CNE to audit self...slowly, very slowly...

CNE’s decision to grant itself an amazing five weeks to conduct its “hot” audit, has rekindled my interest in electoral nuts-and-bolts. The CNE Audit “Instructivo” – the official instructions for the audit – starts on page 58 of this large PDF document. The vote-tallying procedure (acto de escrutinio) is set out from page 74 of the same document.

Sumate’s big beef with the Escrutinio procedure is that nowhere does it instruct polling workers to physically count all the ballot papers issued. Articles 168-174 of the Framework Law on Suffrage and Political Participation, set out the rules for the vote-tallying process in some detail. According to Article 172, the “Acta de Escrutinio” – the Vote Tally Report that serves as an official record of votes in each polling station – must contain both a count of the total number of ballots that were cast and the automated tally of how many votes each candidate and party got.

The Acta de Escrutinio needs to contain both these counts because Article 220, paragraph 1, says an acta is invalid if the two numbers don’t match. The only way to determine whether the two numbers match or not is to actually count the ballot papers.

(Somewhat bizarrely, the law does not say you need a hand-count of how many votes each candidate and each party got, you just need a hand-count of how many ballots were cast in total. This is meant to ensure the automated count does not contain more votes than there are actual ballot papers.)

Under the rules CNE has published, ballot papers will only be counted for a single box in a sample making up about 45% of voting tables. The votes will be audited by a hand-count, but instead of being immediately matched against the automated results, they will be sent to CNE, which will have five weeks to build a data base and conduct a statistical study of discrepancies.

A few things to note about these rules:

  • Since only CNE will get to see the hand-count data, CNE will effectively audit itself – how many companies get away with that?
  • The audit procedure contains no mechanism whereby hand-counts trump automated counts in the event of a discrepancy.
  • By the time the audit is published, 5 weeks after the vote, the election results will long since have been announced and the seats adjudicated – and there’s no mechanism for re-adjudicating a seat in the event that the audit demonstrates fraud.
  • Though not directly related to the audit, the Vote Tallying (escrutinio) procedures AGAIN call for Actas de Escrutinio to be printed after the machine has been connected to CNE headquarters and the data been sent in (article 46, page 75 of the document I linked to above.) This week, Jorge Rodriguez pledged to backtrack on this item. Still, it is hard to figure this sequencing was in there in the first place, given the grave misgivings about bi-directional communications it has always aroused.
  • Frankly, if this is how they propose to go about it, it’s hard to see what good auditing could do. It’s silly to think a procedure this opaque, this slow, and this CNE-centric could really convince anyone who currently doubts CNE’s probity. Ultimately, if the political will was there, CNE could run a credible hot audit, one that would end once and for all opposition misgivings about its fairness. It could, but it hasn’t.