If Pytheas did eventually see the tin mines in Britain, there might have been a point after his initial wonder that he thanked the stars he wasn’t one of the workers who toiled to extract it. The backbreaking labor that went into retrieving tin may have seemed remote when admiring the latest creation of the tinsmiths in Massalia, but Pytheas no doubt appreciated that sooner or later, someone needs to do a lot of hard work to get the process started. Programming is undoubtedly much more rewarding than mining for tin was in 300 BC, but it is probably as badly understood today as tin was to the ancient Greeks.
The evidence of our inability to properly plan and put together computer applications is everywhere in the IT world; constant delays in release dates, budget problems, and never-ending beta processes seem to be the hallmarks of application development. Libraries tend to be somewhat removed from the realities of the software development cycle and it probably is no surprise that developers are a little rare in the library world.
Creating a system can be no less ambitious than extracting tin before the invention of large-scale mining equipment. Linus Torvalds was able to offer enough existing and new building blocks to catch the attention of a ready and willing worldwide audience to put together one of the most impressive collaborative projects in the history of humanity. An Open Source library system has to appeal to a much smaller audience and one that may not be as enabled to build momentum in the way that Linux has come together. An entire community of developers may be galvanized to build the world's next great operating system or spreadsheet, or almost anything else that Microsoft or Oracle actively markets and profits from, but far fewer developers are interested in creating an alternative cataloguing editor or library circulation system. To address this, the PYTHEAS system had utilized as many freely available building blocks as possible but the number of functions an ILS is expected to supply looked daunting when compared to the developers available for the task.
The popularity of XML and the amount of work done on metadata standards make some initial steps look more achievable. The Dublin Core/MARC/GILS Crosswalk16 supplies a straightforward way of mapping MARC to a generally understood metadata format that can form the basis of searching, and the NACO normalization rules17 provide a consistent approach to mapping queries to relational database layouts. Using the Resource Description Framework (RDF)18, which is a standard for describing metadata in XML, records could be described and enhanced in ways that might be difficult in MARC, such as adding annotations and reviews to bibliographic records. Castor makes XML easy to produce and work with at almost every working level, so that XML can be used to “wire in” many other applications (see illustration below).
A system that requires a million lines of code is probably as unworkable to libraries as mining tin without a large group of willing miners, but XML and java-based technologies like EJB open the door to leveraging the efforts of a worldwide community to the task at hand. Phoenician tin was often overpriced and of inferior quality because it went through a series of suppliers on the way to market. Library systems built on web-related standards and close to a rich source like XML may be able to use technologies that have not yet lost their luster.