OPPOSITION REFERENDUM FRAUD THEORIES

C O N F I D E N T I A L CARACAS 002730 SIPDIS NSC FOR CBARTON USCINCSO ALSO FOR POLAD STATE PASS USAID FOR DCHA/OTI E.O. 12958: DECL: 07/12/2014 TAGS: PGOV KDEM PHUM VE SUBJECT: OPPOSITION REFERENDUM FRAUD THEORIES Classified By: Abelardo A. Arias, Acting Deputy Chief of Mission, for R easons 1.4(b) and (d). -------------- Summary -------------- ¶1. (C) The Venezuelan opposition has developed theories -- though no solid evidence -- of alleged fraud and irregularities in the August 15 presidential recall referendum. Among the theories, mathematicians have proposed that the election results were manipulated using a pre-determined percentage by state and a cap on "Si" votes that added votes to the pro-Chavez "No" option. At the operative level, Chavez opponents charge, there was lack of appropriate custody over the software used to program voting machines and an unexplained 17-minute failure of the CANTV transmission network at a critical time. The GOV's alleged manipulation of the electoral registry before August 15, the bulge in the electoral registry a month before the referendum, and unusual changes in poll workers immediately before also boost the opposition's doubts about the referendum. While the opposition's theories, if true, would indicate manipulation and irregularities on a large scale, none alone would appear to be of sufficient magnitude to change the results. End summary. -------------- Fraud Theories Abound -------------- ¶2. (C) In the wake of the opposition's 40-60 defeat in the August 15 recall referendum against President Hugo Chavez, fraud and conspiracy theories have proliferated. Coordinadora representative and Gente de Petroleo leader Juan Fernandez laid out the principal arguments to poloff on August 24. There are two mathematical observations that to the Coordinadora suggest electoral fraud: -- "2,1,0" The opposition claims a state-level analysis suggests that each state was assigned one of three possible percentage outcomes (mathematically expressed as a 2,1,0 series). States with high opposition support, for example, were assigned a closer percentage split between the "Si" and "No" votes (prejudicing the "Si" votes); areas with lower opposition support were assigned wider spreads that gave more votes to "No" than were actually cast. -- "El Tope" Some mathematicians looking at the voting machine level note an unusually high recurrence of similar numbers (affecting between 3,000 and 4,000 machines) for the "Si" votes, leading them to conclude that a cap ("tope") had been applied. According to this theory, votes cast for "Si" in excess of this cap were applied to the "No" option, which did surprisingly well in some pro-opposition strongholds. Fernandez said the mathematicians believe that the cap and "2,1,0" phenomena are connected. Fernandez also said that mathematicians confirmed evidence of tampering by applying the "Law of the First Digit" (aka Benford's Law),the same mathematical principle used to detect Enron's manipulation of energy markets. ¶3. (C) Constitutional lawyer Tulio Alvarez_, who is leading the Coordinadora's special commission to document the alleged fraud, told reporters August 22 the mathematical phenomena would have been undetectable had it not been for the unprecedented 27 percent abstention rates. The high voter turnout, he claimed, forced many more "topes" to be imposed by the machines than what was previously planned. CNE Response: Jorge Rodriguez told reporters August 17 that it is expected that voting tables would have similar results. In presenting examples to support his argument, he showed data from similar tables, not/not from the voting machines, which misrepresented the opposition's argument about similar results at the machine level. International Observers Response: The Carter Center's Jennifer McCoy told reporters August 17 that the cap phenomenon could be seen among "No" results as well. McCoy later sent a letter to the Coordinadora August 26 stating that, after consulting a Stanford mathematician and a Bolivian pollster, the Carter Center concluded that the recurrence of similar numbers is statistically probable. The OAS's Edgardo Ries told poloff August 21 that similar results in a voting center tends to reinforce the validity of the results; it would have been more suspect had voting machines in the same electoral precinct varied substantially. Comment: Post does not have the expertise to judge the merit of these mathematical arguments. -------------- Ghosts in the Machine -------------- ¶4. (C) Fernandez gave two possible explanations for how the mathematical manipulation was carried out. Either the software had been tampered with, he said, or the data transmission was compromised. Fernandez complained that the CNE had not given them full access. He claims Coordinadora informatics specialists were only given limited access to portions of the code. Coordinadora representatives also charged publicly that the CNE distributed thousands of machines to centers prior to the audit. CNE/Smartmatic Response: The CNE's Rodriguez said publicly the Coordinadora had full access to the software used to program the machines and had representatives present when the machines were individually programmed before being sent to the field. Smartmatic officials told emboffs in June that the CNE and political actors would have access (read-only copies) to only about five percent of the total code used to operate the machines, principally because of proprietary reasons. Mugica said the CNE and political actors would be allowed to audit -- but would not be able to alter -- the relevant codes for tabulating votes thanks to a three-step verification process within the company to certify the software was as advertised. Smartmatic officials claim the programs exhibit a unique digital signature that would be altered had the source code been altered. Smartmatic President Antonio Mugica told poloff August 18 his company had given the Coordinadora copies of the digital signature of every machine to check the authenticity. International Observer Response: Observers were not present for most of the software development process, though they have publicly expressed confidence in the performance of the voting machines. Comment: It appears likely the Coordinadora did not have sufficient participation in the development and loading of the software and transportation of the machines to voting centers, making manipulation a possibility. Also, the audit of the machines was a one-day affair with a sample of machines. Some 14,000 of the more than 19,000 machines had already been distributed around the country. ¶5. (C) Fernandez also called into question an unexplained outage of the CANTV transmission network between 8:12 and 8:29 p.m. on election night. A Venezuelan judge documented it at CANTV's request. According to Fernandez, CANTV said it could not account for the data transmission of at least 200 machines, though Fernandez believes there were more. Fernandez said that since the voting machines can receive data as well as transmit, it was possible for a "phantom server" to intercept the data and send altered data to the CNE as well as back to the voting machine. Smartmatic Response: Mugica claimed his machines cannot be remotely re-programmed and noted that the machines are only connected to the network for transmission, amounting to no more than two minutes per machine. International Observer Response: OAS and Carter Center observers found no evidence of manipulation, though Carter Center observer Edgardo Mimica confirmed the outage had taken place. Observers did not perform audits of CNE servers handling the information. Comment: CNE rules had the transmission taking place before/before the tally sheets were printed. The timing is significant because CNE Director Rodriguez announced at 7:50 p.m. that polling stations would stay open until midnight, instead of 8:00 p.m. as previously announced. As CNE public announcements are notorious for not filtering down to the voting centers, it is likely that many of the 19,662 voting machines attempted to connect to the network a few minutes after 8:00 p.m., coinciding with the reported blackout. This fact is significant in that only voting machines closing at 8:00 p.m. would have been affected, and might not be detected by the two supposedly random samples taken by the CNE for audit purposes. -------------- 1-2-Switcharoo -------------- ¶6. (C) A few press reports surfaced on election day that some voters noticed on their paper receipt that their "Si" vote was preceded by the number Q" rather than the number "2", the code assigned to the "Si" option. CNE/Smartmatic Response: Smartmatic officials said such an occurrence was impossible. International Observer Response: Both Mimica and OAS chief observer Edgardo Reis told poloff they did not find any such cases in the 354 machines audited. Comment: This alleged fraud was the product of election day rumors. It probably had some resonance later among Chavez opponents because the results announced by the CNE were the reverse of what exit polls showed. -------------- Masters of The Electoral Universe -------------- ¶7. (C) The Coordinadora also goes back to the weeks before the election with its theories and allegations when the electoral registry plumped up by an unprecedented two million new voters in the two months before the referendum, attributed mainly to the GOV's massive identity card/voter registration drive. Coordinadora electoral chief Amado Dounia estimated prior to the election that a registry of 12 million would give the opposition 3.8 million votes, just over the threshold needed to recall Chavez. He estimated the GOV would get just 3.4 million voters, using available polling data and time studies of the election procedures. The addition of two million new voters (from mostly pro-GOV social classes),however, changed his analysis to show a possible GOV victory. ¶8. (C) Dounia doubted the new voters actually turned out to vote based on his tracking of the flow of voters on election day. The Coordinadora's network of witnesses at each table reported that nearly seven million people had voted by 9:00 p.m. on election day. By midnight, however, the system was projecting 10 million voters. Three million additional votes in three hours, Dounia estimated, would have required all machines to be working at a rate of one minute per vote, much faster than what had generally been observed during that time. Dounia suggests that the new voters were fraudulently added en masse to the "No" option late in electoral day. He had no theory, however, for how such a fraud could be carried out. ¶9. (C) Dounia also told poloff that the CNE changed voting centers for some 300,000 voters just days before the election. Forty thousand of these voters were able to correct the problem, but the remaining 260,000 voters were forced to go to centers far from home or to abstain from voting. Dounia accused the CNE of intentionally trying to eliminate "Yes" votes using this method. The quantity could have been important, he noted, were it a tight race. Additionally, Chavez opponents allege that last minute changes of poll workers affected the efficiency of voting centers. CNE/Smartmatic Response: The electoral registry was a source of contention within the CNE. CNE Director Rodriguez even filed charges against fellow Director Sobella Mejias because the latter allegedly had failed to expunge thousands of deceased persons from the registry. The pro-opposition Mejias claimed she had been shut out of changes to the registry. International Observers: The Carter Center's Mimica told poloff privately the last minute changes to the electoral registry were unfair, though the observers did not criticize the CNE publicly. Comment: The GOV's "Mission Identity" to give identity cards to millions of Venezuelans and foreigners had few controls to ensure integrity of the cards, the only proof needed to vote. The CNE was a doormat to the new names presented by Mission Identity, pouring them into the electoral registry. Also, some 200,000 names were added to the registry days after it was closed. Coordinadora representatives complained to emboffs that the registry was changing daily even the week before the referendum. The GOV's manipulation of the registry was a significant factor in Chavez's victory. -------------- Comment -------------- ¶10. (C) We do not know whether the opposition's allegations, if they turn out to be true, would sufficiently alter the results to overturn Chavez's victory. No one theory alone appears to be of sufficient magnitude to be a silver bullet. The technical arguments raised by the opposition will have to be evaluated by experts in light of realities of the voting system. International observation, though imperfect, did not detect any evidence that would support the opposition's claims, but then again, these would not be things readily evident to persons eyeing the process at the voting precincts. Some opposition allegations would require a significant number of accomplices in the CNE, the military, Smartmatic and even CANTV. It seems reasonable that Smartmatic and CANTV, both firms that have an interest in avoiding allegations of impropriety, would probably not risk company health by engaging in fraud. This would not rule out, however, the actions of specific employees, and in fact, the opposition's theories aim more at the GOV and the CNE, not the companies. Chavez opponents remain intent, nonetheless, on swimming against the strong tide of opinion blessed by international election observer organizations. McFarland NNNN 2004CARACA02730 - CONFIDENTIAL