ANNAPOLIS, Md. — Maryland officials plan to use the state's new touch-screen voting machines (search) in next year's presidential election, despite security flaws in the voting system targeted for fixes.

"We believe we are fully prepared to roll out the [touch-screen] machines for the 2004 presidential primary," said Gilles W. Burger, State Board of Elections chairman.

An independent study of the voting system recommended security fixes ranging from reprogramming part of the systems' software to creating a "chief information systems security officer" at the state elections board.

State Budget Secretary James C. DiPaula Jr. said he expects the report and the new security measures to restore public confidence in the new machines, which the state purchased in July from Diebold Election Systems (search) for $55.6 million.

"We're confident that this will safeguard the entire election process," DiPaula said.

Gov. Robert L. Ehrlich Jr. commissioned Science Applications International Corp. (search) to review the system Aug. 5 after a Johns Hopkins University study found it insecure. After SAIC submitted its results Sept. 2, the elections board worked with Diebold and the state Department of Budget and Management to develop an action plan.

Burger called it the "most robust" risk assessment study ever done in the country for an election organization.

"It seems that the SAIC report pretty much backed our report and came to a lot of the same conclusions," said Avi Rubin, an author of the Hopkins study and a computer science professor.

But Rubin said it's a bad idea for the state to rush forward with the new system until it's confirmed that the security holes have been plugged.

"If they're going to get this report and then go ahead with it anyway," Rubin said, "you have to question their motives for commissioning the report in the first place."

Technical improvements have been finished, Burger said, including encryption for unofficial results that are sent via modems, as well as programming measures that make it harder to create bogus "smartcards."

Other changes, ranging from an independent security review of Diebold software to hiring a security officer, will be implemented starting in October and continue through February and March of next year.

Many sections of the state report were deleted from the public version for security reasons, according to DiPaula. Some of the omitted information appeared to include interviews, as well as the kinds of encryption (search) and the version of Microsoft Windows the voting terminals use.

"The best security is for you not to give a road map to the people who want to do harm," DiPaula said.

"I was very concerned by the redactions in the report," Rubin said, noting that if the security problems truly had been fixed, there would be no risk in leaving them in the report.

"It gives the impression that they're hiding something from the public," Rubin said.

The Hopkins study, which analyzed Diebold voting machine software, said, "voters can trivially cast multiple ballots with no built-in traceability," and "the threats posed by insiders such as poll workers, software developers, and even janitors, is even greater."

Diebold responded that the researchers didn't consider how elections officials use the machines in actual elections.

The SAIC study addressed those problems by not only analyzing the software, Burger and DiPaula said, but considering how Maryland election workers use the equipment.

DiPaula said the difference between the two studies is like examining a car by starting with driver education, rather than merely looking under the hood.

Rubin and other critics had suggested printers might be attached to the voting terminals to create paper records for any possible recount, but the SAIC study counters that printed paper ballots might "still be subject to fraud."

"I stand behind the idea that a voter-verifiable audit trail is very important," Rubin said.

Problems with Florida's voting machines in the 2000 presidential election prompted Maryland to re-examine its voting systems and adopt a plan to create a uniform statewide election system by 2006.

Montgomery, Prince George's, Allegany and Dorchester counties already used Diebold machines in last year's gubernatorial election. Maryland counties are paying half the $55 million cost.

Baltimore City, which won't get the machines until 2005, also will pay half of its system cost.

"We're delighted with the [study] results," said Pat Maurer, community relations director for the National Federation of the Blind (search), a Baltimore-based advocacy group that supports the Diebold machines, which ease voting for the blind.

Organization members, Maurer said, "are all looking forward to voting with the Diebold machines."