Hello,

So far we have been considering both DOI and Handle System (besides LSID and maybe ARK) as potential GUID technologies for biodiversity informatics. However, DOI uses the Handle System as its underlying resolution mechanism, thus sharing many features with Handles. What I wanted to discuss is whether DOI additional features warrant its use by our community, or we should only consider the Handle System as a potential candidate.

The DOI handbook states that the features added by DOI are metadata, policy and business rules [1]. According to the handbook, policy is specifically developed to control how the intellectual-property based businesses using DOI operate in the network [2]. As far as I can tell, the list of current policies [3] looks more like a list of mandatory requirements that all other technologies meet, plus other policies that have little use in the biodiversity informatics field. I believe that the same applies to the business rules they try to enforce.

That leaves us with the DOI metadata infrastructure built on top of the Handle System as the only usable feature uniquely provided by DOI. For one thing, it seems to be possible to perform queries in DOI using the metadata, which is an unique feature among the GUID technologies being evaluated. However, DOI metadata framework has two drawbacks: it is tied up to a particular technology, XML Schema, and (not so serious) DOI impose a mandatory set of metadata core fields, which are geared towards the publishing industry, but resemble the Dublin Core metadata profile.

In any case, a completely separate metadata framework can be implemented on top of the Handle System, either using DOI approach (not recommended - lot of work to reinvent DOI wheel) or using the type and index fields in the handle record (but not so nice because it is based on a proprietary protocol - more on this later). In LSID one can use ANY technology, past, present or future, to implement the metadata framework. And several implementations can be supported simultaneously by the framework. The LSID specs currently *recommends*, but does not mandate the use of RDF for metadata.

In any case this metadata facility comes at a price: the DOI fees. From my professional experience in the field, I would say that the DOI funding model is not suitable for biodiversity informatics. I'm not saying that the other technologies have no setup and maintenance costs. It is just that their cost just doesn't appear as a yearly, per identifier fee. The costs of the other systems are incurred by individual agents (or authorities), usually in the form of in-kind contribution of systems and network administrator's time. To me, that seems like a much more suitable model for our field.

Given those arguments I would conclude that DOI is not suitable for our purposes and that we should really only consider the Handle System as a candidate GUID technology.

Comments are more than welcome.

Best regards,

Ricardo

[1] - DOI Handbook - Section 3.5.1 - Relationship between the DOI System and the Handle System - http://www.doi.org/handbook_2000/resolution.html#3.5.1

[2] - DOI Handbook - Section 6.1 - Policy formulation - http://www.doi.org/handbook_2000/policies.html#6.1

[3] - DOI Handbook - Section 6.3 - Current policies of the IDF - http://www.doi.org/handbook_2000/policies.html#6.3