Florida Commission on Open Government Reform
Sarasota public hearing
Testimony of Maurice Tamman
Online Editor, Herald-Tribune, Sarasota, Fla.
February 12, 2008

My name is Maurice Tamman. I am the online editor of the Herald-Tribune. For most of my 18- year career I have worked as database editor/projects reporter. Before joining the Herald Tribune three years ago, I worked at the Atlanta Journal Constitution. I am a member of Investigative Reporters and Editor and the National Institute of Computer Assisted Reporting. I routinely teach about using databases for both organizations.

Let me start by trying to explain how databases work and how limits on database access are a grave threat to open records.

This is especially important because information is exploded in a database and put into different tables, and only pulled back together when needed.

Let’s take this piece of paper as an example of how information might be organized in a database.

For starters, you might want to keep track of information about when and where this testimony took place. You might store that information in a table called “events” that contains fields with the commission name, the location of the hearing and dates. To differentiate between meetings, each event would be assigned a unique identifier – a series of numbers and letters that will be assigned to each meeting.

You might also want to track information about each speaker, so there would be a second table called “speakers” that would have fields for the speaker’s name, employer, title, employment dates, bio info and contact info. Each speaker would also be assigned a unique identifier.

If you looked at each table separately, matching the elements of one table to the other would be impossible. Realistically, this paper document does it but computers do not store information in databases that way.

What this database would need is a third table to tie everything together. In this case, it might be a table called “schedule.”

In that table there might be date and time fields for each speaker, but there would be no additional information about the location or speaker except fields for the event’s unique identifier and the speaker’s identifier. Those identifying fields would allow all three tables to be tied together.

This is usually called a relational database. It allows information to be stored efficiently and married up when needed.

In the real world, databases can contain hundreds of tables with intricate links tied to specific identifiers.

Today, governments store most of what we view as documents, like this piece of paper, in databases. Gaining access to those databases, all the tables stored in those databases and information about how they all link together, is the most important open records issue in Florida and the nation.

Without access to all three elements, effectively the data inside that database is shielded from public scrutiny.

The matter is further complicated because some of the identifiers used by governments are themselves exempt. So, for the sake of one exemption, such as Social Security numbers, every “document” contained in a database is taken out of the sunshine.

Some agencies will replace those Social Security numbers with a new unique identifier, but usually at great cost. Or, they will truncate the Social Security numbers to four or six digits, which effectively makes them useless because the number could now match numerous records in a table rather than just one.

Using the truncating scrubbing method, for example, one of my children could have the same unique identifier as me. Clearly, that could cause grave errors, but more importantly produces a document that is not even a close facsimile of the document contained in the database.

And the issue is getting worse. In recent years, local governments have privatized much of their database administration to vendors who routinely claim proprietary privileges over the design of the database. Yes, they generally agree to give out the data. But without the definition of each table and the fields contained in those tables and how they link together, it is useless.

Just today, I was talking to an IT employee with a local county who said a contract he signed with a vender barred the county from sharing a requested database’s design information.

This database contains scores, perhaps hundreds of tables. He agrees that we are entitled to the data but the contract bars him from giving me the information that would bring it into the sunshine.

This is not isolated and if we don’t deal with these issues, we will lose all substantive access to public records.

Thank you.

Testimony before Commission on Open Government Reform in Sarasota