To go Open Source or Not?
It is my dream to establish a legal informatics industry. Today, legal informatics is conducted either as an internal function or by consulting firms that specialize in long multi-year projects to build custom solutions. The few commercial products that exist are in the form of proprietory products or web services. Compared to many other forms of informatics, legal informatics has evolved very slowly. Part of the reason for this, of course, is the specialized nature of this field. This is particularly the case with legislative information where each legislature or parliament has long established traditions that are difficult to change.
As with every other informatics field, an industry will be established eventually. The costs of custom built software are simply economicially impractical in many cases, demanding a re-think about how solutions are created. For me, a key part of establishing that industry is the creation of standards. Whether there are official standards or de facto standards, standards will spur on the creation of an industry by creating a common model upon which to build. I have seen this happen in other industries that I have participated in and I don’t see why legal informatics should be any different. Yes, legal informatics is tardy in this regard, but that slowness should not discourage us from making it happen now.
So the question is quite simple. Can an open source solution form the basis of a de facto standard for legal informatics? And if so, what does that solution need to consist of? That is the question we have been wrestling with. There are two sides to this argument. While we might want to promote the establishment of an industry, at the same time we need to provide an economic incentive that will encourage businesses to participate. Could providing too much of an open source solution merely enable existing players to be more efficient, yet continue to work in relative isolation? It seems that a better outcome would be to promote the creation of interoperable products that can be mixed or matched to solve the multitude of needs in this field.
To this end, we’re trying a two pronged approach. First, we are fully supporting the establishment of official standards through bodies such as OASIS. Secondly, understanding that the official standard route is going to be a slow and perhaps arduous process, we’re pushing for the establishment of de facto standards. To this second goal, we have open-sourced our own SLIM model for legislation. It’s a very simple XML model based on 10 years experience building these types of models. While it isn’t a be-all and end-all solution, it is consistent with the current XML thinking and is quite easy to adopt. I have spent a fair amount of time wrestling with how to release it as an open source package recently. There are two questions I have:
- Which model? There are so many to choose from: Creative Commons, GNU, BSD, etc. Which model is permissive enough without discouraging commercial entities from adopting it.
- What aspects should be open source? I think it is quite clear that any and all XML information models should be open. That is in the spirit of XML, is consistent with the public domain nature of legislative information, and will allow the data to be accurately interpreted long after any particular software application of company has run its course. But would providing foundational software packages that are also open source further encourage the adoption of the model? And if that is the case, what foundation software would be beneficial?
At this point we have answered the first question and the first part of the second. We have chosen to release the SLIM XML schema as open source by using a Creative Commons Attribution-ShareAlike license. The rest of the second question remains open. What else we should provide with an open source license? Certainly it cannot be our full software suite. We are a commercial business and we need to make a living. But parts of our packages could be released to promote the adoption of SLIM as a de facto standard of sorts. What do you think?