• Vulnerability revisited: AES house of cards



    By Giovanni Tapang, Ph.D.

    Last weekend, we joined a press conference with the AES watch at UP Diliman where experts pointed out that the 2013 elections turned out to be worse than 2010. Despite the denials of the Comelec and Smartmatic, the very fact that the poll count was stuck around 69 percent a few days after elections. This implies that 30 percent of the PCOS machines did not function as it was supposed to during election day. The results were only transmitted by either replacing the PCOS machine itself or carrying the compact flash (CF) card to another working machine.

    The Smartmatic automated election system (SAES) is a complex collection of several component subsystems that were expected to function as one during election day. These subsystems were expected to count correctly our votes and then pass on verified tallies of the results at each stage of the polling process. The SAES failed to work as expected and was only forced to work through the creativity and ingenuity of the local Board of Election Inspectors (BEI). They had to go through hoops to make the PCOS machine work such as inventive ways like pushing ballots with brooms to make space for more.

    Another ad-hoc adjustment was when some of the BEIs had to trim the ballot’s size in order to fit the PCOS machine. This means that these machines were not made to accommodate the ballots as were supposed to be when they were manufactured.
    In general, what happened at each polling precinct reflected what was done in general by the Comelec to make the 2013 midterm elections work. Chairman Brillantes, Smartmatic and the Comelec had to improvise in order to make the whole election system work.

    Rather than waiting for the ladderized reporting of the certificate of canvass, the Comelec changed its rules and allowed other means for votes to be directly reported to the national canvassing board. Such ad-hoc changes during the canvassing period, the veracity of the final count and tally used as the basis for proclaiming winners of the electoral contest was put into doubt.

    It was not only during the canvassing that such improvisations were done. After the final testing and sealing (FTS), Chairman Brillantes announced that the PCOS machines had been re-calibrated to read the ‘X’ marks by allowing less than the 40% shading that was used last time. The announcement came after the FTS found that a lot of the ballots with ‘X’ marks were scounted despite Comelec’s own information campaign saying that ‘X’s are not allowed. These types of post-facto adjustments of the procedures seem to be the norm rather than exceptions.

    In 2010, there was a massive recall of the compact flash (CF) cards a few days before the elections. This did not happen this time. In 2013 problems involving CF cards occurred during the tallying and transmission of the results. Some PCOS machines failed to transmit because the CF cards were either corrupted or not functioning at all.

    This means that the recall of the CF cards, this time in 2013, happened not before but after the elections. These CF cards had to be replaced and the replacements had to come from Manila, and that delayed transmission further. The 30 percent failed transmission can in part be attributed to this recall.

    This CF card recall opens the possibility that these cards are different from those originally intended for the machines. The configuration stored in these new CF cards could be wrong and result in the miscounting of local and national election results. If one’s name is not properly encoded in the cards, even if it is printed in the ballot, then their votes will not be properly counted and transmitted.

    The percieved 60-30-10 percent ratio in the senatorial race has also raised concern. This has to be verified in a more local setting. We need to get copies of election returns in the precinct level and check if the ratios still hold. Rick Bahague of the Computer Professionals Union is trying to collate these. If you have copies, please send the scans to rick@cp-union.com.

    That the elections in 2013 are a truthful tally of the votes cast in the precinct rests on several assumptions. First,the PCOS machine has the correct and working program to scan the ballot, the correct and working program to tally the results, the correct and working program to transmit the votes. The CF card configuration is also assumed to be correct and the physical card is also assumed to be functional.

    The program to receive the results at the canvassing and consolidation server at the municipalities and provincial centers are also assumed to be working and correct. The same is assumed to be true for the national transparency and Comelec servers for the national level.

    In the 2013 elections, all of these assumptions turn out to be unproven or untested for the Comelec-Smartmatic automated election system. Much like a house of cards, when you take one assumption away, the whole setup falls down. We can thus only surmise that the recent midterm elections are really much worse than the previous ones.


    Please follow our commenting guidelines.

    Comments are closed.