Automated Election Auditing from DRE Logs

1
Automated Election Auditing from DRE Logs
P. Baxter1, A. Edmundson2, K. Ortiz3, A.
Quevedo4, S. Rodríguez5, C. Sturton6, D.
Wagner6 1Clemson University, 2Cornell University,
3University of Puerto Rico at Arecibo, 4Miami
Dade College, 5University of Puerto Rico at
Mayagüez, 6University of California at Berkeley
Overview Voting machines used in any election
generate audit logs that contain an immense
amount of raw data. This study implements ways in
which this data can be analyzed to produce
meaningful and organized reports these reports
can be used to enhance the accuracy and
efficiency of elections. Our study has been
integrated into a public website intended for
election officials.
Results
Problem Electronic voting machines generate large
audit logs containing an immense amount of data.

• What analyses are possible from the given data?
• Can these analyses be made easily available to
  election officials and interested third parties?

Our Contribution We analyze audit logs from the
South Carolina 2010 elections to find
Analyses

• Inaccuracies in election-tally reporting
• Procedural errors that occurred on election day

• Equipment that may need servicing
• Inefficiencies at polling locations that could
  be averted in future elections

Our Tool AuditBear We developed several analyses
and incorporated them into a web-based tool.
Auditing DRE log files in a systematic way can
promote accurate post-election audits. Our tool
standardizes this process in the ESS client
base.

• Recommendations
• Recommendations to make audit files more usable
• There should be more detailed event descriptions
  available.
• Event logs should contain accurate date and time
  logging.
• Additional events should be used for important
  actions.

• Acknowledgements
• Special thanks to Dr. Kristen Gates and the TRUST
  program staff.

Try it! www.audit-bear.org
This work was supported by the TRUST Center and
ACCURATE Center (NSF award number CCF-0424422)