Enough of praise, now for some constructive criticism...
If one uses the 'keyword' field of the Advanced Search form to look for "african red slip", a list of 14 objects is returned. Clicking on the fourth item brings up a useful description of an ARS Hayes 53 plate of the fourth to early fifth century AD, an object purchased in 1989 but undocumented prior to that date (mfa.org:1989.690).
The URL used to access the record is:
This string of characters is so long that nobody would use it as a fixed reference to that object in a publication. Rather, one would fall back on the scholarly practice of using "MFA 1989.690" and leave it up to the reader to track down the object. The MFA site makes this relatively easy in that the Advanced Search page has a field for accession number. If you enter "1989.690" in that field, you end up at a page with the URL:
Although this feature is undocumented, that URL can be shortened to:
That is a reasonable length and means that the MFA's own unique identifier for the object can be made actionable in a fairly clean manner. There is, however, a further issue. That URL does not lead to the full information and image for the object, only to an abbreviated list view. The shortest URL that I could compose to link to the full view was:
I don't know the semantics of the ID field, but it looks like a numeric primary-key imposed by the database system, one that is not otherwise publicly documented by the institution. This is my main criticism: the full information for mfa.org:1989.690 is not directly accessible by its publicly documented unique identifier.
This may seem a subtle point, and I don't mean for it to distract from the obvious benefit provided by the MFA's efforts. But as we move towards an effective infrastructure for digital publication and scholarship, I would like to see URLs such as http://mfa.org/collection/1989.690 instantiated as valid links to useful information.