Aut oma tio n o f Ele ct ion s Ano ba talaga ito?
Agenda Why we need to automate elections How cheating is done The manual election system Alternative solutions
Why do we need to automate elections
Process is too long. It takes 2540 days before national positions can be proclaimed.
To eliminate wholesale cheating, incl. DAGDAG-BAWAS
How is cheating done?
Retail cheating
Vote-buying (carbon paper, ballot marking, “Lanzadera”) Ballot box snatching Ballot box stuffing “Flying voters”
To v ote once & onl once …
y
Wholesale Cheating
Mani pula ti on of pr eci nct counts Votes counted as
Recorded in words as
Marikina 1998
Manila Dist-4 2001
Mani pul ati on of pr ovi nci al r esult s Municipal Certificate of Canvass
Provincial Certificate of Canvass
Zamboanga del Norte, 2001
Manipula tion in t he of ficial municipal & pr ovincial can vas s Official municipal canvass
Official provincial canvass
Illegal votes
Alaminos
2599
12599
10,000
Dagupan
13784
28784
15,000
Illegal votes from 22 other towns in Pangasinan for a senatorial candidate in the 1995 elections equaled 112,994.
Anatomy of DAGDAG-BAWAS
Province 5 Cand . A
%
Cand . B
%
COC: NA MFR EL : NAMFREL T rend:
COC In cr eme nt:
15,925 1,951
20%
58,292
75%
To tal Vot er s 87,520
To ta l % 100%
77,967
89%
9,553
11%
7,142
Province 5 Cand . A
COC:
33,634
38%
49,803
57%
To tal Voter s 87,520
NA MFR EL :
15,925
20%
58,292
75%
77,967
89%
-89%
9,553
11%
NAM FREL Trend:
COC In cr eme nt:
%
1,951
17,709
Cand . B
%
To ta l % 100%
7,142
185%
(8,489)
Province 5 Cand. A
%
Can d. B
%
Total Vo te r s
Tot al %
COC:
33,634
38%
49,803
57%
87,520
100%
NAMFR EL:
15,925
20%
58,292
75%
77,967
89%
-89%
9,553
11%
NAM FR EL T rend:
1,951
COC I ncrem en t:
17,709
Dagdag Bawas:
15,758
Dagdag bawas =
31,389
7,142
185%
(8,489) (15,631)
Dagdag-Bawas Summary %
Total Voters
Province 1
DagdagBawas 103,359
75.8
136,297
Province 2
153,761
64.9
236,768
Province 3
134,505
57.9
232,010
Province 4
60,615
41.33
146,652
Province 5
31,389
35.86
87,520
Province 6
53,798
19.4
277,318
Province 7
38,844
12.35
314,577
Province 8
22,289
9.17
242,944
Estimated Total
598,560
The Manual Election System
Basic Election-related Data
83 Provinces
200 Congressional Districts
1,600 Cities and Municipalities
40,000 Barangays
250,000 precincts
40M+ voters
Elective Positions National Positions President Vice-President 24 Senators (12 elected/3 years) Party List Local Positions Congressman Governor Vice-Governor Provincial Board Mayor Vice-Mayor Councilors
Definition of Terms
BEI
- Board of Election Inspectors (250,000)
CMBOC - City/Municipal Board of Canvassers (1,600)
PBOC
- Provincial/District Board of Canvassers (200)
NBOC
- National Board of Canvassers (Comelec/Congress)
ER
- Election Returns
SOV
- Statement of Votes
COC
- Certificate of Canvass
The Manual Election System 1. Ballots tallied by BEI in each precinct and ERs prepared 2. BEIs bring ERs to CMBOCs 3. CMBOCs canvass ERs and prepare SOVs and COCs; bring them to PBOCs 4. PBOCs canvass COCs and prepare provincial COCs and SOVs; bring them to NBOC 5. NBOC (Comelec) canvasses COCs; Congress canvasses Pres/VP COCs
Manual Tallying/Canvassing Time Line 5-12 hrs
PRECINCT TALLYING
10 days
20
30
40
CITY / MUNICIPAL, PROVINCIAL AND NATIONAL CANVASSING (25 – 40 DAYS)
Given the above time line, it becomes obvious, which phase of the election process should be automated.
So now, we want to apply technology in our elections ... to speed up the process and to be able to proclaim the winning candidates earlier; 2. to minimize, if not eliminate, cheating; Ahh … but we have added a third ... 4. to make the election process transparent to the public. 1.
Election processes that can be automated Voters list Voting Tallying Canvassing Reporting
In automating elections, two issues immediately come to mind:
How do we secure the system?
Which technology should we adopt?
Two ways of securing a system
Fence it in very tightly so no intrusion can ever occur (security by obscurity).
However, implementor must prove to all interested parties that system is indeed extremely secure.
Not easy to convince all; there will always be doubters.
Secure the system, but make a copy of all software and data (read only) accessible to all interested parties and to the public. Proof of veracity and accuracy of results becomes automatic.
We favor this because it is the transparent alternative.
Features of an ideal automated election system for the Philippines
Automates canvassing
Tight security measures
All steps transparent to the voting public
Software used available to the public
Digital counts and results, in all steps, available to the public (any one can do his own tabulation)
Results quickly verifiable all the way to original source documents
Cost-effective (P4-8 billion, depending on the solution)
Minimum or no training required for >40M voters
Minimum or no storage concerns after each election process
Not dependent on the trustworthiness of the implementors
Alternative election automation technologies
DRE (Direct Recording Electronic) System – “touch-screen”
OES (Open Election System) - manual voting & counting, and automated canvassing PC-based data encoding of ERs
OES-OMR (Optical Mark Recognition) System – pre-printed ballots, read by OMRs at the voting centers (schools)
Direct Recording Electronic System 1. 2-4 Units per precinct 2. Touch screen, mouse, or keyboard CITY/MUNICIPAL BOARD OF CANVASSERS
PROVINCIAL BOARD OF CANVASSERS
NATIONAL BOARD OF CANVASSERS
3. Voter’s choices printed for audit purposes 4. At end of voting (3:00pm), ER is printed 5. ER transmitted to CMBOC and NBOC
DOMINANT PARTY
DOMINANT OPPOSITION
CITIZENS ARM
MEDIA & OTHERS
6. NBOC transmits data to interested parties 7. CMBOC produces SOV and COC; transmits to PBOC
PRECINCTS
8. PBOC produces SOV and COC; transmits to NBOC 9. NBOC produces
Direct Recording Electronic System PROs Instantaneous tally of votes at precinct level If all precincts connected, almost instantaneous canvass at City/Mun., Prov., & Natl. levels; ergo, theoretically, national results known 1 hr. after close of voting Less work for BEI With one printer per precinct, printing of 30 copies of ER at precincts is easy No ballot box snatching
CONs Cost prohibitive, estimated at P15-20B (some est. >P30B) Logistics can be a nightmare (750K units to 250K locations) Thousands of technical people req’d (but where to deploy?) BEI training staggering 40 Million voters to be trained Where online connection is unavailable, difficult to secure electronic media (CDs) After each election,
If we can’t see it, we can’t trust it!
TransparentElections.org
Optical Mark Recognition
1. Voters mark preprinted ballots 7
2. Ballot boxes brought to City/Mun Tab Center.
8
3. ERs printed; signed by BEI
5
6
4 5 3 2
4. CMBOC tabs all ERs; produces SOV and COC 5. ERs, SOV & COC sent to PBOC and NBOC 6. NBOC transmits to interested parties 7. PBOC tabs COCs; produces Prov SOVs, COCs 8. PBOC transmits all data to NBOC; produces Nat’l SOV, COC
Optical Mark Recognition PROs
CONs
Ballots are pre-printed so voters simply mark choices Voter training minimal, relative to DRE Faster, because tally of votes automated Less work for BEI at precinct level Cost less than DRE; approx. P8B (using $2,000 OMRs)
Internal tallying. Voters won’t see and may not trust count Wholesale cheating, usually possible only at canvassing level, can happen at precinct level Sensitivity to external marks or smudges Difficult to fairly resolve over-marked ballots Easier to add to undermarked ballots Need to store specialized OMR machines
But … the newspapers reported that the automation of the last ARMM election was successful. Was it?
Problems with DRE
Problems in the initialization of voting machines since some of the BEIs committed repeated errors in punching their pin codes. Designated IT experts had to take over the initialization process to speed up the process because this has caused delay in voting.
Operational delays in starting the machine due to defective DREs which were however immediately replaced.
Incidents of automatic machine shut down while the voters were casting their votes. It was resolved by replacing the electronic voting machines (EVMs).
Many BEIs were unfamiliar with the EVM due to the overnight substitution of BEIs with untrained persons which could have been perpetrated by interested parties.
Problems with DRE (cont.)
Many voters and BEIs were unfamiliar with the system since there was hardly any opportunity to see and test the DRE before the elections, this could be attributable to lack of voter education due to time constraints. There were several instances where illiterate voters and those who were not familiar with the new system were being accompanied by another person inside the precincts as coach. The relationship between the voter and his/her companion was not properly validated. These so called coaches do not only guide the voters inside the voting precincts but even control the hand of the voter as to who to vote. Even some of the BEIs and watchers have been seen coaching the voters as well. On the secrecy of voting, there were no booths to cover the DRE machines enabling the voters of another adjacent DRE machine to see the votes being cast. Size of the candidates’ pictures (too small) made the image unclear.
Problems with OMR 1.
2.
3.
4.
5.
6.
Votes shaded in the OMR ballot were exposed to tampering. Reports of unscrupulous erasures were documented. The distribution of the official OMR Ballots were likewise exposed to the threat of advance shading. The voters would sometimes accidentally scratch or ink-blot the OMR ballots which hampered its optical scanning. The folding and unfolding of OMR ballots resulted to some extent in time inefficiency at the counting centers. In a number of the PPCRV’s poll watchers reports, some BEIs, accidentally perhaps, tore off the bar code of the ballots resulting in their rejection. The BEIs had the lack of procedural knowledge on the disposition of invalid ballots.
Problems with OMR (cont.) 1.
2.
3.
Valid ballots that were crumpled, folded (to fit in the size of the ballot box) and those that contained unnecessary markings or smudges as well as those lightly shaded ballots were rejected, which slowed down the counting. The number of ballots to be counted per ACM was not as it was projected. There are discrepancies in the counting of ballots between those who actually voted with results counted. An example of this was experienced in one of the precincts of Shariff Kabunsuan where the actual number of voters is 371 but the machine counted only 276, there was a discrepancy of 95 ballots papers. But, after the BEIs conducted a recount the machine counted 365. Incidents of over voting in some precincts that used OMR, such as Bumbaran, Lanao del Sur, were also encountered because of BEIs voting in their assigned precincts. In these cases the result was invalidated (treated as zero) and COMELEC had to override it.
Problems with OMR (cont.) 1.
2.
3. 4.
The counting and Canvassing System (CCS) was not programmed to accommodate failure of elections in some municipalities, such as Balindong, Lanao del Sur and Basilan, thus the machine had to be shut down to force the system to close the counting. There were incidents wherein the system would not close the counting and canvassing since it showed that it didn’t count 100% of the total votes from all the precincts though all precincts were able to count the votes. Some ACMs to include laptops and printers overheated, stopped functioning and had to be restarted. Constant paper jamming (of the OMR Ballots). The attached full 196-key Keyboard in the ACM is open to programming intrusion.
But even assuming that use of the DRE and OMR during the ARMM elections were successful …
From Dr. Aviel Rubin’s book, “Brave New Ballot”
“Past performance is no guarantee of future results, especially when it comes to security.”
“Success on a small scale does not guarantee success once the scale of a project is enlarged.”
Besides (and very few people realize this), …
The ARMM election is a nonevent!
If we insist on using DRE …
The COMELEC, with its limited resources (permanent staff of 5,000), will have to worry about running two different systems. And this is apart from the many other duties that they normally perform during elections (cleaning up voters’ lists, training teachers and other deputized agencies, regulating political campaigning, addressing possible violence among partisans, etc. .)
This is the perfect formula for failure!
TransparentElections.org
TransparentElections.org
We are NOT vendors of election systems
We are a team of like-minded IT practitioners who have implemented election-related projects in the past, using ICT
Option 2: Open Election System PC Encoding
1. Votes cast & tallied as in manual voting 2. ERs brought to school encoding (PC) center
CITY/MUNICIPAL BOARD OF CANVASSERS
PROVINCIAL BOARD OF CANVASSERS
3. ERs validated then posted on the web w/ BEIs digital signature
NATIONAL BOARD OF CANVASSERS
4. CMBOC will access database, produce SOV, COC 5. All interested parties may access and process the data by themselves
DOMINANT PARTY
DOMINANT OPPOSITION
CITIZENS ARM
MEDIA & OTHERS
VOTING CENTER
PRECINCTS
ENCODING CENTER
DOMINANT PARTY
DOMINANT OPPOSITION
CITIZENS ARM
MEDIA & OTHERS
6. All interested parties can send SMS to watchers to verify figures 7. PBOCs access DB; produce Prov SOVs and COCs 8. NBOC accesses DB
Open Election System PROs
Most transparent - voters and watchers observe tally at precinct level No need for voter training Once ER is encoded, result (web database) becomes accessible to the public Cost affordable at about P2B (Comelec only buys PCs/servers) PCs/servers can be passed on to DepEd after each election No storage concerns, because machines can be passed on to DepEd Ballot box
CONs
Manual tallying is tedious ERs will have to be encoded Looking for tens of thousands of encoders is a challenge Since it’s still manual tallying, public may think that election is not automated
If we adopt OES … This COMELEC would leave the legacy of making the Philippines probably the first country in the world to use technology to effect transparency in all steps of the election process.
How do we implement OES? COMELEC should … espouse the concept bid out the development of the system and computer programs make system/programs available to IT community and to public adopt good contributions make the system available to all interested parties, free of charge
Once the OES system has been developed, the COMELEC would only need to …
Bid out the PCs, servers, and the communications requirements
Bid out the management and implementation of the project
How do we implement OES? COMELEC should … espouse the concept bid out the development of the system and computer programs make system/programs available to IT community and to public adopt good contributions make the system available to all interested parties, free of charge
If OES is the best system for the Philippines because …
it is the least expensive it makes wholesale cheating extremely difficult to execute all steps are transparent to the voting public
… then what are we worried about?
There were disturbing statements coming from the CAC
"DRE was well-received but was seen by some as too expensive. OMR was cheaper but it still requires human intervention." "DRE is suited for areas where there is good infrastructure including electricity and connectivity. OMR is more suitable for rural areas where infrastructure isn't that reliable."
In its resolution dated 29 November 2008, the CAC recommended the following technologies:
Direct Recording Electronic (DRE) or Precinct Count Optical Scan (PCOS) technology for all areas, subject to the election automation budget of the COMELEC; Central Count Optical Scan (CCOS) technology for all other areas not covered by DRE or PCOS technology; and the public telecommunications networks for the transmission of results.
With all due respect to the CAC, we truly cannot comprehend why DRE remains an option. We cannot understand why there seems to be a compelling reason to use DRE.
It is the most expensive
Using it will be a logistical nightmare
It is the least transparent
IF it is illogical to employ an allDRE system, why then would there be logic in partly employing it?
Reality check … IF the COMELEC has already made the political decision to use DRE or OMR, then following is our recommendation …
Disallow the use of DRE (should be non-negotiable!)
OMR should be voting centerbased (school-based), NOT precinct-based (unless they can find an OMR that’s less than P20,000 each)
Option 2: Open Election System PC Encoding
1. Votes cast & tallied as in manual voting 2. ERs brought to school encoding (PC) center
CITY/MUNICIPAL BOARD OF CANVASSERS
PROVINCIAL BOARD OF CANVASSERS
3. ERs validated then posted on the web w/ BEIs digital signature
NATIONAL BOARD OF CANVASSERS
4. CMBOC will access database, produce SOV, COC 5. All interested parties may access and process the data by themselves
DOMINANT PARTY
DOMINANT OPPOSITION
CITIZENS ARM
MEDIA & OTHERS
VOTING CENTER
PRECINCTS
ENCODING CENTER
DOMINANT PARTY
DOMINANT OPPOSITION
CITIZENS ARM
MEDIA & OTHERS
6. All interested parties can send SMS to watchers to verify figures 7. PBOCs access DB; produce Prov SOVs and COCs 8. NBOC accesses DB
Open Election System - OMR
1. Votes cast & tallied as in manual voting 2. ERs brought to school encoding (OMR) center 3. ERs validated then posted on the web w/ BEIs digital signature
CITY/MUNICIPAL BOARD OF CANVASSERS
PROVINCIAL BOARD OF CANVASSERS
NATIONAL BOARD OF CANVASSERS
4. CMBOC will access database, produce SOV, COC 5. All interested parties may access and process the data by themselves
DOMINANT PARTY
DOMINANT OPPOSITION
CITIZENS ARM
MEDIA & OTHERS
VOTING CENTER
OMR PRECINCTS
DOMINANT PARTY
DOMINANT OPPOSITION
CITIZENS ARM
MEDIA & OTHERS
6. All interested parties can send SMS to watchers to verify figures 7. PBOCs access DB; produce Prov SOVs and COCs 8. NBOC accesses DB for final results
Some final words about DRE …
Peter Erben, who was a speaker in November’s Vendors’ Fair said that electronic voting machines are “..in its infancy with significant problems still facing their widespread use ...”
Kinstall Brace, president of Election Data Services, a company that examines voting machine usage across the country stressed that “From 2004 to 2006, electronic voting machine usage went up and 2006 was the high water mark. Then use came down. From 2006 to 2008, every jurisdiction that has changed has gone to optical scan … and election administrators are now moving their decisions in that direction.” Some parts of Florida and California, all of Connecticut, parts of New York and other jurisdictions around the country, switched from either DRE or lever machines to optical-scan systems. Comm. Rene Sarmiento
And we may become the laughingstock of other countries … for using a technology that is now, because of their sad experience in at least two elections, being discarded in the United States.
Please join …
TransparentElections.o rg … make our elections transparent!
Thank you!