The Automated Election Law, Republic Act 8436 as amended by Republic Act 9369 (RA 8436), requires the conduct of a random manual audit (RMA) on the results generated by the PCOS machine, renamed Vote Counting Machine (VCM), for the 2016 national and local elections.
Section 29 of the law provides: “Where the AES [automated election system]is used, there shall be a random manual audit in one precinct per congressional district randomly chosen by the Commission [on Elections]in each province and city. Any difference between the automated and manual count will result in the determination of root cause and initiate a manual count for those precincts affected by the computer or procedural error.”
The process of verifying the results generated by the PCOS/VCM from randomly selected clustered precincts involves manually assessing and counting what are considered valid marks on the paper ballots. The result of the manual count is compared with the vote count generated by the PCOS/VCM. The accuracy rate of the vote counts generated by the PCOS/VCMs covered by the random manual audit is then computed.
The 2016 Random Manual Audit Committee (RMAC) reported an accuracy rate of 99.9023 percent. How acceptable is the 99.9023 percent accuracy rate?
The 2010, 2013, and 2016 RMAC had arbitrarily set the allowable margin of variance of at most 10 per clustered precinct. The RMACs defined “variance” as any difference between the PCOS/VCM generated count as shown in the election return and the RMA count. The 2010 and 2013 RMAC assumed an accuracy rate of 99 percent. The 2016 RMAC, on the other hand, assumed an accuracy rate of 99.6 percent. No basis can be found for these assumptions in the 2010, 2013, and 2016 RMAC reports. The accuracy baselines were only set by the RMACs during the course of the RMA activity, with the 2010 and 2013 RMACs expressly setting aside the 99.995 percent PCOS accuracy requirement set by the poll body in its Request for Proposal for the Election Automation Project. The 2016 RMAC simply adopted the rationale for the assumptions made by the 2010 and 2013 RMACs.
While RA 8436 declares it as a policy of the State to ensure that all votes are counted and consolidated accurately and transparently, no further action appears to have been done by the 2016 RMAC to determine the cause of the variances between the machine count and the RMA count.
The random manual audit addresses only an audit of the vote counts. The PCOS/VCM used in the last three elections could have been audited to ascertain how exactly it appreciated the marks on the ballot – the size of the marks, the position of the marks, the intensity of the marks, and/or the presence or absence of marks on voting targets (the voting targets are the ovals on the left side of the names of the candidates). The process of vote recording and counting and the generation of the election returns could have also been audited.
While auditing the rest of the AES is not provided by law, some parties or groups have demanded that an audit be conducted on the whole AES, especially following the introduction of a program by Smartmatic’s Marlon Garcia in the transparency server while the AES was in full operation. The program was introduced to correct the spelling of the names of some candidates with the letter “ñ.”
All counting, canvassing and consolidation is hidden behind the veil of automation. RA 8436 provides a transparency mechanism so that accredited groups are able to receive electronic copies of the election returns. But this is not done in (near) real time from the transparency server that is located in the PPCRV operations center which is racing to publish results to stay ahead of the others. The election returns transmitted from the PCOS/VCMs undergo a process of conversion from a format called election markup language to a format referred to as comma separated values. RA 8436, however, does not provide for the provision of electronic copies of the results of canvassing and consolidation to the accredited groups, thus slowing down or preventing the process of verifying the election results generated throughout the hierarchy of canvassing and consolidation. The transparency mechanism is insufficient since it does not allow observation of all processes involved in vote counting and consolidation of results, including transmission.
A post-election audit can be done on the whole AES covering the PCOS/VCM hardware and software, all devices used for canvassing and consolidation of election results at various levels, the transparency server and all network devices, the central server and all network devices, and the meet-me-room network. The audit must cover all electronic logs and records in the devices, including the electronic transmission logs from the PCOS/VCM to the first level of canvassing and consolidation and the electronic transmission logs through the canvassing and consolidation hierarchy.
The 2019 elections is 32 months away. There is still time to prepare legislative proposals that will enhance the audit provision of RA 8436 to include auditing the whole AES. The veil of automation that hides all processes must be lifted. Transparency mechanisms must be put in place so that the election results can be monitored and observed from the vote counting machines all the way through the canvassing and consolidation of results up to the national level, including transmission.