Unfortunately, I missed Tuesday's Boston PHP User Group meeting because it was an excellent one. Our own Steve Zimmerman, along with Gabe Ricard from iMaxWebSolutions.com, Max Carlson from OpenLazlo, and Adam Peller from IBM all talked about AJAX. My colleague, Eric Shea, took notes and posted them on his blog.
Apparently the government listened to my assertion that I would only vote on an open source electronic voting system. This article reports on hearings within the California State Senate about using open source software in election systems. The article also points to several large companies that are actively using open source:
A number of private businesses, including Bank of America, Amazon.com, America Online (AOL), DreamWorks, Charles Schwab, IBM, and Merrill Lynch, have begun using open source software for some applications. Furthermore, the Department of Defense, the State of Massachusetts, and the California Air Resources Board have begun to migrate some of their computer systems from proprietary to open source software.
The Joomla! project just announced the launch of their new Developer Portal. Joomla! already has the Joomla! Forge site which is a SourceForge for Joomla! add-ons and a very well managed and helpful user forum. The goal of the Developer Portal is to be a destination for the Joomla! community. Content includes the blogs of the core development team, the most recent forum posts, working group outlines, documentation, and IRC chats.
In his summary of the SIIA Information Industry Summit, John Blossom makes some interesting observations about the future of the publishing industry. His ideas were largely based by the presence of Craigslist CEO, Jim Buckmaster, and his measured customer focused approach to the content industry that appears to be so successful. Some of the points that John makes that I think are directly relevant to open source.
"Online publishing is swinging the base of the content industry back to small to medium businesses." I think that content is distribution is shifting focus from creating a content destination to a syndication model where you have many publishers that are aggregated in different ways. Using an open source CMS, many of which have syndication support built in, it is inexpensive to establish enough infrastructure to become a node on the content network. Success is determined by the quality of the content rather than the scale of the source. The extreme of this is a simple blog.
"Getting technology right is important, but it's not as important as building trust with your users. " To support this point the John talks about Craigslist with its primitive layout and simplicity. To quote further: "Identifying a core need and filling it in a way that people really like and trust takes just a smidgen of carefully developed technology in many instances." Identifying that core need often requires experimentation and agility that precludes making big bets on technology.
One of the key values of open source is that it is inexpensive to try an idea by putting just enough technology in place test for business value. If the idea is successful, the solution can be scaled and evolved to support additional requirements as they surface. Putting in commercial technologies that are targeted for feature focused "enterprise software selection" diverts attention from the original business idea to the elaborateness of the tools. I think this is a lesson that Web 2.0 has learned from the Internet bubble.
Rocky Burt wrote a nice post on how Plone development is converging with the new Zope 3 architectural patterns through the help of the Five Project. If you are just coming up to speed, Zope 3 is a major change from Zope 2.x and introduces some modern architectual patterns such as content views in addition to incorporating some critical content management functionality from CMF (Content Management Framework, an add-on product on top of Zope) into the core. Here Rocky's summary of the benefits and risks:
Benefits
Of course further development, refactoring, and refining the base architecture generally provides a more robust solution to such an extent that businesses have to spend less time and effort managing and supporting deployed solutions. Having reusable components means less time spent developing new components and more time spent plugging in existing components which proves to take less time and thus less expense.Risks
Migration of existing sites can get quite complicated. Also having to deal with switching frameworks such as making it difficult to maintain backwards compatibility with third-party products. Of course this gets easier as more and more third-party products adopt the same Zope 3 develop techniques which means education of this process moving forward is vitally important.
In the end, he concludes that committing to the Zope 3 architecture and development style is the right approach. I would tend to agree.
I just got around to registering for the Plone Symposium in New Orleans March 8-10. It looks like an excellent program of talks and tutorials. Here are the sessions that I am thinking about attending.
Best Practices: Plone Development
Make Plone Go Fast
Enterprise Search
Introducing CMF 2.0
Open Source in Construction: Digital Plan Room
Mechanisms to Promote Free Market and Open Source Software Innovation
CASE STUDY: Fundable.Org
Applying Plone to Business
CASE STUDY: Disney, An Intranet Success
With the low airfare and hotel prices, it is hard to think of a reason not to go.
James Robertson of Step Two Designs recently published an excellent article on portals. As usual, James has a very pragmatic and even perspective in evaluating the business value of the technology. In addition to helping to define the term "portal" (I would say that over half of the projects that I have worked on that were initially described as portals were actually not portals at all.), the article reminds us that portal integration is only one click deep and is not a silver bullet to the enterprise information problem. But, if executed properly, there is business value in having a central place that summarizes information managed in different places.
One peculiarity that I did notice, was that Jame's definition of customization and personalization are the reverse of what I am used to hearing.
Personalisation allows individual users to tailor the portal for themselves (such as what information is displayed).
Customisation involves the filtering of information and tools to target identified user groups or roles (such as delivering information to specific geographic areas).
Must be a hemisphere thing.
Lars Plougman's story about a lawyer's scheme to work around a document management system points out a huge issue in content management. The natural tension between creators and consumers of content. Content creators create content for their own purposes and don't like to spend the extra effort to add metadata that is only useful to other users and systems. In fact, I often distrust voluntary metadata because, as as I mention in an earlier post, users either neglect or abuse metadata. Consumers of content, however, benefit from good metadata because they allow readers to quickly identify content that is useful to them.
It is interesting that the solution that Lars' story presents uses a third party intermediary to take the content and properly enter it into the system with the correct profile information. The more I think of it, the more I am thinking that this is a good idea and it actually maps to something that I do at Optaros.
We use Subversion as our source control system and have set up a SVN commits mailing list (more about that here). As I read the checkin emails, in addition to seeing what people check in, I can see who has been naughty and nice with their checkin comments (which are metadata!). If someone gets lazy with the comments, I give them a friendly reminder. Otherwise, it would not be until much later, when we need to roll back some code, that we would realize that the comments were not helpful (I know whoever is reading this is thankful that they don't work with me!). Usually just the knowledge that people are looking makes contributors more meticulous about their comments.
The most common way to enforce metadata is through input validation (like in the case of the lawyer's document management system). Users hate it and that is the first thing people point to when they talk about how unusable their system is. It strikes me that making content creators write their own metadata is relatively new and maybe it doesn't work. Formal publishing environments typically have authors write stories and editors compose headlines and place stories. Lawyers are usually pampered with admin staff so they historically have been removed from the mechanics of creating documents and metadata (although that is changing). The librarian profession specializes in metadata and organization. Is it unrealistic to expect content creators to manage their own metadata? Or should we return to inserting metadata specialists into the process after content creation, either as police like me with my commits list, or as metadata editors like the secretary in Lars' story. Can we afford to do that? Can we afford not to?
I am looking forward to seeing your comments.
[2/15/2008 Update: If you want a more up to date view of the marketplace, consider buying Open Source Web Content Management in Java.]
[2/25/2007 Update: The original whitepaper has been taken down from the Optaros site. It is somewhat out of date but there still seems to be a considerable amount of demand for it. You can now download it here]
[9/29/2006 Update: I have started to add updated reviews on this blog. Here is an updated and more complete review of eZ publish.]
[My apologies.... With the re-release of our website, the back door that allowed access to the white paper without registration has been closed. Fortunately, this white paper has been published under the Creative Commons 2.5 license so there are other copies floating around on the web. Maybe under Google keywords Seth Gottlieb CMS Whitepaper?. Also I have also seen some of the individual project reviews reposted on SWiK. So, if you like your information free (as in "libre", not "free weekend at a timeshare promotion") you might consider checking these alternative sites.]
I just published an epic whitepaper where I discuss how selecting an open source CMS is different than a proprietary software selection and summarize 15 open source projects:
Alfresco
Bricolage
Drupal
eZ publish
Lenya
Mambo/Joomla
MediaWiki
Midgard
OpenCMS
phpBB
Plone
Roller
Twiki
TYPO3
Zope CMF
The summaries are not to the depth that you would find in the CMS Report (which reviews Midgard, OpenCMS, Plone, and Zope) but they give a high level view of what the project is about.
Here is the abstract:
The open source community has produced a number of useful, high quality content management systems which presents an opportunity to deliver tailored content management solutions without the high licensing or management fees associated with commercially-licensed or hosted software. However, the sheer number of open source CMS projects and the ineffectualness of traditional commercial software selection techniques can make the task of finding the right open source software an intimidating challenge. The strategy of using feature matrices is particularly ill-suited to open source software selection. A more practical approach is to match your needs to a common business problem that others have solved using open source software and engage with the community to learn about their experiences in implementing the solution. Doing so will take advantage of the unique aspects of open source software: the openness of the user community and the transparency of the development process.
The content management use cases that are particularly well served by open source are: informational websites, online periodicals, collaborative workspaces, and online communities. This paper briefly describes some open source projects that have been successfully applied to support these use cases and gives techniques for how to engage with the community. While open source is frequently and successfully used as an alternative to custom development of unique solutions, the use of open source software will be the topic of another white paper or case study.
Feedback is welcome in the comments. If you want more information, feel free to contact me directly.
Jeff Potts, in his blog ECM Architect makes an excellent point about how migration tools are usually over-sold by software marketing organizations. Jeff rightly says
The *problem*” is, Notes/Domino [and I would extend this to any real CMS platform] is often used to develop complex, highly-customized applications. For those, there's probably no getting around a complete re-development effort if they are to be moved at all.
The best one can hope to do is migrate the content and success will depend on the quality of the content. If the content is well structured, it is often possible to use automated migration logic to map fields from one repository to another. If the content is just free text, the best you can do with automation is to push the content into an equally unstructured system. Of course, then your pristine new system will be as cluttered as your original system - probably more-so since much of the organization that was implied through navigation will be lost.
Not to scare you, but if you are switching platforms, be prepared for the cost of migrating your data, as well as training and loss of productivity. James Robertson of Step Two sets some realistic expectations in his article Spending Patterns During CMS Implementation. On the bright side, this work often forces critical business decisions and activities that directly improve the way content is managed within an organization. For example, deciding how long to keep around content, eliminating redundant or outdated content, organizing content more effectively, establishing ownership and processes for content.... If you want to be successful in this initiative, you would do this work anyway. Otherwise many of the problems that hampered the original system would be ported to the new system where they would be equally problematic.