Carl Bialik for The Wall Street Journal

Joey Chestnut may not have noticed, but Nathan’s Famous at Coney Island now reports calorie counts.

Projections of such laws’ impact — for New York, Los Angeles and all of California — are based largely on a study conducted by researchers from New York City’s public-health department, which critics note was based largely on Subway customers, who may not be typical of chain-restaurant consumers. Such objections were raised by an editor of a Centers for Disease Control and Prevention journal that rejected the study for publication, in an email that surfaced in the New York Sun’s coverage of a legal dispute over the New York law. “It’s not uncommon for scientific submissions to go through a couple of submissions to different journals,” said Lynn Silver, assistant commissioner for chronic disease prevention for New York City’s public-health department.

Projections also are based on the number of restaurants affected, which isn’t always clear. J. Justin Wilson, a senior analyst at the Center for Consumer Freedom, which opposes menu labels, cited government numbers suggesting just one in 15 meals nationwide would be affected. It depends in part on how and where you count: Advocates in New York have cited numbers from market-research firm NPD Group, which has found that about 35% of restaurant traffic in the city’s metropolitan area is in major chains. But that includes places not covered by the law. ReCount, NPD’s restaurant-count service, finds that 17% of restaurants in New York City ZIP codes have 20 or more units nationwide, and therefore would be covered by the proposed federal law. But these restaurants may be bigger, and serve bigger portions, than their independent rivals, and therefore have an outsize impact on public health. Moreover, none of these numbers addresses the question of whether the restaurant diners who change their eating habits are obese or at risk of obesity.

Why exclude independent restaurants? “Mom and pop restaurants produce a much more variable product, and a much higher cost, proportionally, so it makes sense to exclude them,” said Kelle Louaillier, executive director of Corporate Accountability International. “Ideally, we would have information at all restaurants, but we don’t want to put an untenable burden on truly small businesses.

Another aspect of the New York study that may limit its usefulness: It focused on only the meal at hand. Diners tend to compensate at later meals. Still, knowing the calorie count with which you’re compensating, or for which you’ll have to compensate later, is helpful. If you think you’re compensating for a 700-calorie order of fries, you might have a less healthy next meal than you would if you knew those fries contained 1,100 calories — and study subjects do tend to underestimate these numbers. “Students in our research have underestimated their own fast food consumption by well over 200 calories,” said Scot Burton, professor of marketing at the University of Arkansas’s business school.

“Those are issues that need to be looked at,” said Silver of New York’s public-health department. “As we understand better how consumers interact with information, it may give us better leads on how to perfect calorie labeling.”

Some advocates say the laws should exist even if they don’t reduce obesity. “It’s consumers’ right to know,” said Kelly Brownell, director of the Rudd Center for Food Policy and Obesity at Yale University. “I don’t think there’s a burden to the public-health community to have to show an impact.”

Rachel Johnson, a professor of nutrition at the University of Vermont who co-authored a skeptical paper about the effect of menu labeling, said, “There is a need for additional research to determine if ultimately providing calorie labeling in restaurants will help stop the progression of overweight and obesity among the American public. Until this evidence is available, I take a ‘do no harm’ approach to the issue and I do not see that any harm would be done to consumers by providing this important information.”

There’s also some dispute about how much consumers should be told. Philadelphia’s law added saturated and trans fat, sodium and carbohydrate counts to the numbers required on printed menus, and at the time the Center for Science in the Public Interest said in a statement, “We hope it is used as a model for other jurisdictions.” Now the center is backing the calories-only federal law, and Margo Wootan, director of nutrition policy for the center, said, “for many people, having simpler, more straightforward nutritional information is easier to use and understand.”

Philadelphia Councilwoman Blondell Reynolds Brown, a leading advocate of the city’s law that stands to be preempted, disagreed: “What is being done on the federal level would be a step forward for most of America, but it would be a step back for what we are working to accomplish in Philadelphia.” She added, “Stating calories alone, while helpful, is not enough to give families the best picture of what is contained in their purchases.”

And it’s unclear whether restaurants’ calorie counts provide the best numbers. Scripps-commissioned tests found big variations for some menu items. “The majority of restaurants that get into trouble with their declaration of calorie values do so because of serving size,” said Sandy Koch, vice president of Analytical Laboratories Inc., which conducted the tests for Scripps. “They want to have bigger and better meals.”

Mark Schrimsher, who runs the site CalorieLab, agreed. “The numbers are probably accurate for what was tested, but the portion size varies a lot in restaurant food, and it’s often not clear if sides and dressings are included,” he said. He added that outside of jurisdictions that require calorie counts, many restaurants don’t publish their numbers: “For instance, steakhouses tend not to, and casual-dining restaurants (table-service restaurants that serve alcohol) are very spotty.”

Taco Bell, though, disputed the findings by Scripps, citing a small sample size, unclear lab standards and a lack of verification that the items reported tested really were the ones tested. “Unfortunately, Scripps chose not to provide us with important information needed to determine how they arrived at their numbers, so we may never know why their numbers were off,” said spokesman Rob Poetsch.

Mark Mears, senior vice president and chief marketing officer of The Cheesecake Factory Inc., also stood by his company’s numbers. “We use independent third-party testing lab to verify that our nutritionals are indeed what we say they are,” Mears said. “Since we make everything from scratch, there is room for volatility in final nutritionals for a particular dish, since they all are made from scratch.”

Wilson said the law would put an undue burden on chain restaurants, which could react by changing their menus. “What it ultimately means is people will get more processed, pre-prepared meals,” Wilson said.

What do you think? Are calorie counts reliable? Would you take them into account when ordering? What impact might they have on public health? Please let me know in the comments.

Further reading: The Gotham Gazette and Newsweek also have examined menu-label laws.

One of these tools, Benford’s Law, is more commonly associated with accounting. Mark Nigrini, a professor of accounting at the College of New Jersey and popularizer of the law, says he has consulted with companies that have used it successfully. One example: It found too many 4s as leading digits in one employee’s expense accounting, which was caused by profligate expensing of daily Starbucks purchases. Of course, the law couldn’t account for whether that was an authorized repeat expense. (It wasn’t.)

Nigrini is skeptical about the application of the law to vote counts. “This really depends on how the fraud was committed,” Nigrini said. “If someone went in and rigged paper ballots, Benford’s Law will struggle to pick that up. Benford’s Law has potential if they blatantly made up the numbers.”

But why not make up numbers that conform to Benford’s Law? Aaron Clauset, a computer scientist at the Santa Fe Institute, said, “People are very strategic in their behavior, and they can use it to game the system. It sets up a coevolutionary arms race between the detector and the evader.”

Others are skeptical of the possibility of evading detection. “If people are actually cheating, they’re probably doing it in a hurry,” said Andrew Gelman, a statistician at Columbia University. “It may be more challenging to do an artificial set of votes that look realistic from all these perspectives,” added Walter Mebane, a pioneer of using Benford’s Law to test election results and a political scientist at the University of Michigan.

Mebane, a Democrat, first became interested in digging for fraud on November 8, 2000, when he recalls “walking around in a daze” after staying up late the night before and hearing Florida declared for Al Gore, then for George W. Bush, and then for neither. Later he went to Florida and, in 2003, he wrote a paper entitled, “The Wrong Man is President!” While he’s employed a variety of methods, Mebane was prompted to study the digits of election returns by the inquiry of a student in 2005.

Benford’s Law had already gotten a workout. Stephen Ansolabeher, a political scientist at Harvard, recalls using among other tests in an analysis of the 2000 U.S. presidential election. And in the wake of a a disputed Venezuelan recall referendum in 2004. Imre Mikoss, a physicist at Simón Bolívar University in Caracas, was suspicious of Hugo Chávez’s victory. “I saw this election and said, this is the opportunity to prove something with Benford,” Mikoss says now. He found that leading digits of vote counts didn’t conform to Benford’s Law, but his findings were questioned by the Carter Center. “The question is open,” Mikoss says now of his analysis.

Mebane has refined his technique, and in regular updates on his analysis of Iran numbers, he is careful to note what he does and doesn’t know. He took his analysis a step further, showing that the Benford anomalies stem from ballot boxes in Iran where the proportion of invalid votes was particularly low, suggesting that if there was tampering, it happened in these stations with ballots that should have been discarded. (The Iranian consulate to the United Nations didn’t return a call seeking comment.)

His usage of Benford focused on the leading digit of vote counts. Others studied the first digit, or the last digit. Mebane said he welcomed all the participation. “I think it’s great. Things move at light speed here,” Mebane said. “We’re not talking about waiting for papers to go through peer review. Nowadays it’s dueling blogs.”

One risk is that what seems like an anomaly is really the result of lots of analysts running lots of tests. “If you don’t specify ahead of time what you’re looking for, you’d be surprised if you don’t find rare events,” said Steven Miller, a mathematician at Williams College.

Bernd Beber, a political-science graduate student at Columbia University who, with Alexandra Scacco, studied final digits of vote counts, said criticism their findings were cherry-picked is wrong, because he based the chosen tests on prior work on Nigerian elections.

Arlene Ash, a research professor at Boston University’s medical school, finds all the hunting for data anomalies in Iran a bit unseemly. She said that the U.S. election system has plenty of flaws of its own. “One question that comes up rarely is whether we’re sure who won the election,” Ash said. “But something that comes up every day is where we have elections with massive amounts of problems.” Mebane added that the “worst country I’ve encountered” for getting election data is the U.S., in part because of the decentralized election system.

University of California, Berkeley, political scientist Henry E. Brady said he’d like to see some small fraction of the money spent on U.S. elections, perhaps a few million dollars a year, spent on election auditing.

What do you think? What methods are most effective for detecting election fraud? Please let me know in the comments.

Further reading: Many bloggers have crunched numbers on Iran election results. Benford’s Law has also been applied to Bernie Madoff and football. The New York Times spotlighted Benford’s Law in 1998, and USA Today did the same in 2000. One recent study analyzed which numbers people pick when falsifying data.

Clemson and U.S. News & World Report’s college rankings were embroiled in controversy earlier this month when a university staffer claimed the school was taking extreme, possibly manipulative measures to boost its ranking. Inside Higher Ed, which broke the story, followed up by getting the peer evaluations submitted by Clemson to the magazine. The university’s provost ranked 29 schools higher than her own, but the school president ranked Clemson No. 1 in its category, above top public and private schools. Clemson’s president, James F. Barker, said he was following the magazine’s instruction to rank by quality of undergraduate education. “I believe that Clemson does that better than anyone,” he said. Inside Higher Ed also reported on questionable numbers submitted by the University of Southern California, which boosted its engineering school’s ranking. Margery Berti, an associate dean at the engineering school, told Inside Higher Ed that USC was reviewing its numbers.

Fact-checking Web sites helped keep politicians and political candidates on their toes during the last national election cycle. One of these sites, PolitiFact, run by the St. Petersburg Times, won a Pulitzer Prize in April. Now it’s expanding its purview, taking on claims by pundits. Recent pronouncements by Keith Olbermann of MSNBC and Sean Hannity of Fox News — which like the Journal is owned by News Corp. — were rated false.

The Oscars voting system has deep flaws, according to voting-math experts I consulted. Now the Academy of Motion Picture Arts and Sciences is changing things, though not shaking up the underlying voting scheme. Ten films will be nominated instead of five, which could makes things more interesting, if not fairer.

As part of a look at popularity and how it influences people and perpetuates itself, I mentioned last month that menus highlighting certain items as most-popular boost those items’ popularity. In a pair of blog posts, the Baltimore Sun’s Liz Kay delves deeper into menu psychology, writing that a lack of dollar signs boosts spending and that placement on the page can influence sales.

Share a cab with a pair of mathematicians and the discussion about how to split the bill could take longer than the ride. You may also want to avoid taking that cab to a pie shop, because voting-math experts have concluded that it’s a lot harder to split a pie so that everyone gets a fair slice than it is to do the same with a cake.

Associated Press

Did Starbucks’s ubiquity help it win a Zagat poll?

The survey results caught the eye of Nate Silver, the baseball-numbers analyst turned political-numbers analyst whose site, fivethirtyeight.com, aggregated polls and other data to forecast President Obama’s electoral victory last November. Silver wasn’t sure how Zagat compiled votes in its online survey for best coffee, but figured that if the food-guide makers used their usual methodology, it might not mean that Starbucks is favored head to head over runner-up Dunkin’ Donuts, or third-place finisher Peet’s.

For Zagat’s city restaurant guides, diners submit their ratings to Zagat, which compiles the results. The more popular eateries get more votes. Since Starbucks and Dunkin’ Donuts are far more ubiquitous — about 7,000 U.S. locations each for Starbucks and Dunkin’ Donuts, compared to about 200 for Peet’s — far more votes were likely to have come from people who had tried the two big chains than their smaller rival. It could be that those who had tried all three preferred Peet’s, on balance, but their votes would be drowned out by the non-initiates to Peet’s, Silver noted.

I checked with Zagat, and it turns out the survey they used to rate fast-food restaurants was even tougher on the little guy than Silver imagined. “The designation of ‘Best Coffee’ was determined by how many people answered the question: ‘Which fast food chain has the best coffee?’ ” Nicholas Sampogna, a spokesman for Zagat, told me. Respondents were given a list of 30 chains that serve coffee, and the one with the most votes won. Since Peet’s has much lower name recognition — it has retail locations in only six states — it’s impressive that it won 9% of the vote, to 28% for Dunkin’ Donuts and 38% for Starbucks.

That the survey was more of a test of which chain has the most coffee-drinking consumers than which chain has the best coffee was reflected in the San Francisco results. There, where Starbucks’s edge in stores is narrower — roughly 66 to 23, according to the company Web sites — Starbucks beat Peet’s in the survey by a margin of just 50% to 30%.

Lara M. Wyss, a spokeswoman for Starbucks, told me it was “a bit of a false premise to assume that the number of our stores is the reason for the result,” noting that in other categories the winners didn’t always have the most stores. She added, “I’d like to think that the ranking reflects the quality of Starbucks coffee and the taste that millions of regular customers enjoy.”

Critics say the CBO numbers didn’t account for other planned pieces of health-care overhaul, something the CBO noted extensively in its report but that didn’t always make the headlines about it. “The best guess is there was some sort of crossed wire” between the CBO and the committee, according to Paul Van de Water, senior fellow at the Center on Budget and Policy Priorities, who objected to news coverage of the estimates.

Kennedy spokesman Anthony Coley said the numbers were no surprise. “We knew that this process would produce this estimate,” Coley said. “The preliminary analysis has highlighted the importance of filling in those blanks and we will fill them in our upcoming markup. Doing so will lower the bill’s overall cost.”

A Republican aide on Kennedy’s committee disagreed. “The CBO scores are a big warning sign,” the aide said. “They almost single-handedly made the Democrats just stop and give it a second look: What are we drafting? What are we pushing here?”

CBO numbers arise from a computer model the agency built to assess proposed health-care legislation. To devise the model, documented in an October 2007 report, the CBO consulted with outside experts, including Elizabeth McGlynn, associate director of RAND Health, whose site RAND:Health COMPARE analyzes various proposals. She’d like to see competing models share their assessments of the various proposals likely to emerge during debate. If they agree, that would provide some measure of confidence in their findings.

Modelers may take into account the experience of Massachusetts, which has taken steps toward universal coverage and has become a laboratory of sorts for observing consumer behavior in the face of new health-care choices. Jon M. Kingsdale, executive director of the state agency administering new health-care plans, said that for those new enrollees buying unsubsidized plans, about 70% chose the cheapest plan available to them. “We were surprised by the extent to which consumers migrated to our lower tiers,” Kingsdale said. Meanwhile, for subsidized plans, new enrollees chose on “breadth [of coverage], and then price.”

Further reading: I wrote in 2007 about health-insurance numbers. MSNBC previewed the CBO’s big role in the health-care debate, while Julian E. Zelizer argued on Politico that this big role is problematic. Eugene Steuerle, vice president of the Peter G. Peterson Foundation, an anti-deficit group, explains why CBO won’t credit certain proposals for cutting health-care costs. Here’s the latest estimate from the Census Bureau of the number of uninsured people in the U.S., and a comment from Census on the accuracy of the estimate, which often is significantly higher than estimates from states. Census is testing a new source for this data. A 1993 projection of health-care costs from CBO proved overstated. See more CBO reports on health. A report released Tuesday by the business-backed Employment Policies Institute attempts to quantify the voluntarily uninsured.

Jon Cohen, the Post’s polling director, explained the Post’s omission of these polls in a column last month. Both SurveyUSA and PPP use what’s known as Interactive Voice Response (IVR), in which survey respondents press buttons on a phone keypad, much like when interacting with an automated customer-service line. And the Post, like many other media organizations, doesn’t report the results of these polls, even after they proved useful in predicting last year’s presidential election.

“Unfortunately, polls that use recorded voice prompts typically take less care than polls conducted by live interviewers,” Cohen wrote. “Robopolls have limitations. People are generally less tolerant of long interviews with computerized voices.” Cohen added, “There’s a brewing battle among survey researchers about whether polls can be judged by how they perform in the final stages of campaigns. But until more research shows the validity of robopolls, we will continue to steer a careful path in our reporting.”

Tom Jensen, communications director of Public Policy Polling, responded in a blog post, pointing out that IVR polling has had a good track record in recent elections but also saying that the Post’s decision didn’t hurt much. “There’s been wideranging discussion of each of the polls throughout the Virginia blogosphere so basically this is just another case of a newspaper ceding their once upon a time authority to the internet,” Jensen wrote. “No complaints from us on that front.” After the race, he crowed about his firm’s relatively prescient final poll.

Cohen told me that comparing election results to final poll numbers shouldn’t be the sole way to evaluate polls. “I think it’s fundamentally wrong, but probably unavoidable.” He added, “We’re trying to fundamentally understand voters and campaigns, not just win what Gary Langer calls the horse-race lottery.”

Among Cohen’s complaints about IVR polls is that their response rate is lower than live-interview polls, perhaps because people are more prone to hang up on a recorded voice. When I ran that objection by Jay Leve, founder of SurveyUSA, he responded, “Polling exists at the intersection of words and numbers. Cohen understands neither.” Leve also passed along charts comparing his firm’s response rates in the 2000 2008 election with those of PPP and with the two live-interview pollsters active in Virginia (whose final polls showed a much tighter race). SurveyUSA and PPP’s response-rate numbers were comparable to one of those firms; the other declined to report its rates.

When I showed this chart to Cohen, he replied, “Response rate is just one of the components we use to gauge poll quality, and it’s not the main one. That said, Jay brings up the good point that there’s significant variation within mode, not simply across them. One of the things I hope to work together with him and others on is coming up with a set of criteria to distinguish among IVRs the same way we do among those who use live telephone interviewers.” He also said SurveyUSA “deserves great credit” for disclosing its response rates when some of its competitors don’t. (PPP also disclosed its response rates during last year’s election.)

The question of whether Nielsen helped its households obtain and install digital converter boxes occurred to me when I read about a spat between the measurement company and ABC over an alarming drop in ratings for the network’s “World News” program. “Some ABC affiliates switched over from analog to digital before the midnight deadline, and their viewership was not credited properly in Nielsen’s initial report,” the Associated Press reported. “The Nielsen recount found the audience for ‘World News’ last Friday was 6.34 million viewers” — compared to an initial count of 4.1 million viewers.

That’s still below the Friday average of 7.3 million this year, which could reflect popularity of that program among those Nielsen households who haven’t yet been able to get their digital signal up and running. I asked Gary Holmes, a Nielsen spokesman, if the missing 2% might lead to overall declines in TV ratings. He said it’s too soon to say. “It’s hard to compare last weekend to the weekend before because viewing has been declining week by week as we head into summer,” Holmes told me by email. “And as you note, the heaviest TV viewers are the most likely to have transitioned. So when you get down into the 2% range, an NBA playoff game one weekend vs. the next could have more impact on overall viewing than DTV.”

While it might seem odd to equip panelists’ homes with TV meters but not help out with digital-TV equipment, Nielsen deliberately didn’t intervene to maintain its panel’s representativeness of the overall TV-owning population (though there are other methodological issues). “We didn’t do anything to change the behavior of our sample households so that the sample would remain representative of the population as a whole,” Holmes said. “So, for example, if a sample home asked for advice on what to do to get ready for digital, we would just give them the address to the government Web site to check that out.”

Holmes noted that the proportion of Nielsen households without a digital signal is declining, and likely will continue to do so. Also some might be able to pick up analog signals from Mexico or Canada.

Some teachers see Wolfram Alpha as a tool liberating them and their students to focus on broader concepts, just as calculators obviated slide-rule instruction. It could also push math into a more visual realm and away from abstract notation, thanks to its plethora of graphs and charts. “Graphical aspect: I’m wondering how much mathematical notation will survive this big push of graphing and animation,” said Rich Beveridge, a math instructor at Clatsop Community College who has blogged about the new site.

However, Roger Howe, a math professor at Yale university, worries that the basics will be forgotten. “Mathematics doesn’t really become real unless students have a fairly direct contact with it,” Howe says. “Doing a reasonable amount of computation seems to be important for mastering mathematics.”

It’s unclear how much will be lost, and how much that matters. “One worries we’ll lose the underlying intuition,” said Donald Berry, chairman of the department of biostatistics at the University of Texas M.D. Anderson Cancer Center. “I worry about that, but I’m not sure how important that is. We as a species have come to a point where we can do things because of what others have built for us.”

Maria Andersen, a mathematics instructor at Muskegon Community College whose blog has hosted a lively discussion on Wolfram Alpha, is excited to use the new search engine in instruction. “I do see it as being a fantastic tool we can use to explore concepts,” Andersen said. “I can’t morally imagine walking into a classroom and having a student say, ‘Why do I need to take a cube of a binomial when Wolfram Alpha can do it for me?’ ”

She added that teachers would have to change their homework. “I would say that at least 50% of the standard homework problems and assessment questions that would be assigned with traditional textbooks could be answered by Wolfram Alpha,” Andersen said. “… find a math book, open it to a section of problems, and you’ll quickly find lots of examples that Wolfram Alpha will do.”

But Andersen worries about becoming reliant on a single online tool: “What if Wolfram Alpha disappears, after we all shifted to use it?”

“It’s very flattering if people care enough about one’s tools to be concerned about their longevity,” says Stephen Wolfram, the founder and chief executive of Wolfram Research. “What can one guarantee in this world?”

He sees his new tool as a way to broaden access to math and science. “The more people have ready access to knowledge and the more they have the power to do things like the experts do, the more they can feel empowered and get motivated to understand what’s going on,” he said.

Some calculation systems can work too much like a black box, though, according to Colm Mulcahy, a mathematician at Spelman College. “If [students] can get an instant answer, does it add to their understanding or make it so they’re just pushing buttons?” Mulcahy asks. “So many students are obsessed with calculators: 2+2 is 4 because that’s what the calculator says, and if the calculator said otherwise, they’d go with that.” He added, “Unless/until teachers — at all levels — teach and test more on the concepts (in addition to a certain level of computation), our students are doomed.”

All these comments are based on an assumption that Wolfram Alpha will improve. It’s already strong in math. The derivative of 5x is a fixed, universally agreed-upon quantity. And Wolfram Research has extensive experience doing such math with its Mathematica program, a fixture in many college courses.

The GDP of France, however, is continually updated and subject to revision. And the French government might decide tomorrow to add a column to the chart in its regular economic report, tripping up Wolfram Alpha’s effort to peel the numbers out. “Wolfram Alpha seems to be quite poor at doing what it claims to do well — namely, adding value to something like a search engine by being able to carry out computations to generate data that’s not already in place on someone’s web page,” said Jordan Ellenberg, a mathematician at the University of Wisconsin.

Some examples of failures offered by Ellenberg and others, both in math and in other fields:
What proportion of Americans live within three hundred miles of the ocean?
ARIMA (misses the statistical concept)
How long to double your money at 7%
Maximize -x^3+4x^2 on the interval between -1 and 5

Jeff Witmer, who teaches statistics at Oberlin College, says he’s already adjusted his curriculum to a more conceptual plane that wouldn’t be majorly affected by this new tool. “I expect that Wolfram Alpha could be used to help answer questions that I asked on statistics exams 15 years ago, when I had my students do a lot of calculating,” Witmer said. “These days I mostly ask students to interpret calculations that have already been made, which means that Wolfram Alpha would not be helpful.”

Phil Hanser, a statistician with the Brattle Group, criticizes the search engine for not reporting its uncertainty about statistical calculations. “If Alpha is meant to be a pre-eminent mathematics search engine, it should also serve as a model of good mathematics practice, including statistics,” Mr. Hanser says.

Other teachers noted that they’d already been using Mathematica in their teaching. “There’s not a lot new there, besides that it’s free as long as you have Internet access, which is not a small thing,” said Dan Teague, who teaches math at the North Carolina School of Science and Mathematics. (Surprisingly, Wolfram said sales of Mathematica have increased since the launch of the new tool, thanks to increased exposure of his signature software package.)
What do you think? How will this new tool affect teaching? How will you use it, if at all? What searches make it stumble? Please let me know in the comments.

Further reading: The Chronicle of Higher Education looked into Wolfram Alpha’s effect on education. Derek Bruff, a senior lecturer in mathematics at Vanderbilt University, launched a wiki to discuss this issue. The Journal’s Andy Jordan took it for a test drive.

I noticed the site recently via its Facebook application. Intrigued, I read its statement on accuracy, which with good humor says the site is “more accurate than a Magic 8-ball. Less accurate than distributing and collecting 300 million surveys.” The formula, most importantly, makes the dubious but necessary assumption that first and last names are independent, even though Juan Epstein probably is a less frequent combination than the frequency of “Juan” and “Epstein” might suggest. “The site may not be incredibly accurate, but we try to be up front about it,” says John Kramer, creator of How Many of Me.

The site also uses data from the 1990 Census, meaning it doesn’t reflect the latest trends in baby naming. Just two of the most popular names for American girls born last year — according to the Social Security Administration, which reports first names of babies but not first and last names of all living Americans — ranked in the top 100 among all female first names in 1990: Elizabeth (No. 5) and Emily (No. 99). Kramer writes on his site that “the 2000 census did not include name data.” That’s not quite right — the Census Bureau did report the frequency of last names in 2000, but not first names. (When I pointed this out to him, Kramer said he’d update the site to include that data.)

Why not report first names from the census? “We determined that the Census 2000 project would focus on surnames only, because of possible sensitivity perceptions on the part of the public,” a Census Bureau spokesman said. When I asked him to clarify, he said, “it was felt that if you released ‘analyses’ of both first names and last names, some people might question just how ‘confidential’ their census returns were (since this is where the data come from) — that is, even though the analyses would have been ’separate’ ” — meaning the analyses would have counted Juans and Epsteins but not Juan Epsteins. “No decision has been made on whether there will be a 2010 census surnames file,” the spokesman added.

The U.S. Census Bureau is unusual for even releasing as much information as it has, according to Kramer. “You wouldn’t believe how hard it is to find this type of data for other countries,” Kramer said. “I’ve contemplated expanding howmanyofme to include other countries, but the U.S. seems unique in releasing this kind of comprehensive genealogical data free of charge.”

The Century Council has argued that these drivers are the majority of those involved in fatal road accidents, and should be the target of enforcement and treatment. In its campaign, MADD supports tough penalties for first-time offenders whose blood-alcohol content is illegal but below the hard-core threshold. That difference of opinion has been reflected in debates over proposed legislation in California and Wisconsin mandating “ignition interlock,” devices to prevent a car from being operated if the driver has been drinking.

Robert Voas, a senior scientist at the nonprofit Pacific Institute for Research and Evaluation, said hard-core drivers are an elusive category. “The majority of crashes are due to individuals who would not be classified as alcohol-dependent, or alcohol abusers, but do drink,” Voas said. “They’re at a much lower risk per mile than the other group, but there are so many more of them, and they drive so many more miles.”

Some defense attorneys question the use of breath-alcohol tests to measure drivers’ sobriety, pointing to studies finding discrepancies between breath alcohol and blood alcohol (though several studies have shown that breath-alcohol tests understate the underlying blood-alcohol level, on average). Most experts say that the tests have improved greatly since their earlier days. “The accuracy of these devices is quite high,” said Paul Zador, a statistician who has studied alcohol-related fatalities. “They are quite reliable” — and are tested and approved by the National Highway Traffic Safety Administration. Also, some jurisdictions push for blood tests.

Clearly, the state of testing has improved from the days when drunk driving was assessed subjectively by the police, when there was “much more room for guilty people to get off, and for innocent people to be convicted,” according to David Hanson, a sociologist at the State University of New York, Potsdam, who maintains a Web site about drunk-driving issues.

Defense attorneys also have asked breath-test manufacturers for their devices’ source code, which they have resisted. Rankine Forrester, chief executive of Intoximeters Inc., which makes breath-alcohol testers, says he’d rather critics test his devices holistically than go on a “fishing expedition” in the programming code — one examination of a competitor’s source code turned up a number of problem areas. But he also says he’s refused to sell the devices to defense attorneys, which is “not in our interest.” Nonetheless, he says, “We’re a scale company. We don’t care if you’re fat or skinny. Just get on the scale and we’ll give you a number.”

MADD is touting a program that could lead to scales being installed, as a standard feature, in cars. Without such a feature, “there’s not always a connection between how you think you’re doing and what your level is,” said Anthony Liguori, a behavioral pharmacologist at the Wake Forest University School of Medicine. He noted a study finding that college students overestimate their drunkenness.

But some fear that driver awareness of their own blood-alcohol level might push some to drink more. “Without a meter, drivers tend to stop drinking well before they reach the legal limit,” said Erika Chamberlain, an assistant professor of law at the University of Western Ontario. “If they learn they could actually drink more before risking arrest, they might do so. Then again, the meters might also prevent some people from driving after they’ve had too many.”

Ms. Chamberlain and her colleague Robert Solomon, MADD Canada’s legal director, have written several papers calling for a lowering of Canada’s legal limit to 0.05. They note a survey of Canadian police officers showing that many don’t enforce the legal limit of 0.08 but instead create a de facto threshold of 0.10. “We don’t prohibit driving when you’ve drunk a lot,” Solomon said. “We prohibit driving when you’re blitzed.” (Drinkers “begin to show impairment at levels much below the legal limit,” said Vijay Ramchandani, a staff scientist at the National Institute on Alcohol Abuse and Alcoholism.)

Chuck Hurley, chief executive of MADD, said he is comfortable with a BAC limit of 0.08 in the U.S. “All the science indicates the crash risk starts to go way up at 0.08,” Hurley said. Though he noted that many other countries have lower limits, “Most other countries that have a limit of 0.05 have amazing mass-transit systems. That’s how people overconsume and get home.” The U.S. is, by contrast, car-reliant.

“I would have nothing against a 0.05 level,” said Susan Baker, a professor at Johns Hopkins University’s Bloomberg School of Health, who has studied the drinking patterns of drivers killed in crashes. “I would have nothing against making it illegal to have any alcohol in the system. But we’re a long ways from having that in the United States.”

Further reading: Telling people they’re drinking alcohol, and rewarding them for avoiding risk, can have surprising results. Passengers report more drunk-driving incidents than drivers do — and both report far more than the number of arrests. (I questioned self-reported drunk-driving rates last year.) Here are the Century Council’s and The Insurance Institute for Highway Safety’s rather different takes on the issue of hard-core drunk driving. The Century Council sponsors a site for drivers to consult before they drink. How Stuff Works explains breath tests.