 |
 |
[For the record, copied from RDA-LIST, Nov. 4] After the new master plan had been publicized, I've had exchanges with various people about it. Mac referred to parts of this. Enthusiasm seems to be buildung up only very slowly, if at all... A plan of this caliber ought to make a real splash in the community. This is not just any old paper but a highly important one of potentially far-reaching ramifications and a high impact on the quality of the stuff we are working with, and thus the quality of our work, from now on into an indefinite future. We all expect this quality to improve, of course. Is this expectation justified by the Framework? For one thing: that plan puts all eggs into one basket in committing itself to Web standards like XML and RDF when, far and wide, there is no large-scale bibliographic database that serves real-life library work while being based on those. Correct me if I'm wrong. What with Linked Data and RDF, those are offsprings of the Semantic Web movement. In that arena, it is taken for granted that everything comes for free. Content standards that are not openly available will meet zero acceptance, may they use RDF or not. Of course, as was discussed yesterday, the maintenance of an open standard takes a long-term commitment. And for the data itself, what is OCLC's view on the matter of liberal access via triplestores? Now XML and RDF are not brand-new, and there certainly have been lots of attempts to employ them in a grand way, even some at very prestigeous places. Where are the success stories and the smoothly running new-age engines based on the results? I'm asking this not for the first time, but up until now got no answers in the forums. Certainly, library systems need to be able to export and import XML and RDF structures, side by side with many others. With the appropriate tools and interfaces, library catalogs need never show anybody, except those working on their upkeep, what their data looks like internally or how they communicate among each other. Even today, not every library system uses MARC internally. They just all of them are able to swallow it and spew it out. (No mean feat, I think, even today. Even something like VuFind takes in MARC and nothing else.) RDF triples in huge depositories called triplestores are static copies, they need to be frequently refreshed. Is that realistic? Will it really be useful and attractive for end-users if every library rig up their own triplestores - or should OCLC do that for all of them? Even now, OCLC could already be doing a *much* better job of letting end-users access structured data in many useful ways, XML structured and otherwise, out of the live database, not a stale copy. So: RDF is welcome as an addition, a special export product, but not suitable for internal purposes and much too clumsy for bulk communication. (JSON seems to be gaining ground now) Secondly, there is no need for there to be one and *only* one exchange standard. If some community needs some peculiar different format XYZ, there may be tools that take in MARC and serve up XYZ. On a per record or result set basis, web services can do that nicely, with no one caring what the original was looking like. If we create more and flexible standards for web services, these might solve or support most of the requirements our catalogs of the future are expected to fulfil for end-users and exchange, even with MARC inside. Web services are flexible, easily extended and modified, with no need to tinker with internal or communication structures. And the plan itself says that MARC21 should be retained as an exchange format for as long as necessary. So why not first create an alternative format, test it up and down any number of years, improve it or add yet another better design, and so on. And creating and enhancing web services standards all the while, as the *primary means* of access to library data from any outside agents. This can begin right now and it has begun in many places, so one should look at ways to coordinate and standardize some of this work. Eventually, let the market decide, let the better concept win or let it take over step by step as it gains acceptance. MARC may or may not fade away in the process, sine ira et studio. Anyway, two years to achieve "credible progress", in this field? How's that defined, BTW, how will it be determined? And what does it mean to "Demonstrate credible progress"? Which of the many aspects of format features and uses will that include? (About involvement of NISO, there's another thread in this forum) And thirdly, data input and editing may use any modern techniques available today, hiding all the ghastly stuff involved with MARC under layers and subwindows of pulldowns and radio buttons and plain language labeled input fields. No playground for RDF and XML here. Ask the vendors why they don't provide that. But don't forget to evaluate the economy of a new catalogers' interface - and what it means to have different ones on sytems A, B and C - in comparison with the universal interface everybody is used to now. If you want to move away from plain tagged editing, it becomes lots more difficult to create a standard. One reason is that interface techniques keep evolving ... Oh I forget: RDA's trouble with MARC was what led to this plan in the first place. Well, that is not MARC's fault but the one of the particular setup that was used for the test. It did not use capabilities and provisions that are in fact there in MARC, like the use of identifiers for authority headings, and record linking for multi-part resources; the part-whole relationship wasn't considered at all. The test, in short, was a much too timid and superficial exercise to base any overall judgement about "RDA in MARC" on. Or had the test, to begin with, been designed so as to be able to then say, "See how inadequate MARC is!"? MARC does have its flaws, I'm really no fan of it as it is now, let alone the curent practice, and I have written up and published a long list of flaws. With some, I don't know why they haven't long since been solved. They may, however, be cured without sacrificing the economy of MARC, without dismissing the entire concept and logic before something demonstrably more economical and logical has been found and proven. Briefly: We can set up our entire enterprise so that, internally, we have the full benefit of an economical format that fits all our numerous and highly diverse management purposes which are of no interest to end-users. Externally, no one needs to be confronted with our internal format, but there can be an increasing variety of options to choose from, all derived from the same internal format. (ISO2709, BTW, is *not* among the flaws and issues. It is a very marginal issue of a purely internal nature and is in no way related to MARC as a content standard. MARC can perfectly well work without ISO, no one needs to bother with it except the few systems that are still unable to ingest anything else, and they can use MarcEdit to get what they want. Abandoning ISO in favor of the external format MarcEdit uses, you get rid of the 9999 character field length limit as well.) B.Eversberg.