Numbers Rule Your World Part 3

In his native Tampa, Bill Poe was well known as a former mayor and a gifted insurance man. He founded Poe & a.s.sociates in 1956, and under his leaders.h.i.+p, it became the largest insurance brokerage in Florida. In 1993, he negotiated a merger with Brown & Brown, the next largest compet.i.tor. Three years later, Poe retired, selling his shares for $40 million. But not for long did he stand on the sidelines. Partnering with his son, Bill Jr., he next created Southern Family Insurance Company, jumping from selling to underwriting insurance. As underwriters, the Poes took on the tasks of evaluating risks, setting premium amounts, and reserving surplus in order to cover claims.

At the time, national insurance giants like State Farm and Allstate were ruminating openly about quitting the Florida market after taking knee-bending losses from Hurricane Andrew in 1992. Having inherited hundreds of thousands of policies from failed property insurers, the state government realized it lacked sufficient capital to cover potential losses and was offering to pay start-up companies to "take out" and a.s.sume these policies. Poe was one of the early takers; on day one, Southern Family burst out with seventy thousand customers, who forked over $32 million of annual premiums to insure $6 billion worth of property, primarily in coastal areas, where the chance of a hurricane landing was the highest. For his trouble, the Florida state paid Poe over $7 million, or about $100 per policy transferred. In a startling move, Poe slashed rates by 30 percent for these customers, even as other take-out companies generally hiked prices, believing that the government had artificially capped rates before, due to political pressure. By positioning itself as a price leader and expanding through acquisitions, Poe Financial Group, operating under Southern Family and two other brands, emerged as Florida's biggest privately held property insurer by 2004, when it serviced 330,000 policies, collected premiums on the order of $300 million per year, and took on $70 billion of exposure. These results represented a phenomenal growth of 40 percent per year for eight straight years. What's more, the business produced over $100 million of profits since inception.

Then an ill wind named Wilma cleared the deck. Hurricane Wilma arrived at the tail end of two consecutive devastating seasons in which eight hurricanes battered the Florida coast. Poe Financial barely survived 2004: after paying out $800 million in claims, the company showed less than $50 million of capital on its balance sheet. In 2005, it frantically searched for fresh funds while also winning new customers to ama.s.s premiums. Neither Bill Poe nor others predicted that one horror would segue to another so that in twenty-four months, his customers suffered losses topping $2.6 billion. It was just a matter of time before Poe became insolvent. In 2005, the upstart lost more money than it had ever made in its ten-year history.

Pondering his misfortune, Bill Poe said, "This was the most unusual wind pattern in the world. The underwriter on that issue is G.o.d." Before 2004, few would have bet against Poe. With over forty years of experience, he knew the Florida market as well as anyone. Having served as Tampa's mayor, he had reliable connections and understood the regulatory environment. Almost all his customers were sourced from the state under its generous takeout policy. He then transferred most of his exposure to foreigners via reinsurance contracts. (Reinsurers insure the insurers.) As required by law, he used quant.i.tative models to demonstrate that the firm held sufficient surplus to cope with a 100-year storm. Poe evidently followed the rules, yet the company crashed. What went wrong?

By any measure, the 2004 and 2005 seasons were extraordinary. Charley, Frances, Ivan, Jeanne, Dennis, Katrina, Rita, and Wilma: these were ill winds that showed no mercy. The Atlantic basin endured the most storms, the most hurricanes, and the most destructive hurricanes, including the single most expensive (Katrina) and the strongest (Wilma) since records began in 1850. The losses suffered in 20042005 erased all the profits earned by insurers in Florida since 1993. In response, the national insurance giants began to rewrite the terms of business with their customers. Property owners with existing contracts were socked with harsh price increases of as much as 200 percent, dramatic reductions in coverage, sharp hikes in deductibles, or all of these. For example, Allstate b.u.mped rates by 8.6 percent in 2005 and 24 percent in 2006. In addition to cherry-picking the lower-risk customers, the insurers terminated outright at least half a million policies, while no other private insurer stepped forward to fill the void. When Florida prevented State Farm from charging an extra 47 percent in late 2008, the company announced its intention to abandon every one of its 700,000 customers. A sure symptom of dysfunction was when the market said no at any price. Industry leaders were acting as if hurricane risk in Florida was uninsurable.

In an editorial t.i.tled "A Failed Insurance Market," the St. Petersburg Times St. Petersburg Times concluded, "Hope and prayer for another hurricane-free season are the only things standing between Florida and financial disaster." The market was in a crisis-an existential crisis, to be exact. At this very moment, having witnessed the ravages of deadly storms, eager customers of hurricane insurance were easy to come by. Curiously, private enterprises no longer desired to play a role, and they called on the government to pick up the slack. The problem went far beyond the collapse of Poe Financial. If hurricane risk was insurable before, why did it suddenly become uninsurable? concluded, "Hope and prayer for another hurricane-free season are the only things standing between Florida and financial disaster." The market was in a crisis-an existential crisis, to be exact. At this very moment, having witnessed the ravages of deadly storms, eager customers of hurricane insurance were easy to come by. Curiously, private enterprises no longer desired to play a role, and they called on the government to pick up the slack. The problem went far beyond the collapse of Poe Financial. If hurricane risk was insurable before, why did it suddenly become uninsurable?

To get a satisfactory answer, we must first ask why anyone buys insurance and what keeps someone buying. Insurance is a thoroughly human response to the challenge of hazards: random events like hailstorms and rockfalls deplete wealth arbitrarily. People who believe in insurance contribute to a pool of funds in advance, to be drawn upon by the unfortunate afflicted. Insurance sidesteps the hopeless problem of predicting victims; instead, we let the lucky majority subsidize the unlucky few. This form of subsidy works because no member of a risk pool is immune to misfortune, and anyone can be either a donor or a beneficiary. In this sense, the arrangement is fair. By contrast, a progressive tax system or a bailout of Wall Street stipulates a subsidy scheme under which winners and losers are predetermined. Who would voluntarily partic.i.p.ate in such an arrangement, knowing one is almost sure to lose?

Automotive insurance is an example of a well-functioning market. In s.h.i.+fting their risks to insurers, drivers agree to pay premiums, most of which acc.u.mulate in a surplus account. Professionally qualified actuaries set the rates of payment at levels that should cover potential claims on average, as well as in any year. Beating the average is not good enough (witness Bill Poe); if and when payouts empty the cash vault, it is game over. Thus, a lot of care goes into designing risk pools. All members should face similar degrees of risk, or else some will tap the surplus more extensively than others. Small gaps in risk exposure are bridged by having less-safe drivers pay higher rates. Large gaps chase away safe drivers, who may feel they are paying more than their fair share, and also attract more risky drivers hoping to receive more than they put in. Since actual automotive claims have held steady over time, the actuaries are reasonably confident in their projections of future losses. In particular, they know that only a small portion of their policyholders will file claims in any given year; if everyone needed a payout immediately, the insurers would surely go bust.

Sh.e.l.l-shocked by the staggering losses in Florida, the hurricane insurers argued that they must have mispriced their product for years. The modeling firms that supplied them with estimates of expected losses were quick to second this view. Ernst Rauch from Munich Re, one of the two largest reinsurers in the world, reported, "Commercial modeling software of the 2005 generation put the annual loss expectancy for the USA Hurricane risk at $6 to $8 billion," which, in hindsight, ought to be described as catastrophically wrong. Based on this a.s.sumption, the insurance community requested annual payments of $10 billion from policyholders in Florida, only to suffer an actual loss of $36 billion in a mere two seasons. Warren Buffett, who knows a thing or two about the insurance business, cautioned against unfounded optimism: "Too often, insurers behave like the fellow in a switch-blade fight who, after his opponent has taken a mighty swipe at his throat, exclaimed, 'You never touched me.' His adversary's reply: 'Just wait until you try to shake your head.'"

Writer Michael Lewis, in a feature article in New York Times Magazine New York Times Magazine, vividly charted how quant.i.tative storm models rose to prominence in the mid-1990s after they had correctly predicted the unprecedented losses due to Hurricane Andrew. Their track record since then has been rather unglamorous. After admitting that it significantly underestimated the impact of the 20042005 hurricanes, Risk Management Solutions, the leading modeling firm, made fundamental changes to its methodology, incorporating the judgment of a panel of experts. Many of these scientists opined that climate change has rendered Atlantic hurricanes more ferocious and more frequent. With each refinement, the modelers lifted their estimates of future losses, which justified higher and higher premiums. By the late 2000s, the insurance industry a.s.serted that to have a chance of covering projected losses, companies would have to set annual payments above what customers could reasonably afford.

Another reason why the industry grossly misjudged the level of risk was its fixation with the "100-year" storm, commonly defined as a storm of such power that it will appear only once every 100 years. In Florida, all underwriters must prove that their capital structure can withstand a 100-year storm. This rule appeared watertight as even Hurricane Andrew, which cost insurers $17 billion in 1992, was a 60-year storm. (A 100-year storm, while producing more damage than a 60-year storm, is less likely to occur.) It was unimaginable for two once-in-a-century disasters to happen in adjacent years.

Statisticians tell us that the concept of a 100-year storm concerns probability, not frequency; the yardstick is economic losses, not calendar years. They say the 100-year hurricane is one that wreaks more economic destruction than 99 percent of hurricanes in history. In any given year, the chance is 1 percent that a hurricane making landfall will produce greater losses than 99 percent of past hurricanes. This last statement is often equated to "once in a century." But consider this: if, for each year, the probability of a 100-year hurricane is 1 percent, then over ten years, the chance of seeing one or more 100-year hurricanes must c.u.mulate to more than 1 percent (a simplified calculation puts the risk at about 10 percent). Thus, we should be less than surprised when powerful hurricanes strike Florida in succession. The "100-year" hurricane is a misnomer, granting us a false sense of security.

From the industry's perspective, the 20042005 disasters exposed the inadequacy of the prevailing insurance rates, which relied on wayward projections of the frequency of storms and the intensity of losses. The insurers also discovered that their customers could not bear the full cost of the insurance, so they could see no profit motive, and thus the insurance market failed.

Statisticians have something else to add to this story: Natural-disaster insurers, unlike automotive insurers, have no choice but to accept risks that are concentrated in vulnerable geographies. This agglomeration of risk became more and more severe as the existing risk pools disintegrated after the 20042005 seasons.

That Poe Financial shuttered in Wilma's wake had much to do with its ill-advised concentration of risks in South Florida. The two infamous seasons yielded 120,000 claims. This is to say, almost two out of five Poe customers dipped into the surplus at the same time at the same time. Such a ma.s.sive coincidental draw on liquidity would have sunk any insurer. For this, Poe had only its flawed vision to blame. The take-out deals with the state government bloated its accounts with customers from previously failed insurers, chiefly coastal properties with the greatest chance of damage. Worse still, by competing on price, Poe violated Warren Buffett's first principle of underwriting. He once advised that if "winning" is equated with market share rather than profits, trouble awaits; "no" must be an important part of any underwriter's vocabulary. Even the last-ditch expansion effort after the 2004 hurricanes hastened Poe's demise: with new customers came more revenues but also more exposure and further acc.u.mulation of risk, compounding the original problem.

Because natural disasters strike locally, these risk pools are much less spatially diversified than those a.s.sembled by automotive insurers. Nonetheless, the principle of insurance still applies: these pools must contain some customers who would not require payment during hurricane seasons, or else simultaneous claims would bury the insurers. Traditionally, such customers included policyholders in inland Florida, other states, and foreign countries. To hear it from Bob Hartwig, the industry economist, because all Floridians belonged to the same risk pool, trailer-park grandmas were subsidizing wealthy owners of estates by the sea. Meanwhile, large national insurers borrowed surplus from other lines of business and income from other states to pay claims in Florida. When primary insurers transferred risk to reinsurers, they in effect bought the right to utilize pools of funds contributed by policyholders in other countries such as those insuring against windstorms in Europe or earthquakes in j.a.pan. This means whenever a hurricane strikes Florida, the Europeans and j.a.panese effectively subsidize U.S. reconstruction projects.

Signs after 2005 indicated that none of the three groups would stick around, and for good reason. The skewed statistics of past mega-catastrophes revealed the inconvenient truth about the winners and losers of the existing insurance arrangements. Of the ten most costly disasters between 1970 and 2005, the top eight occurred in the United States, producing over 90 percent of the $176 million total insured losses; the other two included a typhoon in j.a.pan and a windstorm in Europe. Six of those eight American disasters were Atlantic hurricanes, all of which pa.s.sed through Florida. Towers Perrin, the global management and actuarial consultancy, estimated that the United States accounted for half of the premiums but three-quarters of the losses in the London reinsurance market. If this level of imbalance persists, other partic.i.p.ants will perceive the arrangement to be unfair.

Already the market has adjusted to the new reality in various ways. The national insurance giants created Florida-only subsidiaries known as "pup companies" to protect their corporate parents, signaling their reluctance to share surplus across state lines. Reinsurers doubled rates they charge hurricane insurers based on the view that hurricane intensity and frequency have grown irreversibly. Unless the recent trend of foreigners subsidizing American disaster claims is reversed, it is hard to see how foreigners will stay in the reinsurance schemes without demanding eye-popping rate hikes, if at all.

Florida's inland residents are also likely to bail out of the risk pool, now that its inherent inequity has been laid bare. In the case of Poe Financial, 40 percent of the customer base cleaned up ten years' worth of surplus within two seasons. When coastal and inland properties are treated similarly, every policyholder pitches in the same premium, based on the average average exposure to risk. In reality, inland properties have a much lower exposure than coastal buildings, so the low-risk residents are subsidizing the high-risk coast dwellers. As partic.i.p.ants have widely varying exposure, the winners and losers in this risk pool are predetermined, and not surprisingly, the low-risk group of customers rejects this subsidy structure as inequitable and illegitimate. Undifferentiated risk pools could not hold together because of large group differences; as unhappy customers decamped, the remaining risk became more and more concentrated. exposure to risk. In reality, inland properties have a much lower exposure than coastal buildings, so the low-risk residents are subsidizing the high-risk coast dwellers. As partic.i.p.ants have widely varying exposure, the winners and losers in this risk pool are predetermined, and not surprisingly, the low-risk group of customers rejects this subsidy structure as inequitable and illegitimate. Undifferentiated risk pools could not hold together because of large group differences; as unhappy customers decamped, the remaining risk became more and more concentrated.

Just as the Golden Rule lawsuit prompted developers of standardized tests to start treating high-ability and low-ability students as distinct groups, in the aftermath of the jaw-dropping losses from 20042005, the insurance industry reacted by splitting the risk pool into two, a strategy statisticians refer to as stratification. After all, one can forcibly mix oil and water temporarily, but they eventually separate into layers. The insurers now fight for low-risk customers while ceding the high-risk coastal properties. How are the two groups faring under this new arrangement?

Recall that Poe Financial Group was the poster child for the take-out program initiated by the Florida government after 1992 to divest high-risk policies. The government's buildup of 1.5 million accounts was reduced to 815,000 by February 2005. Almost all of Bill Poe's customers came from this incentive program. With Poe's insolvency, it was a case of return to sender: in one swoop, Citizens Property Insurance Corporation, the state-run caretaker, added 330,000 customers to its roster. Citizens was ill placed to handle this influx: it was about half a billion dollars in the red after the 2004 hurricane season, and the fiscal deficit ballooned to $1.7 billion by the end of 2005, even before absorbing Poe.

Having announced its conviction to bail out the Poe customers, the government was forced to raise the payout amount year after year as closed cases were reopened and new claims materialized. By late 2008, Citizens was collecting $4 billion of annual premiums but covered exposure worth $440 billion. The ent.i.ty would have folded if it were a private concern. Instead, Florida legislators plugged the hole by levying a series of "a.s.sessment fees." After appropriating $920 million from the 2006 state budget, they still had to charge every buyer of homeowner or automotive insurance in the state a fee equal to 6.8 percent of premiums. For those with property insurance, this came on top of a 2 percent levy. Further, a charge of 2.5 percent would then kick in for ten years. In 2007, another 1.4 percent had to be a.s.sessed. In addition, having raised nearly $2 billion by selling bonds in 2006 and 2008, the state covered interest payments by taking another 1 percent of premiums for the next eight years. A reporter reminded her fellow Floridians in 2008, "That sucking sound you hear is a 3-year-old hurricane pulling money out of your wallet."

In a time-honored game of pa.s.sing the buck, the unwanted risks were shoved from owners of seaside properties to private insurers to state-run Citizens to Bill Poe back to Citizens and finally to Floridians. For those living on the coast, nothing has changed: every year, they wish for a quiet hurricane season. When the insurers raise rates, they put up with it; when one insurer takes off, they scramble to another. As for the inland residents, they subsidized the coastal customers when private insurers used to place both groups in the same risk pool; now the state government takes their money via a.s.sessments and sends it to the coast. Same difference.

In an ominous development, the state government rightly got scared of the possible insolvency of Citizens should another mega-hurricane strike Florida but then promptly repeated its mistake of enticing start-up companies to take out policies from the caretaker agency. From 2006 to 2008, 800,000 policies were taken out and taken up by bit players who received "D" (weak) or "E" (very weak) grades from rating agencies, due to inadequate capital and inexperience. The state regulator took a farcical position: "Could there be another monster like Katrina or worse? G.o.d forbid if there were. All bets would be off at that point." We've seen this script before, and it did not have a happy ending.

And the troubles do not end there. One might expect the devastation wrought by Hurricane Wilma to have deterred Floridians from moving to the coast, but the trend has showed no sign of easing. Economists blame "moral hazard": coastal dwellers are less worried about hurricanes because they expect the state to continue the unending bailouts and offer an implicit backstop. New residents believe that if their homes should fall, someone else will pay for the reconstruction, so why not enjoy living along the coast? This alarming trend further aggravates the economics of hurricane insurance. The already heavy concentration of simultaneous risks has gained even more heft. The increased population density inflates real estate prices so that the number of risks multiplied by the value of each exposure has expanded by leaps and bounds. Many scientists consider the "lemming-like march to the sea" as a much more serious threat than the recent rise in intensity and frequency of hurricanes. By 2006, Florida owned $2.5 trillion of insured properties, the largest value of any state, eclipsing even New York. Storm models predicted that a hurricane of Katrina's strength would produce upwards of $100 billion in losses if it traversed the Miami area, repeating the disaster of 1926. While this threat looms, the insurance market in Florida stands paralyzed.

The harrowing hurricane seasons of 20042005 awakened the disaster insurance industry to an essential reality: under existing risk pools, customers with low-risk inland properties were sure losers, and those with high-risk coastal properties sure winners. This group difference threatened the viability of the insurance arrangements because the cross-subsidies no longer appeared fair. The big insurers reacted by imposing stunning rate hikes, especially on the high-risk group, in effect shutting them out. When the state regulator objected, they relinquished the entire market. Inevitably, the state of Florida a.s.sumed the role of insurer of last resort, which did nothing to relieve the low-risk group from subsidizing the coastal property owners. If the state must play such a role, then it must provide incentives to slow the migration of people and wealth to the vulnerable coastline. If the state cannot or will not stop the unfair cross-subsidies, it must at least respect the low-risk residents by working to ease their burden. When past lessons are not learned, the next disaster is only a matter of time.

On the surface, the insurer and the test developer have nothing in common: one a.s.sembles a profitable roster of customers, while the other constructs a fair set of test questions. From a statistical point of view, they both have to grapple with the issue of group differences. In both cases, variability across groups was seen as undesirable. Test developers made a breakthrough when they understood that they could not directly compare black students with white students; by contrast, an insurance market verged on disintegrating once insurers realized that they could not treat every customer as the average customer. The crucial decision in these situations is whether or not one should aggregate groups. That is the dilemma of being lumped together.

4.

Timid Testers / Magic La.s.sos The Sway of Being Asymmetric I want to talk about the positive, not the negative, about this issue.

-MARK M MCGWIRE, PROFESSIONAL BASEBALL PLAYER You don't even have to know what it is you're searching for. But you'll know it once you find it.

-CRAIG N NORRIS, CEO CEO OF OF A ATTENSITY Awatershed occurred in the early 2000s, when the baseball players' union finally acquiesced to a steroid-testing program. Baseball fans were losing faith in the integrity of the national pastime, and the scandal was fueled by a pair of sensational books: Juiced Juiced, in which All-Star slugger Jose Canseco exposed himself as the "G.o.dfather of Steroids" and outed several well-loved ballplayers as dopers, and Game of Shadows Game of Shadows, in which two San Francisco Chronicle San Francisco Chronicle reporters laid bare the federal investigation of BALCO, a California supplier of steroids to many elite athletes, including baseball players. It was no longer possible to deny that performance-enhancing drugs had infiltrated the sport, just as they had cycling and track. reporters laid bare the federal investigation of BALCO, a California supplier of steroids to many elite athletes, including baseball players. It was no longer possible to deny that performance-enhancing drugs had infiltrated the sport, just as they had cycling and track.

Drug use is harmful to the athlete and makes a mockery of the sporting spirit. Most sports have adopted an anti-doping code written by the World Anti-Doping Agency (WADA), requiring athletes to submit urine or blood samples for testing. Baseball's steroid-testing program, however, does not meet the more stringent international standard, as grudging players opt to take one small step at a time. For example, Major League Baseball (MLB) does not test for human growth hormone (HGH), a powerful drug that burst on the scene in the late 1990s. Boston Red Sox star Mike Lowell explained why: "[HGH testing] has to be 100 percent accurate, because if it's 99 percent accurate, there are going to be seven false positives in big league baseball, and what if any of those names is one of the major names?"

Baseball salary statistics fanned Lowell's uneasiness: the average MLB player earned almost $2.5 million in 2005, and Lowell's team, the Red Sox, was one of the league's wealthiest, paying more than $4 million per athlete. With so much money at stake, it is no wonder that ballplayers are worried sick about false-positive errors in steroid testing-about drug-free athletes who get falsely accused. However, with the focus squarely on this one aspect of testing, the anti-doping community has unintentionally abetted the drug cheats. In this chapter, we will learn how.

A variation of the steroid detection problem disquieted American troops stationed in Iraq and Afghanistan: how to screen large numbers of local job applicants for past, present, or future a.s.sociation with insurgency. At Camp Cropper in Iraq, David Thompson led a team of interrogators who relied on their training, real-world experience, and gut instinct to "filter truth from fiction." This lie detection process proved imperfect as insurgents continued to target Thompson's soldiers. With each phase of the war, he noticed that the recruits arrived younger, less experienced, and thus less prepared for immediate success in the counterintelligence work.

Imagine the thrill when he received a s.h.i.+pment of portable lie detectors in 2007. The portable lie detector is a handheld computer with fingertip electrodes for measuring skin conductivity and pulse rate. An operator asks the subject a list of yes/no questions and taps in the answers; within minutes, the computer processes the data from the electrodes and renders a verdict: green for truthful, red for deceptive, and yellow for inconclusive. Like the conventional polygraph, this miniature version actually detects anxiety, which can be induced by either the act of lying or the fear of getting caught lying, depending on which expert we believe. Unlike users of the conventional polygraph, Thompson's interrogators no longer require expertise or real-life experience; the computer program, optimized by physicists at Johns Hopkins University, removes the human element from counterintelligence screening. Each portable lie detector costs $7,250 plus an annual maintenance fee of $600.

With many American lives at stake, it is no wonder that army leaders worry about false-negative errors in interrogations-about insurgents who evade detection during screening. They instructed the Johns Hopkins researchers to calibrate the gadget so that those who receive green lights have virtually zero chance of having lied. However, statisticians tell us that by focusing exclusively on this one aspect of accuracy, the technologists have unintentionally impaired the ability of this gadget to pick out potential suspects, its primary function. In this chapter, we will learn how.

Steroid testing and lie detection are both technologies that cla.s.sify people into types: dopers versus clean athletes, liars versus truth tellers. Critics fault drug tests for destroying careers (false positives) and polygraphs for missing potential criminals (false negatives). Statisticians point out that such technologies face an undesirable but also unavoidable trade-off between the two types of errors. Any detection system can be calibrated, but different settings merely redistribute errors between false positives and false negatives; it is impossible to simultaneously reduce both. As an a.n.a.logy, consider a baseball hitter of average skill level: he can swing the bats more aggressively, in which case he will register more strikeouts, or less aggressively, in which case he will hit fewer home runs. More importantly, some errors are more visible and more costly than others. Such asymmetry provides strong incentives for drug testers and law enforcement officers to focus their energies on one type of error while the other aspect is neglected and unnoticed. The sway of being asymmetric is the subject of this chapter.

Mike Lowell, the professional third baseman, was considered toast in Florida before he became the toast of Boston. When the Florida Marlins traded Lowell to the Red Sox in 2005 after two consecutive under-par seasons, they felt the veteran player had seen his best years, and eagerly unloaded his $9 million annual salary to the Boston club. They could scarcely have imagined that Lowell would be named Most Valuable Player in the 2007 World Series, capping off a banner year in which he broke personal records in hits, runs batted in, batting average, and OPS (on-base plus slugging percentage) at the somewhat-old age of thirty-three. Thanks to the one scintillating season, the Red Sox rewarded Lowell with a hefty pay hike to $12 million a year. Mike Lowell's climb from the humble son of a Cuban exile and a cancer survivor to the top flight of baseball was the substance of American dreams.

For Major League Baseball, 2007 was a year in which dreams threatened to turn into mirages. While Lowell's performance on the field exhilarated Boston fans, baseball was sagging under the weight of a steroid scandal that could expose some of the game's biggest stars as frauds. The seeds were sown in 1998, when Mark McGwire and Sammy Sosa captivated crowds with a superhuman, down-to-the-wire home run chase, both of them surpa.s.sing Roger Maris's record, previously regarded as sacrosanct. The cynics whispered about rampant abuse of steroids by ballplayers inflating their statistics artificially.

Then tantalizing clues started to emerge. A bottle of androstenedione, a stimulant banned in the Olympics, was spotted inside McGwire's locker. When Barry Bonds shattered McGwire's record three years later, skeptics noted that the thirty-seven-year-old appeared bigger, stronger, and better than his younger self. Old-school baseball watchers wanted to embellish Bonds's record with a big black mark even before any proof came to light. Then, in 2003, federal investigators found physical evidence that a California lab, BALCO, had supplied performance-enhancing drugs to elite athletes. The used syringes, doping schedules, and client records entangled numerous ballplayers, among them super-star Bonds. Bonds later admitted to using two substances from BALCO, known as the "clear" and the "cream," but maintained he thought they were flaxseed oil and antiarthritis balm. Top track coach Trevor Graham anonymously sent a syringe of the "clear" to anti-doping authorities in a sacrificial act to knock out a rival team of athletes he knew to be fellow BALCO clients, leading to its identification as THG (tetrahydrogestrinone), a designer steroid engineered by chemists to be undetectable by testing labs.

Next, 3,000-hit man Jose Canseco blew the lid off when he alleged in his sensational 2005 book, Juiced Juiced, that four out of every five ballplayers, including McGwire and Jason Giambi, used steroids. Later that year, prompted by President George W. Bush's State of the Union address, legislators conducted congressional hearings, which were most notable for McGwire's refrain, "I am not here to talk about the past. I want to talk about the positive, not the negative, about this issue." Another star slugger, Rafael Palmeiro, declared to Congress, "I have never used steroids, period. I don't know how to say it any more clearly than that. Never." Six months later, he tested positive for stanozolol, the same steroid found in sprinter Ben Johnson's urine at the 1988 Seoul Olympics. In 2007, young Rick Ankiel provided baseball fans with an unlikely feel-good story: the pitcher for the St. Louis Cardinals, who in his rookie season had inexplicably lost his ability to throw strikes during the World Series, resurrected his career by winning a starting job as a major-league hitter. The excitement faded, however, when an investigation linked Ankiel to a Florida clinic suspected of distributing HGH to professional athletes (the league did not discipline him).

All of these threads were expected to converge in December 2007, when Senator George Mitch.e.l.l issued his report on the steroid scandal. The nature of the evidence was as yet unknown, and speculation was running rife that scores of players would be named and vilified.

Such was the backdrop when Mike Lowell spoke to the Boston chapter of the Baseball Writers a.s.sociation. As befitted a college graduate in finance, he gave a succinct, a.n.a.lytical rationale for why the players, led by union chief Donald Fehr, have long vacillated on the issue of steroid testing: "[HGH testing] has to be 100 percent accurate, because if it's 99 percent accurate, there are going to be seven false positives in big league baseball, and what if any of those names is one of the major names? You've scarred that person's career for life. You can't come back and say 'Sorry, we've made a mistake,' because you just destroyed that person's career."There's got to be 100 percent accuracy, and that's why Donald Fehr puts himself in a position where he's responsible for the seven false positives, not the 693 that test OK. Because, G.o.d forbid, what if it was [Hall of Fame shortstop] Cal Ripken, you know what I mean? Doesn't that put a big black mark on his career? That's where I think the union has to make sure the test is 100 percent, no chance of a false positive. Some people have said 90 percent [accuracy]. That's 70 [false positives]. That's three full rosters."

A false positive occurs when an athlete who did not cheat is falsely found to have cheated. "Being called a cheater, knowing I didn't cheat, it's the worst feeling in the world," opined Tyler Hamilton, the American cyclist who famously and valiantly endured a painful collarbone injury to finish fourth in the 2003 Tour de France. Adored by cycling fans as a straitlaced "Boy Scout" and respected by colleagues as an all-around nice guy, Hamilton was poised to emerge from the shadow of Lance Armstrong, his mentor, in 2004. His income topped $1 million, including salary and endors.e.m.e.nt deals with Nike, Oakley, and other brands. In Athens that year, he became the first American to win an Olympic road race in twenty years.

Suddenly and cruelly, his world came cras.h.i.+ng down as the drug testers at the Tour of Spain found foreign blood mixed with Hamilton's. The outlawed practice of transfusing another person's blood cells into one's blood, called blood doping, raises the amount of oxygen carried in one's bloodstream, a key driver of cycling ability. Subsequently, it was disclosed that the "A" sample collected from Hamilton in Athens had also raised a red flag, but the rules barred further action when the lab technician inadvertently froze the "B" sample. (To increase accuracy, testing labs divide blood or urine samples into "A" and "B" halves and declare a positive only when both parts test positive.) Defiant, Hamilton demurred, "We don't have an answer to why there was a positive test at the moment. If I knew the answer, I'd have it tattooed on my arm. The bottom line, I didn't blood dope." His lawyers advanced the possibility that the foreign blood belonged to a "vanis.h.i.+ng twin," an abortive twin brother who might have shared a womb with Hamilton.

Hamilton's defense was as standard as it came for accused athletes: he never cheated, so the positive finding must have been a false false positive, which could be explained by alternative causes, such as vanished twins. The arbitrator eventually rejected Hamilton's appeal, and he served a two-year ban from cycling. If he were indeed a clean athlete, his plight would justify Mike Lowell's worry about false positives ruining the supernova careers of elite athletes. positive, which could be explained by alternative causes, such as vanished twins. The arbitrator eventually rejected Hamilton's appeal, and he served a two-year ban from cycling. If he were indeed a clean athlete, his plight would justify Mike Lowell's worry about false positives ruining the supernova careers of elite athletes.

Accused athletes have had trouble shaking off the stigma of testing positive-and for good reason. Travis Tygart, CEO of the U.S. Anti-Doping Agency (USADA) once observed, "Denial is the common currency of the guilty as well as the innocent." Such behavior is consistent with their incentives. The truly innocent athlete, upon failing a drug test, should immediately hire a lawyer, find character witnesses, seek alternative explanations, challenge the lab procedures, and fight tooth and nail. The guilty athlete frequently pursues the same course of action while voluntarily withdrawing from compet.i.tion. At worst, he loses the argument and gets suspended for two years, backdated to when he ceases competing, disregarding when his final appeal falls apart. If lucky, he may find exoneration via a botched procedure, a spoiled sample, or a sympathetic arbitrating judge. By a strategy of denial, the guilty athlete spends most of the suspension clearing his name with a remote chance of redemption; if he instead chooses not to contest the positive finding, humiliation arrives instantly, but reinstatement no sooner.

Consequently, the true positive and the false positive, once commingled, are difficult to set apart. That is why millionaire athletes like Mike Lowell demand nothing less than a test that is "100 percent accurate, [with] no chance of a false positive."

If only real life were so perfect.

Statisticians say even if a test committed zero false positives, it would be far from "100 percent accurate" because of false-negative errors. Athletes only complain about false positives; the media wax on about false positives. We are missing the big story of steroid testing: false negatives false negatives.

In an impa.s.sioned letter written in his own defense, cyclist Tyler Hamilton a.s.serted, "I have been tested over 50 times throughout my career and this is the first time I have ever even been questioned." Actually, entire pelotons have been pedaling under a stubborn fog of suspicion ever since a string of sensational drug busts. .h.i.t the sport's marquee event, the Tour de France, in the 1990s and 2000s. Second-guessing hard-to-believe performances has become a spectator sport. Protesting and proving their innocence has become a second job for many cyclists.

Hamilton's heroics at the Tour de France happened under the watch of Bjarne Riis, the owner of the CSC-sponsored team and a former champion from Denmark. Accusations also trailed Riis for a decade after his memorable and devastating victory in 1996. One observer recalled the occasion: "Through a series of a dozen or so brutally calculated accelerations and decelerations, he ripped the legs off his compet.i.tion, knocking a few loose with each surge, until he ended up alone and very far ahead." This feat Riis accomplished at the age of thirty-three as the leader of a then-unheralded Telekom team, the first time he had commandeered any team in ten years of racing. As owner of the CSC team, he raised many eyebrows by a.s.sembling a front-ranked club within two years. As a rider, Riis always met his doubters with one answer: I have never tested positive. He attributed his success to bee pollen, training methods, and "Thinking to Win," and he incorporated these elements into the CSC program. In 2004, Riis denounced cheaters in a public letter: "It is-to put it mildly-extremely frustrating to experience how some within the sport do not live up to the wish that most of us have for a healthy, sound, professional sport."

Marion Jones, the superstar sprinter and Vogue Vogue cover girl, was certified clean in every one of about 160 samples she provided in her ill.u.s.trious career. She reached dizzying heights at the 2000 Sydney Olympics, earning five medals, three of them gold. Her annual earnings, which included race bonuses and endors.e.m.e.nt deals, exceeded $1 million. But rumors hounded her, not least because of the company she kept, particularly ex-husband C. J. Hunter, ex-boyfriend Tim Montgomery, and coach Trevor Graham. Hunter, a shot put champion, was busted when he tested positive for the steroid nandrolone at a thousand times the normal level. Montgomery, one-time 100-meter world record holder, and Graham, celebrated coach of numerous tainted athletes, were key figures in the BALCO scandal. cover girl, was certified clean in every one of about 160 samples she provided in her ill.u.s.trious career. She reached dizzying heights at the 2000 Sydney Olympics, earning five medals, three of them gold. Her annual earnings, which included race bonuses and endors.e.m.e.nt deals, exceeded $1 million. But rumors hounded her, not least because of the company she kept, particularly ex-husband C. J. Hunter, ex-boyfriend Tim Montgomery, and coach Trevor Graham. Hunter, a shot put champion, was busted when he tested positive for the steroid nandrolone at a thousand times the normal level. Montgomery, one-time 100-meter world record holder, and Graham, celebrated coach of numerous tainted athletes, were key figures in the BALCO scandal.

Notwithstanding, Jones adamantly denied ever using drugs. In her autobiography, she shouted in large, red, capital letters: "I HAVE ALWAYS BEEN UNEQUIVOCAL IN MY OPINION: I AM AGAINST PERFORMANCE-ENHANCING DRUGS. I HAVE NEVER TAKEN THEM AND I NEVER WILL TAKE THEM."

In case we missed the message, she splattered the words all over page 173. When the owner of BALCO, Victor Conte, alleged that she doped "before, during and after the 2000 Olympics," Jones served up a $25 million defamation lawsuit. In an interview with "20/20," the public-affairs program, Conte described how Jones "did the injection with me sitting right there next to her. . . . [She] didn't like to inject in the stomach area. . . . She would do it in her quad." Her lawyers dismissed Conte as "simply not credible." When ex-husband Hunter also implicated her, she called him a vindictive liar. At a time when droves of elite track athletes were being exposed, such as double world sprint champion Kelli White, 200-meter world champion Mich.e.l.le Collins, 1,500-meter world record holder Regina Jacobs, British sprinter and European record holder Dwain Chambers, and Olympic gold medalist twins Calvin and Alvin Harrison, Marion Jones stood tall, never testing positive once. She decried unjust treatment by the media, complaining, "The athletes who have not tested positive have been dragged through the mud."

Then in August 2006, Jones got a scare: an "A" sample she had provided at the U.S. champions.h.i.+ps tested positive for EPO (erythropoietin), a high-tech replica of blood doping that obviates nasty blood transfusions. Her detractors pounced while she continued to deny. Her then coach, Steve Ridd.i.c.k, declared, "I would stake my life on it she did not take EPO." Within a month, the "B" sample was ruled inconclusive, so for the moment, Jones's integrity remained intact.

Now her supporters had their turn to gloat. They attacked the validity of the EPO test as the positive finding of the "A" sample was overturned by the inconclusive "B" sample. Her lawyer lamented, "Marion was wrongfully accused of a doping violation and her reputation was unfairly questioned." For her part, Jones reiterated, "I have always maintained that I have never ever taken performance-enhancing drugs, and I am pleased that a scientific process has now demonstrated that fact."

What was the common link between Marion Jones and Bjarne Riis? They both scaled the summit of their sport, they both pa.s.sed every drug test along the way, they both reaped the riches afforded superstars, and they both swept aside accusations with strident public declarations of honesty. Last but not least, they were disgraceful drug cheats both. Ten years after his stirring victory, long after he had quit compet.i.tive cycling, Riis publicly confessed to extensive doping, including EPO, HGH, and cortisone. Jones eventually confessed in 2007, but only after federal prosecutors had hauled her into court on perjury charges stemming from the many lies she sprang at BALCO investigators. She spent six months in jail. (By contrast, Tyler Hamilton, the clean-cut Olympic champion, never admitted to cheating; on returning to compet.i.tive cycling after the 2005 suspension, he repeatedly failed drug tests, and the eight-year ban he was served in 2009 effectively ended his career.) Supporters of Marion Jones pointed to her ordeal at the 2006 U.S. Champions.h.i.+ps as a real example of a false-positive error. In reality, Jones tested negative at the meet owing to the inconclusive "B" sample, which acted as an automatic protector against false positives. Given subsequent events, one must turn this around and ask if, rather than a false positive, the finding was a false negative! That scenario was more likely, even though Jones had only owned up to unintentional doping between September 2000 and July 2001 (that is, not in 2006). Like Barry Bonds, she claimed to have believed the "clear" was flaxseed oil until the BALCO scandal broke. Her many accusers, however, insisted that she had had full knowledge. Only Jones and her staff would know the whole truth.

London's Daily Telegraph Daily Telegraph, in a.s.sessing Jones's extravagant demise, offered this perspective: "The inconvenient truth that testing negative means nothing is really the key finding of the Marion Jones episode." The reporter was pitch perfect in her evaluation, and one wonders why few others picked up on the story. Statistical a.n.a.lysis shows that in steroid testing, a negative finding has far less value than a positive. As shown in Figure 4-1 Figure 4-1, for each doper caught red-handed (true positive), one should expect about ten others to have escaped scot-free (false negatives). False negatives, not false positives, are the true story of steroid testing.

In particular, pay attention to these two numbers: the proportion of samples declared positive by drug-testing laboratories and the proportion of athletes believed to be using steroids. Of thousands of tests conducted each year, typically 1 percent of samples are declared positive. Therefore, if 10 percent of athletes are drug cheats, then the vast majority-at least 9 percent of them-would have tested negative, and they would have been false negatives. (If the athletes were right that some of the positive findings were errors, then even more dopers would have been missed.)

The issue of false negatives has largely been ignored by the media and is virtually untouched by athletes. Tyler Hamilton, Marion Jones, and others contend that every negative result helped prove their innocence but no positive finding could corroborate their guilt. Channeling Mark McGwire, they might well have chanted, "We are here to talk about the false positive, not the false negative, about steroid testing." Mike Lowell expressed the same belief in numerical terms. Since three full rosters equaled 70 guys, and there were thirty teams in Major League Baseball, he a.s.sumed a total of 700 players, each tested once annually. Of the 700 players, 693 "tested OK," meaning they were clean; the other 7 were false positives, meaning that they too were clean. In other words, all 700 were clean, but 7 hard-luck players tested positive erroneously. With no dopers, of course, every positive result must be a false positive. In Lowell's world, the only acceptable test was the one that gave only negatives.

Figure 4-1 How Steroid Tests Miss Ten Dopers for Each One Caught How Steroid Tests Miss Ten Dopers for Each One Caught [image]

Some dismiss false negatives as victimless errors. Not true. As Michael Johnson, the superlative sprinter with the golden Nike spikes, wrote, "the athletes who finished behind [the winner who cheated] will never experience the glory or recoup the financial benefit they deserved for their hard work." To his credit, Johnson saw the problem of false negatives. A count of the victims of Marion Jones had to start with her relay teammates (who were required to return their medals), and then there were all the silver medalists who should have won gold, all the bronze medalists silver, and all the fourth-place finishers bronze. All waited seven years to learn they had been cheated. (In a sardonic twist, some "victims" turned out to be cheats, too. For example, four of the other seven finalists who raced with Ben Johnson have since been exposed as dopers.) Many athletes get away with cheating. In the anti-doping community, this statement is not controversial. In a review of drug testing for the New York Times New York Times, Professor Charles Yesalis disclosed, "It is virtually impossible to mistakenly identify a substance if a person tests positive for it. [However,] it has been proven that testing cannot catch all substance abusers." Dr. Rasmus Damsgaard, who ran anti-doping programs for professional skiing and cycling teams, estimated that "maybe hundreds, maybe even thousands of EPO positive samples are lying around in WADA-accredited labs," that is, after having pa.s.sed testing. Poring over past doping cases, perhaps David Letterman would feel inspired to make one of his famous Top Ten lists for easy tips to produce a false negative. If so, he might consult the following methods that were actually used, as presented by the athletes with firsthand experience: 10. When the tester is looking away, stir in a little whiskey, and shake it. (Irish swimmer Mich.e.l.le Smith)9. Misdirect the testers to the wrong place, and then stage a motorcycle accident to avoid the out-of-compet.i.tion test. (Sprinter Konstantinos Kederis, also known as "the greatest living Greek")8. Hold a friend's pee inside your body, release quickly when the tester shows up. I get extra credit for being cooperative. (Russian track star Yelena Soboleva and six teammates)7. Believe in human frailty. If a clueless lab technician freezes one of the samples, the lab cannot run a test on it. (Tyler Hamilton)6. It's all about timing! Know how long it takes for the stuff to clear. (American sprinter Kelli White)5. Easy does it for guys. Wear a prosthetic and give them fake pee. (Customers of the Whizzinator and similar products)4. Be ahead of the curve; use only the newest designer stuff. They don't know what it is, so they don't test for it, wink wink. (BALCO athletes)3. It's a natural high. The testosterone is all yours. You're just more manly than the compet.i.tion. (American cyclist Floyd Landis)2. It's so easy to walk right through the front door. Apply for the pa.s.s to cheat; it's called the therapeutic-use exemption. You have asthma, you can dope. (Many athletes) And the number one easy tip to produce a false negative is . . . to sit back and relax, as timid testers will let cheaters ride off into the sunset for fear of wrongly besmirching honest athletes.

Could this really be true? Conventional wisdom says testers and cheaters engage in a high-tech cat-and-mouse game, in which self-righteous testers, eager-perhaps overly eager-to catch the cheaters, tend to cast a wide net, trapping many innocents. However, once we understand the incentives causing testers to look the other way, the game appears to play out differently. The testers are timid because they are swayed by asymmetric costs from the two types of error. A false positive-in fact, any positive, as Tygart observed-will be rigorously litigated by the accused. An overturned positive publicly humiliates the anti-doping authorities and diminishes the credibility of the testing program. By contrast, negative findings can only be proven false if athletes, like Riis, step forward to confess, so most false negatives never see the light of day. Athletes and testers alike can hide behind the anonymity of the false negative. The testers are timid because the false-negative error imparts negligible cost while the false positive can be highly public and highly toxic. Since anti-doping agencies pursue only the strongest cases, no wonder they have won almost all of them. Our criminal-justice system makes the same judgment call: a few murderers escape punishment so that very few innocent people are sent to the chair.

Timid testers are eager to minimize false positives. This objective doesn't conflict with the desire to avoid false-negative mistakes, or does it? Statisticians tell us it definitely does. Testers face an unsavory trade-off: fewer false positives mean more false negatives; fewer false negatives mean more false positives. We next consider how this trade-off manifests itself in steroid tests, using the hematocrit test for catching blood dopers as an ill.u.s.tration.

While no fewer than nine Tour de France champions have been exposed as frauds since 1975, Bjarne Riis had the distinction of carrying the nickname Mr. Sixty Percent, a sarcastic reference to his alleged hematocrit level. The hematocrit level measures the density of red blood cells as a proportion of blood volume. Normal males register a level of about 46 percent. A level of 50 percent or higher is considered abnormal, and 60 percent is insane, because blood becomes too viscous, putting enormous pressure on the heart. The effect is like sucking thick milk shake up a narrow straw.

Before more advanced tests became available, the International Cycling Union (known by its French initials, UCI) used the hematocrit test to identify suspected EPO abusers. EPO is a hormone, naturally secreted by the kidney, which stimulates the growth of red blood cells. By injecting synthetic EPO, endurance athletes boost their red cell counts (and hematocrit levels), raising the oxygen-carrying capacity of their blood. Training at alt.i.tude provides similar benefits but is inconvenient and reportedly less effective. EPO is basically the modern form of blood doping. It is also a potent killer: doctors suspect that the reason why, in recent years, a number of young cyclists-fit men in their prime-have suffered fatal heart attacks in their sleep is EPO abuse. There is a saying that "Cyclists live to ride during the day and then ride to stay alive at night." Allegedly, some EPO users wake up multiple times a night and jump on the exercise machines to get their heart rates up!

UCI used to disqualify everyone with a hematocrit level above 50 percent. This policy sidelined the dopers but also those with a "natural high"; for example, about 20 percent of people living in highlands have natural red cell densities above 50 percent, and they fell victim to the false positive. Antic.i.p.ating this, UCI did not regard the positives as doping violations but instead considered the disqualification a measure to protect the health of cyclists.

Statisticians tell us it was the threshold of 50 percent used to separate natural from unnatural that fixed the unavoidable tradeoff between the two types of errors. If the testers had used 60 percent, they could have reduced false positives, but without a doubt, more dopers would have evaded detection. Similarly, lowering the disqualifying hematocrit level would decrease false negatives at the expense of allowing more false positives.

Nowadays, WADA relies on a more sophisticated, more accurate urine test for EPO. Whether steroid tests measure hematocrit level or another indicator, the principle of all the tests is the same. By setting the threshold of separation, anti-doping authorities explicitly calibrate the tests according to their tolerance for each type of error. Because false positives make for lousy publicity, containing these errors is a top priority. However, this policy inevitably means some drug cheats are let loose, especially since testers can hide behind the false negatives, which are invisible. Due to the sway of asymmetric costs, testers are timid.

Even though anti-doping agencies are inclined to minimize false positives, many athletes echo Mike Lowell's fear. Rafael Palmeiro believed an injection of vitamin B12 caused a false positive. Tyler Hamilton said his vanished twin caused a false positive. Floyd Landis, the American cyclist, claimed a few beers caused his testosterone level to spike. Petr Korda, the Czech tennis player, believed eating veal from calves fed with nandrolone caused his positive. David Martinez, the Spanish discus thrower, citing pigs, even raised one on nandrolone to prove his point. Ben Johnson has long claimed that Carl Lewis spiked his energy drinks, causing the positive. Justin Gatlin, another American sprinter, said his ma.s.seuse rubbed steroid-laced cream on his legs. Zach Lund, the American skeleton rider, knew it was a baldness cure that put finasteride in his body. The list goes on. caused a false positive. Tyler Hamilton said his vanished twin caused a false positive. Floyd Landis, the American cyclist, claimed a few beers caused his testosterone level to spike. Petr Korda, the Czech tennis player, believed eating veal from calves fed with nandrolone caused his positive. David Martinez, the Spanish discus thrower, citing pigs, even raised one on nandrolone to prove his point. Ben Johnson has long claimed that Carl Lewis spiked his energy drinks, causing the positive. Justin Gatlin, another American sprinter, said his ma.s.seuse rubbed steroid-laced cream on his legs. Zach Lund, the American skeleton rider, knew it was a baldness cure that put finasteride in his body. The list goes on.

We may choose to believe all, or some, of these claims. It does not matter, because from a scientific point of view, none of these results actually const.i.tuted a false positive! To see why, we must separate the test of chemistry from the test of integrity. In the test of chemistry, a false positive occurs when the testers report finding an illegal substance that, in fact, does not exist in the sample. In each of the cases just described, the athlete tacitly admitted the presence of the banned drug, so each case was a true positive, not-withstanding the colorful explanations for how steroids entered their bodies. How then do we a.s.sess the claims about nutritional supplements, spiked drinks, and so on? Can we take the athletes' words for it? This is no longer a question of the science; it falls into the realm of lie detection, which is where we now head.

The modern-day polygraph machine is a fear-inducing briefcase filled with medical diagnostic tools, including a chest cord for measuring breathing, an arm cuff for gauging blood pressure and pulse rate, and fingertip electrodes for sensing skin conductivity. Forensic psychologists believe that the act of lying or the fear of getting caught lying raises anxiety, and scientifically speaking, the polygraph detects anxiety, not actual deception. William Marston, a Harvard-trained psychologist who was the first to relate truth telling and blood pressure variations, failed to popularize the concept in the early twentieth century, but he ultimately achieved immortality by creating the comic book heroine Wonder Woman, who not coincidentally wielded a Magic La.s.so that "makes all who are encircled in it tell the truth."

Rather than a "lie detector," the polygraph is simply an instrument of data collection: the data can suggest signs of anxiety, but deception is just one of various causes of elevated blood pressure, hastened breaths, and so on. Therefore, the role of the polygraph examiner is paramount in judging which apprehensive subject is a liar and which is not. A polygraph machine without a competent examiner is like a stock chart without a seasoned a.n.a.lyst: reams of numbers without any significance.

The typical examiner is a retired employee of a law enforcement or intelligence agency that utilizes polygraphs in conducting investigations. He or she has undertaken professional training, such as the thirteen-week course at the Polygraph Inst.i.tute of the Department of Defense, followed by a six-month interns.h.i.+p. This person has developed confidence in his or her ability to interpret fluctuations of various bodily metrics, such as pulse rate and blood pressure. To establish a baseline level of anxiety, the examiner usually engages the subject in extensive pretest conversation, including previewing the test questions. The actual test begins when the examiner feels that the subject is completely at ease. Whenever the examiner senses signs of deception, he or she suspends the test to elicit the subject's state of mind. Throughout the test, the examiner looks out for countermeasures-tactics that the subject may use to derail the examiner, such as biting one's tongue, controlling breathing, contracting certain muscles, deliberately stepping on a hidden tack inside one's shoe, counting backward, and any number of other tactics directors have popularized on film.

In 2005, the simmering steroid scandal in Major League Baseball boiled over as six-time All-Star slugger and self-appointed G.o.dfather of Steroids Jose Canseco ignited the blowtorch known as Juiced Juiced, his finger-pointing expose of the steroid subculture in baseball locker rooms. Canseco dropped bombsh.e.l.ls such as "If I had to guess, I'd say eight out of every ten players had kits in their lockers filled with growth hormones, steroids, supplements." Of greatest interest to the media, though, he named names, starting with fan favorites, including Mark McGwire, Juan Gonzalez, Ivan Rodriguez, Rafael Palmeiro, and Jason Giambi. Regarding McGwire, Canseco recounted, "Mark and I would duck into a stall in the men's room, load up our syringes and inject ourselves. I would often inject Mark." The author of Juiced Juiced was promptly branded as vindictive, despicable, and delusional. was promptly branded as vindictive, despicable, and delusional.

h.e.l.l hath no fury like a G.o.dfather scorned. Three years after Juiced Juiced, Canesco reloaded his blowtorch, delivering a sequel called Vindicated Vindicated. Far from repenting, he reiterated his earlier claims and even advanced some new ones, including an educated guess that Alex Rodriguez, one of the most bankable stars of baseball, was a doper.

To clear any doubts, Canseco released verbatim tran