of users and understanding where they're coming from
- Why not just rely on companies to solve this for us?
- Companies can ultimately do what they wish, can turn off services etc
- Right to Replicate - providing a pathway for communities to run their own infrastructure if needed
- Transparency
- Where are the places that companies are intentionally not transparent? Can 2i2c lean into those aspects of transparency to differentiate itself?
- Commitment to modularity and "right to replicate" even if 2i2c goes away
- Build a platform for open science
- Demonstrate why the community-based approach is the right approach for our values and the communities we serve
- Tension between products and services
- Something like mybinder.org is a product
- Something like "run a JupyterHub for a community" is a service
- Products tend to be "many users who don't know us" - less of a "collaboration", puts users into "consumer" buckets
- Example of a product
- GitHub
- Example of a service
- Old-school GitLab
- Atlassian (local install of bitbucket)
- Tension between high-touch (but more expensive, less-scalable), low-touch (but more producty, less collaborative w/ users)

## What are major focus areas that will have a lot of impact?

### Summary

- Linking identities with other services
- Leading community efforts to build unified experiences
- Building confidence in our infrastructure

### Raw notes

- Credentials management
- https://github.com/jupyterlab/jupyterlab-git/issues/1044
- Would allow 2i2c services to integrate with pre-existing services
- "Link my account with XXX"
- Examples of things to interoperate with
- GitHub
- Globus
- Is this something we can provide leadership on in the community?
- Help lead jupyter++ communities towards the "platform for open science" vision
- We want a platform to be modular and composable, and largely driven by the community
- We can help coordinating and guiding stakeholders in that community
- Working with these communities to develop a strategy around a unified experience for open science
- Position ourselves to also _execute_ on this strategy
- It is hard to coordinate in these multi-stakeholder communities, can we put efforts into this?
- Authentication via university log-in
- Could allow us to integrate with the other services that universities provide
- What authentication services feed into the same vision and values of 2i2c? (cross-organization, agnostic to vendor, etc)
- ORCID?
- How can we group users if the authentication we use does not have a notion of "groups" already?
- Could we have a "2i2c identity"?
- What ties together all the different hubs, services, etc?
- Users could still have flexibiltiy to use different auth providers, but there'd be one database that hubs pull users from
- Centralized space on the web for a user that is independent of any one hub
- Link to other services (e.g., GitHub, ORCID, etc) at this central place
- Would be a big technological investment, potentially a slippery slope
- Would communities feel comfortable centralizing their identity on a single organization like 2i2c?
- Could this be a "linking identity" problem rather than a "centralizing identity" problem?
- I have an identity in a JupyterHub, I have an identity elsewhere (e.g., GitHub), how can I quickly link them?
- Another pathway: build confidence in our infrastructure
- Automation, testing, etc
- Reduce complexity of our infrastructure
- A key issue for this: https://github.com/2i2c-org/pilot-hubs/issues/502
- Important for maintenance burden
- Important for right to replicate