The performance history (1883-1985) of the Met was compiled in print and released in the massive Annals of the Metropolitan Opera in 1989. It was published in two volumes. The first was Chronology 1883-1985, while the second was Tables 1883-1985. I had a copy of the print edition that I used frequently. In 2005 the Met put the searchable Met Opera Database online, using the BiblioTech/Inmagic system.
That online database was decidedly old-fashioned software by today’s standards, but it was extremely good at the sort of relational searching an opera researcher wants to do: performer + opera + role + date range, with each additional criterion narrowing the existing result. It was free. I found it very easy to use. It was far better than any other online opera archive that I searched.
In May 2016, the company Access Innovations publicly announced that it had been hired by the Metropolitan Opera to redesign the Met archive database. They said that although the Met’s data were comprehensive, the database needed to be modernized to provide better search functionality for opera fans and researchers. They planned a new custom XML schema intended to make such things as operas, performers, roles, and performance locations searchable. The 2016 announcement establishes that modernization and improved searching were explicitly among the objectives.
What resulted was the opposite of what was intended. The new archives are buggy and very difficult to use. For example, when I put a singer’s name in and then refined the results by asking how many times he’s sung the title role in Gounod’s Faust, it listed every performance of Faust in the Met’s history. The previous Archives did exactly what was asked of it. There are ways to get around this problem, but they are not intuitive and are user-unfriendly. The data themselves appear to be intact. The problem is largely the query interface and the way successive filters are combined.
There was another problem that could easily have been avoided. If you want to know the names of the production team, you have to go to the first performance of that production. Under the old system, the production team was listed for every performance of that production. Take the Met’s 2008 production of Doctor Atomic. The first performance, October 13, 2008, is marked “New Production,” and the search result identifies production personnel such as Penny Woolcock, Brian MacDevitt, Andrew Dawson, and Mark Grimmer. But open the very next performance, October 18, and the page lists the cast and conductor but not the production team at all. If you are searching for the production team for a production that was first staged years ago, you’re going to have to spend a lot of time trying to find the team that created it.
The same thing occurs with Satyagraha. The first performance on April 11, 2008, identifies Phelim McDermott, Julian Crouch, Kevin Pollard, Paule Constable, Leo Warner, Mark Grimmer, and 59 Productions among the debuting production personnel. Subsequent performances are listed simply as performances of Satyagraha, without carrying those production credits forward.
This is a poor design decision because production information belongs logically to the production, not just to its premiere performance. If I am looking at Tosca on February 17, 2010, what I want to know is: Who sang? Who conducted? Whose production was this? Who designed the sets, costumes, and lighting?
The fact that the production originally opened three, ten, or twenty years earlier should be irrelevant to the user. The archive plainly has the information necessary to associate a performance with its production-or, at the very least, it should be structured to do so. The interface simply isn’t displaying that information on subsequent performances.
The old archive did exactly the sensible thing: it repeated the production team on every performance belonging to that staging. That saved the researcher from having to determine when the production premiered, navigate backward, sometimes many seasons, and then retrieve the credits there.
It also creates a more serious historical problem. Suppose you are examining a particular performance from 1998. Unless you already know that the production originated in, say, 1987, the present archive gives you no immediately obvious indication that you need to go searching eleven years backward for the designer’s name. A casual researcher could quite reasonably conclude that the information simply isn’t in the database.
The redesign of the Met’s Archives was explicitly undertaken to make searching better. That makes the present problems ironic. The designers clearly understood that this was unusually complicated information. One of them specifically pointed out that you cannot assume an opera has only one librettist or that every Met performance occurred in New York, and said that the database had to accommodate these complexities while making the details searchable.
Yet we have now identified precisely the sorts of relationships that the replacement handles poorly: Singer → opera → role can lose the singer restriction. And: Performance → production → director/designers no longer automatically display the production personnel attached to that staging. Those are not exotic requirements. They are fundamental relationships in an opera-performance database.
The most frustrating aspect of all this is that the information has not disappeared. It is still there. What has deteriorated is the user’s ability to retrieve it intelligently. A redesign undertaken expressly to improve access has instead made some of the most elementary historical questions harder to answer. After nearly a decade, the Met should fix it. The financially challenged Met likely spent a lot of money to convert an online archive that was working well into one that was poorer than its earlier iteration.
