hile playing bystander at a recount can be excruciatingly boring, it has its moments. For one thing, there's something reassuring about watching a trio of Democratic, Republican and Naderite observers intensely scrutinizing document after document and broadly agreeing with each other on the intentions of each voter. In addition, the monotonous seriousness of the undertaking is frequently relieved by evidence of the determined individuality of the American voter--the write-in votes for "God," the straight-ticket Republican voter who deviated only to write in Ralph Nader's name and the editorialist who left Bush's name alone but pointedly and emphatically crossed out Cheney's. Punctuating the hushed, at times reverent atmosphere of the counting hall in a nondescript corner room in New Hampshire's low-security Legislative Office Building is the occasional ejaculation "Object!" by an official observer who has found fault with an incorrectly or ambiguously marked ballot. The fate of these challenged documents is generally left to the seasoned eye of the secretary of state, in this case New Hampshire's William Gardner, a fourteen-term Democrat who is widely respected and appears studiously fair. Of course, the objectivity of the process will depend greatly on local conditions. In Florida 2000 and Ohio 2004, many questioned the neutrality of election officials who were also self-avowed partisans.

New Hampshire may not be typical of other states, in part because it is so small, but it has some important lessons to offer:

 § Transparency is the only way to go. New Hampshire resolutely refuses to use any ballot-counting technology that does not leave a paper trail; it will not use the "black box" touch-screens that have been installed throughout the country, and now record the votes of 29 percent of Americans. Thus every New Hampshire voter fills out a paper ballot. Only the counting mechanism varies--some localities use optical scanners from Diebold or another vendor, and others stick with hand counts. This means that as long as a candidate can be found to request a recount and someone can be found to pay for it, the citizens of New Hampshire can be fairly sure that their votes will be properly counted.

§ Third parties do play honest broker. Only a candidate can ask for a recall. With Bush having no incentive to do so, and Kerry having no interest in contesting the results in a state he won, there would be no advocate for accountability if Ralph Nader had not been in the race.

§ We should be deluged with statistics, but we aren't. New Hampshire is a distinct rarity, in that it posts election totals, right down to the precinct level, on the Internet--for anyone to see.

But New Hampshire is far from infallible. For one thing, thanks to a complicated formula reflecting Republicans' majority-party status, the GOP has in recent years always been listed first on the ballot. Hence, general election ballot choices come in the following order: -Straight-ticket Republican
-Straight-ticket Democratic
-Presidential race, Republican candidate
-Presidential race, Democratic candidate
-Statewide race, Republican candidate
-Statewide race, Democratic candidate
-Local race, Republican candidates (in multimember districts, in alphabetical order)
-Local race, Democratic candidates (in multimember districts, in alphabetical order)

Numerous studies have found that ballot sequence determines preference in enough cases to make a decisive difference, especially in close races. And as Paul Twomey, a Democratic activist and attorney handling a case about ballot order, put it to me, "A Democratic [state representative] candidate whose name starts with a W doesn't have a chance in hell of getting elected." That's an exaggeration, but only just. Randomly assigning ballot slots would certainly help eliminate this bias.

The straight-vote system is itself deeply problematical. When a person marks the "Straight Republican" or "Straight Democratic" option, all unmarked specific races are assigned to that party's candidate. Even if a voter marks a "straight" ticket and then marks several races for the opposing party's candidate, all unmarked spaces are assumed to be votes for the favored party. Officials I spoke to in New Hampshire conceded that, in all likelihood, many of those choosing the straight-ticket option hadn't read or properly absorbed the instructions, and so wrongly assumed they were merely indicating their party affiliation or registration rather than their intention to give all their votes to one party.

** When the results from the first two precincts did not bear out Ida Briggs's theory that something was amiss, Briggs wasn't the only bewildered party. Democrats and Naderites wondered how, in a state where Kerry did well, urban Democrats could have favored Bush--especially since these anomalous voters tended to have voted on Diebold equipment.

The best guess I heard was that although many of those urban Bush voters were Democrats, they were socially conservative, and many were Catholics who had been targeted by implicitly anti-Kerry letters from their bishop and leafleting campaigns in church parking lots. And, said State Senator Lou D'Allesandro, a Democrat from Manchester, despite the vaunted Democratic get-out-the-vote effort, the Republicans did a better job of getting out their base--even in urban areas, and even including sympathetic Democrats. Another factor was that some wards had substantial population growth since the last presidential election, and as long as the new residents were an unknown factor, it didn't make sense to assume anything about them. In other precincts, Kerry did fairly well among moderate Republicans who couldn't stomach Bush but who were not especially socially conservative.

Nonetheless, when the counters return after Thanksgiving, they'll still have some technical problems to resolve. The hand count of a third precinct showed roughly 100 fewer presidential votes than the optical-scan machines had, and will likely have to be recounted yet again. And in a fourth one, a local Republican candidate being recounted was awarded 105 more votes than he had before. Was the problem Diebold or somebody in the counting room? The answer will soon be clear.

What's already evident is this: As a country, we've never come to terms with the fact that the entire voting and tabulation process, electronic or otherwise, is rife with potential errors, many of them emanating from human beings, whether programmers or counters. Or that, imperfect and "inefficient" as they may be, hand-marked paper ballots offer the best possibility of finding those errors. And rebuilding trust in the process.