I’m in DC for the annual fall meeting of the Coalition for Networked Information. This time the opening plenary is a discussion moderated by Cliff Lynch, CNI’s Executive Director, and including Tom Cramer (Chief Technology Strategist, Stanford University Libraries), Michelle Kimpton (Chief Executive Officer, DuraSpace), and James Hilton (Dean of Libraries & Vice Provost for Digital Educational Initiatives, University of Michigan).
Cliff – Notable successes in people launching community source projects over the last 10-15 years. But the landscape is changing: economic model, speed of development are looking shakey, accellerated move to single or multi-tennant arrangements run elsewhere. Where does this leave you when you come to the point in the lifecycle when you need to think about new systems? How do we engage with new opportunities in the MOOC and Unizen space?
Community source – what is it and does it really have a future?
James Hilton – Community source not going away. Is community source the same as open source? Open source often used synonymous with developer-autonomous centric. Kuali as compared with Sakai. How do you organize the labor that produces an outcome? We have many more tools to tune development – different organizational models can work.
Michele Kimpton (Durapsace) – Tuning of community development model. If you want to collaborate and develop code together, that’s a community model. Code doesn’t advance if people are doing customization at their institutions themselves. Need to invest and be transparent to advance the code base.
Tom Cramer – Many forms of community – one form is to have a centralized organization, but just as many examples where the community is grass-roots driven from the edges, like Fedora.
Is there a trend taking us towards or away from the grass roots model to funding a central model?
Tom Cramer – Examples on both sides. Central organization can bring focus, but so can grassroots – e.g. BlackLight faceted browser for SOLR.
James H – as scale of investment goes up the pressure to organize and centralize goes up.
Is the presence of serious commercial players a factor in central vs. grass-roots?
Tom Cramer – if there’s an absence of commercial players that can buy space and time for grass-roots organizing. Central authority can make missteps, whether community or commercial.
James – Unizen is trying to organize community effort around content and analytics standards. Made a decision to adopt commercial software – in part because they wanted the speed that came with that. Contingent on contracts giving the control needed.
Michele Kimpton – Two models – when commercial entity makes product open-source that gives an exit strategy, but it’s not community controlled. Really serving core paying customers.
Tom Cramer – Community source projects have failed where they’ve been gated communities – fail to channel the interests outside the gates. Also true of vendor solutions – unless you can tap the bigger market it will be a problem.
James – Unizen focus is on creating relays that will be as agnostic as possible. Community development is in building workflows using repositories, not in refining the LMS. It’s not about software, it’s about business and economic models.
Cliff – move to talk about the move of software from local to redundant network hosting. There seems to be a big move in that direction. Seeing what would have been community source before now taking on character as community service – like DPN, APT, etc. How does that change the landscape?
James – makes you ask the question – what parts do I need to control, what do I not need? In Unizen trying to figure out what parts need control. The LMS is core infrastructure – go for economy of scale. Focus control on building digital workflows, helping humanists and research scientists know where stuff goes.
Michele Kimpton – 1700 institutions running DSpace – difficult to upgrade to new releases. Duraspace wanted to provide pathway for smaller institutions to run latest code. Cloud infrastructure will flip IT in academic environments on its head. Will be hard to justify building data centers when they can buy IT as a service and buy only what they need. Can keep the same governing process and openness.
Tom Cramer – Running data center and installing and maintaining software is not the core competency. It’s higher up in the stack providing value to the community. Where do you want to maintain control? Curation, discovery, preservation.
Cliff – A lot of this software is getting big, volatile, and complex enough (especially in the security environment) doing maintenance and configuration management is getting to be troublesome. But if you’re out in the cloud you put need to do version control and validation – is that a worthwhile tradeoff?
James – if you’re committed to running everything in this compliance environment, that is all you will do. What to we value as academic institutions? What do we bring that’s unique?
CLiff – Barrier to innovation is everybody forking off code and doing local adaptations. Sense is in the future with networked software as service that area of variation really goes down. Can diffuse innovations faster.
Tom Cramer – Perhaps getting better at managing diversity. Seen lots of good examples that different communities are good at putting enhancements back into the code base. Separate question than running software as a service. In commercial world looking at securing different layers and diffusing innovation at those layers.
Michele Kimpton – There has been a lot of customization of both DSpace and Fedora, and that leads to frustration in upgrading. But customizations are needed. Part of the beauty of more innovation is you can look at aggregations across instances in the cloud – e.g. how do we aggregate pushing content into DPLA or DPN? Easier to do from cloud to cloud.
Tom Cramer – It’s standardization that enables that, not just cloud. e.g. standardized APIs.
Cliff – standards – are the places where standards are most applicable changing? Used to be notion of standards that allowed replacement of building blocks within a system. Now that you move into a world of aggregated things standards don’t mean as much – may work just fine to be expedient.
James – challenge is how do you move standards at the pace of technology?
Tom Cramer – role for standards based on size of the pool you want to swim in. There are important communities of practice around loose coupling whether informal or formal. Look at the numbers of people using SOLR for searching.
Cliff – Puzzle about how patterns of innovation change. Community source projects from grassroots where there is considerable technical expertise at the participating instituitons. If we think about collective service-based aggregations do local technical experts become scarcer and does that imply less diversity of innovation?
James – if we can move innovation up the stack life gets better.
Tom Cramer – you don’t need to know how to run a server to have technical expertise. Successful solutions will figure out way to tap innovation coming from the edges. Be the community you want to be.
Cliff – you can look back and see the evolution – used to be many organizations that had huge knowledge of global networking, but now it’s held in fewer insittutions.
Michele Kimpton – if the developer can focus on developing and not setting up server and talking to IT, it increases innovation. Can throw things up and see if they work? Capital costs to innovation are so much lower. That’s why in the commercial space you see cloud-based services spawning all over the place.
Discussion of contracting and procurement – the legal folks have the same challenge we do in figuring where we really need to be unique flowers. We all have indemnification and state rules. We don’t need 50 different ways to say it.