Sequoia Voting System database laid bare but no secrets found

David M Williams
Wednesday, 21 October 2009 19:36

Opinion and Analysis

At first glance, it appears there is no code within the database at all. Browsing the appropriate section in SQL Server’s Management Studio shows no stored procedures, views or triggers at all.

Yet, Jim March’s output from the Linux strings command yielded actual SQL statements. The examples he posted were nothing unusual despite the odd inference March drew that they somehow ‘controlled’ the flow of the election.

Rather, they appeared to be nothing more than regular old maintenance scripts which create the tables in the first instance and define the fields.

With some work the same code can be found. March stated that Sequoia redacted the database before sending it to EDS and it seems the nature of this redaction was not to ‘vandalize’ the binary backup file but instead to drop all the programmability.

Unfortunately for Sequoia their database administrator did not think to compact the database before backing it up. SQL Server – like most databases – does not automatically shrink its files.

Like your computer’s hard disk a database does not truly remove data when deleted but instead marks the space it held as free for re-use. Thus, the code remained in the database although invisible to Management Studio and can be inspected via other means.

Nevertheless, the code does not perform any nefarious task but simply serves to create the 88 database tables found within, with names like AUDIO, BALLOT_CONTEST, BALLOT_CONTEST_POSITION, BALLOT_PRECINCT, BALLOT_STYLE, CANDIDATE, CONTEST, EVENT_LOG, LAYOUT, PROVISIONAL_VOTE, REGISTRATION, TALLY_BLANK_BALLOT, VOTER and so forth.

So, the structure of the database file is uninteresting – being untampered with. The programmability content of the database is uninteresting. What’s more, the data content of the database is similarly uninteresting.

Within table VOTER we find records of voters but with such non-identifying fields as VOTER_ID, SERIAL_NUMBER and PRECINCT_ID. I can tell you that VOTER_ID 885 has SERIAL_NUMBER 41970 and is in PRECINCT_ID 594 but that doesn’t tell me who the person is or who they voted for, or even if they voted at all.

Mokurai’s DailyKos story has hit the Internet with prominent sites like SlashDot featuring it but yet the claims raised are sensational and baseless.

While I can respect the very good intentions of the EDS and Jim March there actually is no grounds for the criticism being levied against Sequoia.

The database file has not been vandalised, and the fact March couldn’t restore the database should have tipped him off from the start he didn’t actually have the technical literacy to analyse what had been supplied.

I’m bound to receive criticism for this article by open source voting advocates. Yet, we can’t advance a cause by putting forth false accusations.

There are many good reasons why voting system ought not to be closed source without embarrassing ourselves by making others up.