Computer tallies can't be trusted


First published: Sunday, July 26, 2009


Marge Acosta's July 10th argument ("Paper ballots an accurate system") rests on the unproven assumption that computerized ballot scanners will count those ballots accurately. Many of our nation's top computer scientists would take exception to that notion.


Scientists at the National Institute of Standards and Technology have advised the U.S. Election Assistance Commission, and Congress, that computer voting systems cannot be built to high levels of reliability or security. This leaves full or partial hand counts of paper ballots as the only way to ascertain the true winners of elections.


New York's rules for manual auditing of paper ballots are being written, but the standards that the state Board of Elections so far has agreed to are woefully inadequate. The state seems unwilling to look for vote-counting discrepancies that could easily change the winners of elections. Instead, it would have us all trust the unobservable computer software.


Unlike lever voting machines, certification and "accuracy testing" of stored-program computers is unreliable. Even if a computer passes such tests, we can not predict how the system will behave the next time it's used -- in a real election.


Software-based vote counting is fraught with potential for error, wildly more so than our lever voting system now in danger of being discarded. This immense yet subtle potential for error requires rigorous audits designed to find miscounted votes -- with the paper ballots under continuous observation. Without such auditing and observation, there can be no confidence in election results.


Howard Stanislevic


E-Voter Education Project

New York City


All Times Union materials copyright 1996-2009, Capital Newspapers Division of The Hearst Corporation, Albany, NY

Hearst Newspapers