http://seattletimes.nwsource.com/html/opinion/2001918663_bevharris04.htmlGuest columnist/ A simple way to make computer voting safer
By Bev Harris Special to The Times
Paperless touch-screen voting systems have triggered a controversy -invisible ballot systems may represent the biggest bamboozling in the history of voting. But even if we make vapor-ballot systems disappear, problems with computerized vote-counting will remain.
If we are going to use any form of computerized vote-tallying, we need to implement procedures to mitigate risks. One of the most important procedures, after a voter-verified paper ballot for auditing, is to post polling- place results.
In America, casting a vote is a private matter. The counting of the vote, however, has always been (and should remain) a public matter.
When we shifted from neighborhood-based to centralized counting, fewer eyes watched our votes. Then we privatized vote-counting altogether: Even when you vote on paper, your vote is counted by a computer, programmed by a private company, and your paper ballot is hardly ever examined. Instead, a computer interprets your vote. Our county officials are prohibited from examining the source code these
computers use, and it is the source code that tells the computer how to count your vote.
Even Washington's secretary of state does not examine source code. Nor does the state elections director. Instead, everyone in Washington relies on a small Alabama branch office of Ciber, Inc., where a motorcycle enthusiast named Shawn Southworth examines the source code.
Rather than direct examination, state and local officials rely on a ballot- sampling procedure called the "logic and accuracy" (L&A) test, saying it proves the machines count accurately. But we now know that at least 100 elections have been miscounted by these computers despite L&A tests.
Following a blistering report from Johns Hopkins and Rice university researchers, Maryland and Ohio commissioned independent studies of Diebold's software and found it to be riddled with problems. This software is used in King, Chelan, San Juan and Klickitat counties in our state.
Ohio commissioned an independent study of Sequoia's system, which found several critical security flaws. Sequoia's central tally software, it turns out, is even easier to tamper with than Diebold's. Sequoia is used in Snohomish County.
Diebold announced that it fixed the flaws, but it turns out that they were not corrected: A document called "release notes" details each change made in upgraded software, and Diebold's release notes for the system used in King County show that the flaws weren't corrected. A second report commissioned by the state of Maryland confirms that flaws still exist.
Sequoia has promised to correct its flaws, but has yet to provide any release notes to show that it has done so.
Computer enthusiasts can now verify these flaws for themselves. The Diebold central tally program is posted on a Web site run by California computer programmer Jim March, and the Sequoia program is now available on BlackBoxVoting.org.
Here's a simple remedy: Post the polling-place tallies in public, before the electronic votes are sent to central count, and match polling-place reports with the central count. Amazingly, Washington state does not require this, but county officials have the authority to do so, and we should demand it.
We vote at local polling places. Our votes are collected on electronic "ballot boxes," in the form of memory cards and cartridges. The information on these electronic ballot boxes is transferred to the county's central tally program. If someone switches the electronic ballot box (about the size of a credit card), or takes advantage of tamper-friendly features in the central tally programs, your vote can easily be changed.
Posting the polling-place tapes will be quick, easy and cheap.Diebold machines have an internal printer. Sequoia touch-screen machines have a port to which a printer can be attached. Both systems can print results at the polling place. This takes about 60 seconds and costs almost nothing.
Elections officials say that polling-place tallies won't match central tallies because they like to mix in other kinds of votes at central count, like absentee, provisional or challenge ballots. But vote-counting is just bookkeeping. If election officials commingle the data, they need to correct their bookkeeping procedures.
In Washington, we spot-check results. But Diebold's program has a specific flaw that survives spot checks even when the totals are wrong: Diebold's tally system uses two different sets of books - which don't have to match. In Sequoia's program, you can paste in vote-shaving code that will pass a spot-check while changing totals.
Some states, like Alabama, already require polling-place tapes. California Secretary of State Kevin Shelley has also directedcounties to post polling- place tapes. We should insist on polling-place printouts in Washington state.
Bev Harris, based in Renton, runs BlackBoxVoting.org, a national watchdog group that promotes auditable and secure voting. Harris is the author of "Black Box Voting: Ballot- Tampering in the 21st Century" (Talion Publishing).