Archive for 23rd March 2010

Community Source – Universities Building Open Source Software (Book Excerpt)

Copyright, Charles Severance 2010 – All Rights Reserved
From time to time I will put up draft excerpts of the book I am writing about Sakai. Feel free to comment and/or correct where I am mistaken by sending me E-Mail.

The Sakai project was formed in a moment of transition between a hierarchical / centrally controlled approach in campus information technology infrastructure and an organic / distributed approach to coordinating across a community of like-minded individuals and organizations. As a result, the Sakai effort has always been at the boundary between old-school approaches and new-age approaches to technology development and deployment. At times Sakai has achieved great success through a blend of old-school and new-age approaches and at other times operating at that boundary has led to great long-lasting conflict and stresses in the community.

By the year 2000, the concept of open source as an approach to software development was well established with the Linux operating system and Apache Foundation projects as solid sustainable examples. These efforts collected the talents of volunteer developers from around the world with relatively loose leadership and a commitment to working together to solve a common need. Generally the developers who worked in these projects fell into one of several categories: (1) volunteers who had paying day jobs who worked on the software in their spare time, (2) consultants or small companies who made their living doing consulting on the open source software and gained competitive advantage from their involvement in the project, or (3) individuals hired by large companies such as IBM who were given release time to contribute to these projects to support the projects and insure that the company had a voice in the projects going forward.

Many universities used open source software throughout their enterprise since the early 1990’s. Open source software was ideal or University use because it was low-cost and allowed University technology staff to make small changes to the software where some particular or unique feature was needed. Open Source gave Universities a sense of “control” over their destiny and future costs when it came to their information technology solutions. Open source software also allowed a certain level of “agility” as technology needs shifted throughout the 1990’s as things like the Internet and World Wide Web became part of the required campus information technology suite of services.

However, few universities were regular contributors to open source efforts. Universities typically felt that the software and other intellectual property produced by their employees had potential value and if a staff member built something that was valuable, then the university wanted to profit from that creation. After all, the university had paid the person’s salary while they were doing the work. It made perfect sense to a university administrator or attorney but was very frustrating to individual university employees who yearned to work with their colleagues around the world as part of a community effort.

This led to Universities writing a lot of software for their own use but not sharing that software with other universities unless there was some profit to be made on the interaction. And because no University was willing to invest the time and staff in making their software commercial-quality, most University-developed software was “just barely good enough” for local use.

One of the most common examples of “locally developed” software in the late 1990’s in use at Universities was the campus-wide Learning Management System. Learning Management Systems were pretty basic software and allowed instructors to distribute materials for students and interact using e-mail, chat, or threaded discussion forums with the students. These systems were simple enough that it only took a small amount of resources to get a basic system developed, up and running with a team of 1-2 developers in with less than a year of effort. Often the efforts were done “on the side” or “below the radar” of the typical campus IT operations.

In some cases these university-developed course management systems developed to the point where were purchased and turned into today’s commercial Learning Management Systems. The WebCT commercial LMS product was based on software developed at the University of British Columbia in 1995. The initial Blackboard product was based on a system developed at Cornell University in 1997. The ANGEL Learning system was created in 2000 based on technology developed at Indiana University-Purdue University at Indianapolis (IUPUI). The Prometheus system was developed at George Washington University and later purchased by Blackboard in 2002.

Often the universities would make some money in these transactions, but the real winners were the companies that took the software, cleaned it up and began to sell it to all of the other universities who were growing tired of their local home-grown systems. These companies started building market share and applying economies of scale to their software development. In time these companies began merging and buying one another to become ever larger and more pervasive in the marketplace. At the time of this writing, Blackboard has purchased Prometheus, WebCT, and ANGEL resulting in a very large market share.

??? Did D2L come from McGill University ?? When/How ??

Stanford University developed CourseWork system in 2001 and began to share the software with other Universities around the world. Also in In 2001, the Moodle project also starts with a simple LMS system with an open source license. The MIT Open Knowledge Initiative (OKI) was a project funded by the Andrew W. Mellon Foundation in 2001 to try to bring order to the chaos of so many independent LMS projects and so many divergent Learning Management Systems at so many Universities. Other projects such as Boddington at the University of Leeds, OLAT at the University of Zurich and CHEF from the University of Michigan were pursuing an open source approach and tried to convince other schools that their solutions were something that could be adopted.

From 2001 through 2003, the MIT OKI project regularly brought together many of the leading thinkers about LMS systems and technology from Universities around the world. The OKI meetings and discussions began to form a community of technical staff who slowly started to know one another and realized that even though they worked at many different organizations, that they were all facing the same problems and challenges.

As the OKI project funding was ending in 2003, several of the participants in the OKI efforts decided that perhaps they should band together to form a consortium and work more closely together to develop and release a shared Learning Management System that they would all work on collectively and all use in production at their institutions. By pooling resources, the partners would gain much greater leverage and each school would not have to solve the entire software development, testing, and maintenance tasks.

The goal of the Sakai project was to take the “best of breed” of the current university-developed learning management systems and produce one system that included the best of each of the systems. As a key founding principle, the Sakai project was going to operate on open source principles and every school that received Sakai funding was required to agree to give away all their rights to commercial gain from the software that they produced as part of the Sakai project.

Demanding these open source principles was quite necessary because university adopters of “free” software had see the pattern more than once where a piece of software started out as a “free and collective effort” and then once it had a few customers, the university which owned the software sold it a commercial company along with the customers who had adopted the software. The university that had originally written the program typically made some money and was giving the right to use the software forever. but the adopting schools were given no such deal. They were usually forced to pay the new owner of their software to continue to use it.

So Sakai was to be owned by no university – it was to be owned by the collective. That way all the participants in the project could be assured that the software would stay free forever and that no school would profit from participation in Sakai by selling the adopters of the software to a commercial vendor.

The University of Michigan was selected as the lead institution for the Sakai project and the Principle Investigator for the Andrew F. Mellon Foundation grant was Joseph Hardin and I was to the the Chief Architect for the project. The three other partner schools were Indiana University, MIT, and Stanford University. All the schools had a very strong track record for leadership in software for teaching and learning. The Sakai project also included the uPortal project as well as continued funding for the OKI project.

As a condition of being a partner in the project, each school was required to sign the agreement that they would forgo any commercial gain from the software developed as part of the Sakai project. This agreement was relatively easy for the University of Michigan and Indiana University to sign, but both Stanford and MIT had made significant revenue from licensing software over the years so it was a pretty impressive that the decided to agree to the terms and join the project. There was a fifth school who was considered as a potential partner who wanted a few weasel words put into their contract terms for the intellectual property. We just said “no thank you” and went ahead with the four core schools. A four-way split would be more lucrative than a five-way split so there was little reason to compromise the core principe of giving away the intellectual property forever.

A key element of the Sakai proposal was the notion of “community source”. We wanted to operate using open source principles but with more central coordination than was typical in open source projects. The idea was that each of the four schools would contribute five staff members to a central pool of talent that would be centrally managed in order to build a coherent product that could meet the collective needs of all for schools.

The combination of the outstanding team of schools, the community source approach, and the fact that it was a good time to try to build cross-school coordination in the area of Learning Management Systems led to the Andrew F. Mellon Foundation awarding the University of Michigan and its partners $2.4 million over a two-year period starting in january 2004 to build the “one open source LMS” that could bring the fragmented higher education market with each building its own independent LMS system together.

The plan seemed simple enough and almost certain to succeed.

Copyright, Charles Severance 2010 – All Rights Reserved