Errol Morris is one of the world's great digressive talkers, and once he gets started it's almost impossible to stop him. "I don't trust people who don't talk too much," he often says. "You never know what they're thinking."

In a crisp white shirt and khaki shorts, the filmmaker looks like a cross between a tropical explorer and a speed-chess player from Harvard Square. He drums on the floor with the toes of his tennis sneakers and jumps up every few minutes to read aloud from a seven-part essay he's writing for Zoom, his idiosyncratic blog on the New York Times Web site. His chief obsession this afternoon is Laysan, a coral island in the middle of the Pacific that was once a major source of albatross eggs. The eggs are important, Morris tells me, because they are an excellent source of albumen, a key ingredient in a photo-emulsion process used in 19th-century photography.

Every few minutes, his business manager, Ann Petrone, pokes her head in to remind Morris to talk about his new documentary, Standard Operating Procedure. The film, which investigates the strange history of the torture photographs taken at Abu Ghraib prison, is Morris' first since his Academy Award-winning The Fog of War, an intimate and chilling portrait of former secretary of defense Robert McNamara.

In the grip of his current obsession with albatross eggs, Morris ignores her. She reminds him to talk about his new movie once, twice, and then a third time before she finally gives up and goes home.

When the coast is clear, he chuckles, then reaches for a gorgeous coffee table book of nature photographs titled Archipelago and spreads it open on his desk. Morris was initially rejected by every college he applied to, and he was later thrown out of graduate programs at UC Berkeley and Princeton. He remains a failed graduate student at heart, delighting in the pure play of ideas, with the secondary aim of exasperating any responsible adults in the room.

Archipelago, he explains, chronicles an attempt to clean up the island and turn it into a pristine habitat for the endangered albatross. The book tells the story of two birds, Bandy and Shed Bird. Morris is particularly fixated on Shed Bird, who was about to fledge when he toppled over, dead. "And they do a necropsy on Shed Bird, and they remove all this from his stomach," he says, jabbing his finger at the next photograph in the book, which shows a treasury of hundreds of small stones, a walnut, and other objects that might have come from the pants pockets of a young boy. But also among the debris are two cigarette lighters, part of a shotgun shell, and what appears to be a crack pipe.

Any attempts to keep the world pure will always end in failure, Morris explains. In this case, it turns out there are ocean currents that sweep a huge amount of garbage from the Pacific Rim and bring it out to Laysan. "I'm a little disturbed by hopefulness," he says. "It doesn't seem to be the ending that I'm looking for. Shed Bird dies for our sins. I'm a Jew, but I can still get behind that sort of thing."

Morris is sitting in the dark in his second-floor production office in Cambridge, Massachusetts. A stuffed horse head is mounted on the wall behind him. On the windowsill farthest from his desk is a stuffed marabou. Wheezing on the chair beneath the windowsill is Jackpot, a French bulldog who receives a nightly knockout dose of Valium because he snores heavily yet insists on sleeping in the same bed with Morris and his wife, Julia.

The basic facts about Laysan are already familiar to readers of the filmmaker's Times blog, which is less an actual blog and more a place where the filmmaker occasionally posts smart, funny ruminations about the number of cannonballs in Roger Fenton's famous photographs from the Battle of Sebastopol and other seemingly recondite subjects. Many of Morris' blog posts reflect his interest in the ways that photographs presented as pure, objective documentation of reality are often staged and manipulated, like those from Abu Ghraib. Morris answers his more lively and particular commenters at length, weaving together the comments and annotating them in a loopy, digressive, but rigorous way that a friend, writer Ron Rosenbaum, has identified as an entirely new form of essay. Much of the material will be available later this year in book form under the title Which Came First, the Chicken or the Egg?

Morris' philosophically engaged skepticism about our ability to see the world as it really is gives his native sense of irony an added bite. What makes his films so compulsively watchable is the way they combine intensely scrupulous and detailed investigations of other people's investigations with a celebration of the irreducible strangeness of the way human beings perceive reality. The filmmaker's capacity for radical doubt is counterbalanced by his palpable desire to know.

The range and variety of his work is astonishing. There is the funny, life-embracing fan of inventors and crackpots and the funny, dark chronicler of murder and torture. There is the Errol Morris who delights in the world-building imaginations of scientists like Stephen Hawking, star of a film version of A Brief History of Time (1991), or MIT robot designer Rodney Brooks, one of the creative minds at the center of Fast, Cheap & Out of Control (1997); and there's the Errol Morris who seems to perversely seek out unlovable subjects like convicted cop killer Randall Dale Adams in The Thin Blue Line (1988), or electric chair designer and Holocaust denier Fred Leuchter Jr. in Mr. Death (1999). Morris' first — and some would argue greatest — film, Gates of Heaven (1980), combined all of these strands in an unlikely exploration of the pet cemetery business in California that critic Roger Ebert has called one of the 10 greatest films of all time.

Perhaps the most concise statement of the filmmaker's highly analytical yet ultimately absurdist approach to the universe was made in the first three minutes of his second documentary, a willfully perverse chronicle of the residents of Vernon, Florida. (The town's name is the title of the 1982 movie.) The speaker was a local eccentric named Albert Bitterling. "Reality?" Bitterling wonders, as if genuinely considering the idea for the first time. "You mean this is the real world? Huh? Ha ha ha! I never thought of that."

Morris lives in a beautiful 19th-century house near Inman Square, in the middle of a block of wood-frame multifamily homes favored by middle-income couples and graduate students at Harvard and MIT. His wife, Julia, who has helped produce many of his movies, used to work for an art museum. Their home holds a collection of odd objects that she gives Morris as presents and that sometimes cause problems at customs.

In the sitting room near the kitchen is a basket of eggs. There are quail eggs, owl eggs, and a cassowary egg. A relative of the ostrich, the cassowary is a large flightless bird that eats fruit and is famous for getting hit by traffic and for producing large, dense scats. When I ask Julia about the occasion for this gift, she politely demurs. "You don't need an occasion for a cassowary egg," she explains.

Julia and Errol met in Madison, Wisconsin, when Errol was researching his first movie, about serial killer Ed Gein, with whom the director conducted a series of interviews. "I was talking to a mass murderer, but I was thinking of you," he told Julia on one of their first dates. The movie was never made.

In addition to producing many of Morris' best-known documentaries, Julia also named the Interrotron, a somewhat diabolical invention that links two cameras and two teleprompters so that the subject of an interview is never looking directly at Morris but rather at a screen on which Morris can project his own face, or the subject's face, or the Abu Ghraib photographs, or anything else that he pleases, which disconcerts his subjects and gives his interviews their distinctive intensity. It's a technological expression of some of Morris' characteristic habits of mind — combining his interest in the inherently unreliable and self-referential nature of storytelling with his love of gadgets and the gleefully skewed sadism that informed the golden era of human psychology experiments. "I'm not saying you can't get information by torturing people. I don't know," Morris says when I ask him whether he thinks torture can ever be useful or necessary. "I mean, I like to think that I torture people in my own quiet way."

Morris also expresses himself by playing his cellos, which he keeps in the living room. The one he plays for me is the instrument his mother bought him when he was a kid — the same one he plays on the soundtrack of Standard Operating Procedure. At Juilliard, Morris had the same cello teacher as Yo-Yo Ma, whom the filmmaker later met when he directed Ma in one of the original Apple "Switch" commercials. He has directed commercials for high tech clients like Apple, Dell, and Hewlett-Packard as well as Miller beer and Brawny paper towels. "The Miller beer commercials are what I expect to be remembered for," he says. "I've done over 100, and in the end it's always volume that counts."

Morris also shared a music teacher with the composer Philip Glass, who wrote the nerve-racking scores for three of the filmmaker's best-known movies: The Thin Blue Line, A Brief History of Time, and The Fog of War. "I was very interested that someone who had once been at Juilliard would end up making movies," Glass says when I ask why he chose to collaborate with Morris. "I had never worked with a filmmaker who could take a score and play it on the piano. And let me tell you, it is not an advantage for a composer." Morris kept sending Glass back to write more cues, and still more cues, until the composer started to get cranky.

The Thin Blue Line, a reinvestigation of a closed murder case that saved an innocent man from life in prison, was the first time Morris used the mix of in-depth interviewing, slow-motion reenactments, and fanciful special effects that has changed the way both fiction and nonfiction films are shot. The carefully balanced artifice of his approach confused and annoyed documentary purists so much that Morris had to wait nearly two decades, and make five more films, before receiving his first Academy Award nomination.

The use of the invented footage, which inspired a generation of forensically inclined television shows, was the filmmaker's way of illustrating the unreliable and often deliberately deceptive nature of his subjects' recollections — in this case, their recollections of the events that landed Randall Adams in prison. Taking as literally as possible the accounts of eyewitnesses to a Dallas police officer's murder, Morris meticulously re-created their stories using actors on stage sets, then slowed the action to a dreamlike, spooky crawl.

The odd blend of deadpan humor and investigative seriousness that runs through these shots, which ultimately proved Adams' innocence and got him released from prison, is also a part of a larger meditation on the untrustworthy nature of memory and language. Morris' use of re-creations suggests that if you take anything anyone says literally, act it out, and then look at it very closely, it's hard not to laugh. At the same time, by repeating different versions of the same event, it is also possible to arrive at a more or less accurate picture of what really happened.

Perhaps the most memorable of Morris' slo-mo mini-essays on the flawed nature of human perception is the milk-shake shot — a half-empty chocolate shake from Burger King splattering on a dark, lonely road moments after the shooting of police officer Robert Wood. The scene helped Adams prove his innocence by persuasively suggesting through repeated reenactments that — contrary to her sworn testimony — Wood's partner, Teresa Turko, never got out of the car and never could have identified Adams as the shooter. Instead, she tossed her drink out the window and drove from the scene, leaving her partner bleeding to death on the highway.

Morris' vivid re-creations of the courtroom testimony draw the viewer's attention to lies and evasions in the spoken accounts of the crime while, at the same time, insisting that behind the swirl of conflicting narratives there is a solid, verifiable reality that can be discovered by a patient investigator.

According to Ron Rosenbaum, Morris has long proposed doing a Rashomon-like movie about Kurosawa's Rashomon to illustrate his annoyance with the slovenly belief that the original somehow proves the postmodern idea that there is no concrete reality beyond the shifting and partial perspectives of the participants. "The last time I had dinner with him, he was talking about hiring a crime-scene investigator to reinvestigate the crime in Rashomon," Rosenbaum says. "Errol believes that there is a single truth, but complexity derives from the fact that the nature of truth is often elusive and may never be found. But that doesn't mean you should stop searching for it."

When he filmed The Thin Blue Line in 1987, Morris says, the fastest available camera ran at 120 frames per second, roughly five times normal speed. The Phantom digital cameras he used for the slow-motion scenes in Standard Operating Procedure shot 1,000 fps, which allowed him to ramp the shots, drawing a curve on a computer and varying the speed during a single take. "You can go from 300 to 1,000 and you don't even see the difference. It's all so slow," Morris says.

The slow motion in Standard Operating Procedure is more richly detailed than the homemade effects in The Thin Blue Line, showing the viewer the slavering guard dogs as they appeared to the Iraqi prisoners and illustrating in loving, color-saturated impasto the random humiliations to which the prisoners were subjected. The Abu Ghraib reenactments, the director makes a point of telling me, were all filmed on the same sound stage in Los Angeles that was used to shoot I Love Lucy.

Morris credits Susan Sontag for being one of the few to notice that the Abu Ghraib photographs were not found artifacts of torture, as most Americans and the rest of the world believed, but scenes deliberately created for the camera by the guards themselves. Without the presence of a camera, the most famous of the digitized moments from Abu Ghraib — the prisoner called Gilligan standing on a box, Lynndie England leading a naked detainee on a leash — would most likely never have taken place.

Like all of Morris' best films, Standard Operating Procedure is a philosophical comedy about an unexpected subject — a premeditated assault on the viewer's expectations of what's true. The famous torture photographs from Abu Ghraib are not photographs of inmates being tortured. They were deliberately created by the guards, whose reasons for posing the pictures were complex, layered, weird, and in some unexpected cases even admirable. Morris reinterprets the infamous pictures as a kind of highly sexualized samizdat parody of the bizarre and even more terrifying reality inside and outside the prison's walls. He also suggests that some of the photos were taken on purpose to expose abuses but wound up landing the photographers in jail. It is one of the outstanding ironies of Morris' story that the photographs, which were seen by the world as documentary evidence of torture, were used as a way to distract attention from the brutal crimes that took place off-camera. While the low-ranking soldiers caught in the staged pictures went to prison, the teams of professional Army and CIA interrogators who actually tortured and murdered prisoners inside Abu Ghraib were never identified or punished.

Reenactments and even staged photos, Morris says, can best be understood as a kind of philosopher's tool that allows us to road test our own addled thinking about the world. "It's not as though Errol Morris' takes this veridical picture of reality that would exist otherwise and sullies it with his damn slo-mo and his reenacted material and his use of actors and his construction of sets," he says. "We manage to survive in the world because we don't think about it carefully. Or scrutinize it. Otherwise, it would just be too horrific."

The 230 photographs Morris examines so intensely in Standard Operating Procedure, which formed nearly the entire case against the soldiers who went to prison for the abuses, are merely a fraction of the many thousands of Abu Ghraib photographs. "Even I have thousands of Abu Ghraib photographs," Morris says enthusiastically, spreading his hands apart to indicate their extent. His point is that we live in a solid world, but our perception of it is flawed and partial:

Q: But don't you think that we manage to keep up a coherent narrative of who we are and where we've been?

A: No! We remember things selectively. We experience things selectively. We think about things selectively. We live in a kind of incomplete, patchwork-quilt universe. A bric-a-brac. Assembled in some higgledy-piggledy way.

He pauses for a moment. A new link in the digressive chain is forming. Morris starts talking about a new idea with the excess energy and over-enthusiasm of the philosophy grad student who will sooner or later get an ashtray thrown at his head by a grumpy professor. (It happened to him at Princeton, he says.)

"In 1974, this guy Brandon Carter came up with what he called the anthropic principle," Morris continues. "There was the thought that the reason the world looks the way it does is because we're in it, but that doesn't mean that in the areas we're not in, the world actually is the way it is here."

We think about Abu Ghraib the way that we do because of a particular grouping of photographs. But the photographs that we have are in turn only a small sample of the larger universe of photographs taken in Abu Ghraib. "You know, someone hands you this pile of photographs, and they say, This is the world,'" Morris explains. "And you say, Oh, thank you very much. I see. This is the world.' But you look no further! Abu Ghraib exists in part because of that phenomenon."

Errol Morris is watching a videoclip on his iPhone in the lobby of the Mercer Hotel in Manhattan. The walls are painted white, and there are soft couches to sit on, soft light from standing lamps near the couches, and soft conversation. Only guests of the hotel are allowed to order food in the lobby, which is stuffed with creative types on expense accounts talking about Fashion Week. Morris is in New York to direct a commercial. We are talking about the dessert menu and the Saddam-hanging video, which he references at the end of Standard Operating Procedure in a shot of a noose and a trapdoor opening in a scaffold.

While Morris has avoided looking at online videos of jihadist beheadings, like the murders of Nick Berg and Daniel Pearl, he watched the Saddam-hanging video immediately and repeatedly. As an advertisement for the US war effort, Morris says, the Saddam-hanging video was not particularly well scripted. Saddam looks like he is going to a premiere of La Boheme. "He has his beautiful scarf, his tailored overcoat, and he's standing there really quite elegantly. I would even go so far as to say regally," Morris points out. "He's clearly angry, but completely composed."

Meanwhile, on the platform with him are a lot of crazy-looking people, many wearing ski masks. "You know, the whole term ski mask is itself a joke," he says. "If you saw somebody — you're skiing in Vail, and you saw someone on the trail coming from behind you wearing a ski mask, you would go batshit. You'd try to call homeland security, the police. God help us! You know, it's someone skiing with a ski mask! It's a fucking terrorist!" He's high on the energy of his riff. "No self-respecting skier would ever be caught dead with a ski mask. It would be risky. You'd get shot," he says, bringing his voice back down to match the quiet conversations of the white-walled room. "And what you see, for all intents and purposes, is a bunch of terrorists who are disorganized, confused, angry, screaming, vindictive. And you see this guy, you know, ready to go to the opera in the middle of it."

Morris notes how amazing it is to be able to watch, on a phone, the video of the hanging of Saddam Hussein, filmed by someone on their phone, and to discover there an essential truth about a war whose only real purpose was to kill Saddam. "I know I like being able to access all kinds of strange information easily," he says when I ask how the Web has changed him. He seems to have the kind of mind, after all, that might drown itself in too much data. In fact, he says, the happy randomness of online research has helped him to start writing again for the first time since he left graduate school. What he's doing with Zoom would be much more difficult, if not impossible, in the bad old days before Google.

Still, Morris reminds me, iPhones and Google are useful but not sufficient tools for understanding reality. He recalls coming across what appeared to be an underlined word in a document shown in a book by Jean-Claude Pressac, a onetime Holocaust denier who published an extensive work on documents related to the design and construction of the gas chambers at Auschwitz. Morris was researching his film Mr. Death. "You see the printed document in the book, and you can see that it's been underlined, and the underline looks like someone really underlined it," he remembers. "And the word is vergasungskeller. Gas chamber. OK? Now, I started to get curious. Was this underlined on the original? Or on a copy? When was it underlined? And why was it underlined? The Nazis were very careful about not using explicit references to the fact that they were gassing people." His hunch was that the emphatic underlining was intended to draw the original author's attention to a mistake in the document.

When he went to the building archive at Auschwitz, he got to see the original — which was in fact not the original but the original carbon copy that had been retained in the archive, the document itself having been sent to Berlin, where it was presumably destroyed. The underlining was in red pencil, which had bled through the carbon. At the top of the page, in the same red pencil, which was not visible in the black-and- white version reproduced in Pressac's book, was a dated instruction to the SS captain who authored the document, indicating that using the word vergasungskeller was something he should never have done and he should be careful never to do again. By looking in the archive, rather than online, Morris found a rare, handwritten reminder that the Nazis consciously worked to eliminate from official documents all mention of what they were doing — this absence being one of the chief arguments used by the Holocaust deniers that Morris portrayed in Mr. Death.

Reality is stubborn and troublesome and often un-Googleable. And Errol Morris wouldn't have it any other way. The principle of uncertainty that guides the filmmaker's investigations is also the source of the rich and irrepressible humor that runs through his work. Through our embrace of the life-giving habits of wonder and doubt, we may be lucky enough to discover that the joke is always and inevitably on us.

David Samuels (www.threeminutehappiness.com) is the author of two new books:The Runner and Only Love Can Break Your Heart.

Jameel Ahed says he didn't really read the email. He was preoccupied with trying to solve a few electrical problems on the robots he hoped his company would sell to the US Army for as much as $300 million, one of the largest robot orders in history. So he didn't pay much attention to the fact that iRobot, his former employer and chief competitor for the Army contract, was suing him for stealing their designs. And he didn't read the attachment ordering him to preserve any evidence related to the case.

Nevertheless, a few hours later that evening, Ahed began cleaning out his suburban Chicago office. He gathered up circuit boards and electronic components and threw them into a box marked "iRobot" — maker of the cute little Roomba vacuum cleaner and sophisticated military robots. He stuffed the box into a green duffel bag, carried it outside, and put it in the trunk of a white Saturn registered to the parents of Kimberly Hill, his girlfriend and the chief operating officer of his company, Robotic FX.

Around midnight that night, August 17, 2007, Ahed and Hill drove to her apartment on Chicago's North Side. The next morning, they went out to breakfast. When they came out of the restaurant, the Saturn's trunk was ajar. Ahed made sure the duffel bag was still inside. Then they drove to a Staples, where he bought a shredder and a disc-erasing program. On the way back, Ahed stepped out of the Saturn, opened the trunk, and emptied the contents of the bag into a curbside dumpster a block from Hill's apartment.

Ahed didn't know that a private investigator was parked nearby. The PI, hired that morning, photographed the dumpster drop and recovered all of the gear. Then he reported back to his client: iRobot.

Ahed was 20 years old when he joined iRobot in May 1999, a biomedical engineering student at the University of Illinois on a summer internship. In those days, the company was just 80 or so geeks in the Boston exurbs designing toys for Hasbro and doing research for Darpa. Ahed stood out. He was hardworking, flirtatious, and outgoing. He dressed well (for an engineer, anyway) and drove a DeLorean. The only criticism anyone could level was that he had the arrogance of youth — he seemed to think he was the smartest kid in the room, that he could solve any problem. When the summer ended, iRobot gave him some robot parts to play with at school; when he graduated, they gave him a job.

The company was a classic startup — almost a cliché. Everyone hung out together. Everyone worked constantly. "We'd sleep in the office, by the front door, so the first people arriving in the morning would hit us on the head and wake us up," says Tom Frost, Ahed's former supervisor. They watched movies at Ahed's apartment and test-drove robots at Frost's parents' house on the beach in Rhode Island.

The range of projects at iRobot was staggering: an automated industrial floor cleaner for Johnson Wax (now SC Johnson), makers of Windex and Pledge; a baby doll that laughed when its feet were tickled; and Urbie, a small, rugged military robot with flippers that enabled it to climb stairs and right itself when turned upside down.

But Ahed was dissatisfied. He missed his hometown of Chicago, and he wanted to be his own boss — or at least to rise in the company a whole lot faster. Building radio antennas wasn't enough. In December 2001, he bought the domain name roboticfx.com, planning to launch his own startup.

Then Frost chose Ahed to work on the electrical team for the PackBot — a descendent of the Urbie for which the company had high hopes. Ahed made two requests. He asked for a raise and for the machine's full electrical schematics. That set off alarm bells. Ahed already had access to the PackBot's mechanical designs, which were kept on a shared server. If he got hold of the complete electricals, he'd be able to make a PackBot on his own.

Frost said no, and Ahed exploded. "I was left without the tools that I needed to do my tasks," he said later in a videotaped deposition. "That frustration took me over the top ... I told Tom if I didn't have — if I didn't get help, that I didn't want to stay." Ahed gave his two weeks' notice.

Before he left, a company staffer demanded that he sign a final confidentiality agreement. Ahed complained but signed. The next day, an email was sent at 10:18 pm from his still-active iRobot account to his new Robotic FX address detailing how the PackBot's batteries were made. Shortly thereafter, Ahed packed up and returned to Chicago.

In July 2002, a month after Ahed quit, Frost took five PackBot prototypes to Afghanistan. Near the mountainous Pakistani border, the Army's 101st Airborne sent the squat, rectangular machines to map suspected terrorist compounds and find weapons caches. Normally, flesh-and-blood scouts would risk their lives to do these jobs. Now the 101st was risking only hardware. The soldiers liked the robots so much, they wouldn't let Frost take them back home with him. Stateside, Frost and his Army liaison became popular figures on the Pentagon briefing circuit, even pitching the robots to defense secretary Donald Rumsfeld.

By the summer of 2003, iRobot had orders for a few dozen PackBots at around $60,000 each. And that looked to be just a warm-up. Advanced robots are the heart of the Army's $200 billion Future Combat Systems modernization project. By 2030, 15 brigades are to be reinforced by more than 1,200 little robot buddies.

The firm's civilian business was taking off, too. In 2002, engineers applied what they'd learned building the SC Johnson floor-cleaner to a disc-shaped bot they called the Roomba. The quirky $200 robovac became the must-have gadget that holiday season, eventually selling more than 2.5 million units. On November 9, 2005, iRobot went public and emerged with a market capitalization of more than $620 million. The PackBot rang the Nasdaq's opening bell.

Ahed, meanwhile, was working out of a small office and basement workshop attached to his father's dental practice. He was intent on creating a lighter, cheaper, easier-to-operate version of his former employer's machines. And he was smart about it. He used inexpensive commercial processors instead of robot-specific chips. He designed custom fixtures so he could injection-mold thermoplastic rubber parts — much lighter than the PackBot's aluminum components. By 2004, Ahed had a bare-bones prototype he called the Negotiator. It weighed just 20 pounds and cost less than $30,000 — half what iRobot was charging for a comparable early version of the PackBot.

The Negotiator's driving system also beat the PackBot's, which at the time used two hockey-puck-like discs mounted on a console — a nonintuitive and complicated setup. Early models of the Negotiator worked with a simple joystick. Later, Ahed incorporated an ingenious puppet controller, similar to what doctors use to teleoperate surgical machines. Pinch your fingers and the Negotiator's claw pinches, too. (The latest PackBot works with joysticks or a PlayStation-style D-pad.)

Ahed debuted the Negotiator at a trade show in April 2004. He made his first sale three months later, and a few months after that he sold six Negotiators to the Illinois State Police for about $20,000 each. Then, in September 2005, an iRobot sales rep saw a Negotiator at a police conference in Miami. Like the PackBot, it was about 2 feet long and a foot wide, and it, too, had tanklike treads with ovoid flippers. He couldn't get his cell phone out fast enough. "You've got to look at this, quick," he told Frost. "This thing looks a hell of a lot like ours."

Eventually, iRobot managed to buy a Negotiator through a third party. Frost and his team were struck by how closely the wheels and treads resembled the PackBot's. "It was upsetting," Frost says. "He worked for me. We'd been to each other's homes. Gotten our asses kicked in softball together." In February 2007, iRobot's lawyers sent a cease-and-desist letter to Ahed, demanding that he stop making and marketing the Negotiator.

Five months later, in July, the Army announced its biggest ground robot contract ever. The so-called xBot deal would be worth up to $300 million and cover as many as 3,000 units — a surge. Robots were already a familiar sight on battlefields. In 2004, the Defense Department had 162 robots in Iraq and Afghanistan; in 2007 there were more than 5,000, from a variety of vendors. Most of the bots in the field were tiny — really just souped-up, radio-controlled toy trucks with cameras attached. But the Army wanted to deploy robots tough enough for combat conditions. And with the number of IEDs spiking, they wanted the machines yesterday. The first 101 would be due six months after the contract was awarded.

Overseeing the contract was a Marine colonel named Ed Ward at the Robotic Systems Joint Project Office in Redstone Arsenal, Alabama. A former helicopter pilot and business consultant, Ward was famous for establishing a cheap, reliable repair shop for bomb-handling robots in Iraq in just a few weeks. He wanted to bring that same kind of economy and speed to the robots themselves, and he thought $120,000 a pop for the upgraded, fully loaded PackBot was a lot to pay. "Look, if I see something strange down the road and I want to take a look at it, I don't need all the bells and whistles," Ward says. "I just need mobility and a good camera."

Rather than go through a months-long bidding process, the JPO announced a drive-off, a sort of robot ropes course, to be held in August. And to bring prices down, the Army would then hold a reverse auction, encouraging sellers to underbid each other.

For iRobot, this was a crucial opportunity, a chance to get a bigger, more lucrative foothold in the military market. At first glance, the company seemed to have every advantage over the competition. The xBot specs called for a robot weighing 50 pounds or less that could hurdle obstacles and scope out potential bombs. The PackBot was just a few pounds over and could already do all that.

But when iRobot's people started digging into the details of the contract, they were spooked. "Our initial reaction was, whaaaat?" Frost says. The request for proposals didn't mention reliability, even though the machines were destined for Iraq and Afghanistan. The reverse auction didn't make sense to them, either. "It's not unusual for a commodity like road salt, where price is the only discriminant," says Joe Dyer, a retired vice admiral who now oversees iRobot's government business. "But a tightly integrated technology? If there's a precedent, I'm unaware of it."

The company plowed ahead, eventually entering two machines — a modified PackBot and a model tailored for the Future Combat Systems program. There were so many modifications to be made, like switching to off-the-shelf encoders, gears, and cameras, that engineers began working around-the-clock shifts. In August, Frost and his crew flew to Huntsville, Alabama, home of the JPO and the site of the drive-off. They tipped the beds over in one of their hotel rooms and converted the place into a makeshift repair shop.

Race day began with a weigh-in and a quick inspection. Soldiers confiscated the engineers' cell phones and cameras to make sure that they couldn't contact (or even see) any of the other competitors. They led the PackBot crew into the 104-degree Alabama summer heat and onto the test track, an old firing range ringed with hills.

Each robot was allowed 90 minutes to complete the course. The PackBot followed a zigzagging series of green flags against grassy fields — a challenge for electronic eyes — and identified mock IEDs in trash piles along the way. It barreled across sandpits, gravel, and rock. Sitting in 6 inches of water, the robot lifted a 5-pound simulated pipe bomb and spun it around, showing how well its arm could move. It then drove up a ramp and into the back of a tractor-trailer, where it picked out coat hangers, bowling pins, and rocket launchers amid the shadows. Finally, the PackBot had to show how fast it could sprint — and how slow it could creep along. "That was the worst part," says PackBot driver John Souliere. "As the clock is ticking down, we're watching it go as slowly as it can." But the bot made it with 12 minutes to spare. Frost and his team returned to Boston exhausted and triumphant.

And the Negotiator? Ahed and his representatives rebuffed requests for comment (about this and every other aspect of this story). In fact, no one will describe the Negotiator's performance on the record: Army officials won't even reveal how many teams competed in the xBot trial. But emails recovered from one of Ahed's hard drives say that the Negotiator lost power before it finished the course. Ahed apparently swapped out the battery and asked for a retest. He got it, and the Negotiator completed the course.

A few days later, an official told Frost's crew that they'd have to run the course again. One of the tests relied too much on knowledge of military gear, he said. That was odd — how much wartime experience do you need to distinguish a rocket launcher from a bowling pin? Frost piled his team back on the plane, but bad weather grounded them in DC. They rented a car and sped down to Alabama. The shape and sequence of the test course had been slightly rearranged. This time, the PackBot finished with 40 minutes left on the clock.

Meanwhile, as Frost's team was packing to return to Alabama for this second audition, iRobot's lawyers emailed Ahed to notify him of their lawsuits filed in Alabama and Massachusetts for patent and trade secret theft. The following morning, a Saturday, the private investigators snapped the pictures of Ahed's dumpster run.

Ahed spent the rest of that weekend scrubbing files from his hard drives and destroying the CD-ROMs he had collected from the office. Many of them contained information about early Negotiator designs. He fed so many CDs into the shredder, it finally gave out. He bought a second one.

On Monday, August 20, a district court judge in Alabama issued a temporary restraining order against Robotic FX, demanding that it preserve "all evidence, information, data, and documents." The next day, iRobot's lawyers showed up at Robotic FX's office, US marshals in tow.

They searched the office, then Ahed's parents' house, and finally headed to Hill's apartment in Chicago. Ahed and Hill got there first, barely, and went in through the back door. Hill let the marshals in the front while Ahed ran to the bedroom. His laptop was there, its screen displaying a message that the hard drive had been wiped clean. He unplugged it, put it in a case, and slid the case underneath the bed. Then he went to the living room, where the marshals were asking Hill where the computers were. She didn't mention the laptop. They found it anyway.

And with that, iRobot figured the competition to supply the Army's unmanned surge was over. Sure, the xBot contract was supposed to go to the lowest bidder, in theory. But the iRobot team was certain the military wouldn't give the contract to a "gentleman of such questionable deportment," as Dyer told me. For that matter, how was Ahed going to make 3,000 robots anyway? "We were very, very suspect of our competitor's ability to produce both quantity and quality," Dyer says. "So we decided, we're not going to underbid a six-person company working out of Dad's dental shop." iRobot made what it considered a reasonable offer: $286 million. Ahed entered what he must have thought was an opening counterbid: $285 million. iRobot didn't bother to respond.

On September 14, 2007, the Army awarded the five-year xBot contract to Ahed for $279.9 million (Ahed knocked off the extra $5.1 million to sweeten the deal).

iRobot went into battle mode. The company filed official protests with the Army and petitioned the Massachusetts judge for a preliminary injunction to put the brakes on the xBot contract.

The Army fought back. Ward argued that there were too many bombs going off in Iraq to stop things now. "The number one cause of Soldier and Marine casualties are IEDs, and this system allows the Soldier and Marine to complete their mission while keeping them out of harm's way," he wrote to the Massachusetts court. "Without the xBot, the combat units will suffer additional casualties."

On September 24, Ahed took the stand in US district court in Boston. He admitted to sweeping his office for old discs and circuit boards — because he no longer wanted "to have any iRobot memorabilia" lying around. Oh, and also because of his concern that he was being followed: "I was afraid someone was going to come and take them."

Military brass didn't buy it. On October 23, Joanne Byrd, a contracting officer with the Army, concluded that there were "questions regarding the award" and that a "reassessment" of Robotic FX was necessary. The bid was officially frozen.

The next week, judge Nancy Gertner granted iRobot a preliminary injunction and ordered the suit to go to trial. She mostly ignored the question of whether the Negotiator's tracks were rip-offs of the PackBot's, focusing instead on Ahed's "highly suspect" behavior. In mid-December, the Army canceled the Robotic FX contract and handed the deal to iRobot for $286 million.

And that, it seemed, was that. The robot thief got his due. Soldiers would have their machines. iRobot got the contract. "This had more twists and turns than a John Grisham novel," Dyer says. "But now it's done."

Except the story has one more potboiler twist. Yes, Jameel Ahed almost stopped his former bosses from getting an important military contract, and he probably used some of their designs to do it. He certainly tried to destroy evidence to that effect. But why he thought he could get away with it has an explanation that was barely mentioned in court.

Ward had been a Negotiator fan since June 2006, smitten by its easy-to-use controller, its sharp-eyed camera, and, of course, its minuscule price tag. "It changed the whole perspective on what robots could cost. This was dramatically different, an order of magnitude different, from what we had seen," one military official says. "It had the potential of changing the whole landscape of robotics." The JPO ordered several of the machines just to play around with, though it wouldn't agree to a later request from Robotic FX that the government protect the company from future patent lawsuits.

In meetings with robot makers, military officials liked to bring up the Negotiator — and its price. "They absolutely used it as a club against us," a former iRobot employee says. When the xBot competition came along, that club turned into a sledgehammer. The xBot specs essentially asked for a smaller, lighter, stripped-down PackBot — in other words, a Negotiator. The reverse auction put a premium on low cost. It was as if the specs had been written for Ahed.

Yet to iRobot, Ahed didn't seem like much of a threat. The company had swatted away knockoffs before. And how could Ahed hope to fill the Pentagon's robot needs from his father's basement? Some iRobot employees hung pictures of the Negotiator around the office as objects of ridicule.

What iRobot didn't know was that Ahed had a partner. In court, Ahed admitted that a large defense contractor had "helped us with some components" and a bit of cash. In truth, the relationship between the two companies — not to mention their relationship with the military — went much deeper.

In return for an interview with an executive of that defense contractor, I agreed not to name him or the company. I can say that the firm is big enough to make robots in quantity — something iRobot had only barely figured out how to do after years of supplying the government. The executive wanted the infantry robotics contract badly; delivering robots to soldiers was an opportunity for decades of steady cash. "We needed that product," he tells me. The company didn't have a robot that could run the Alabama course.

Meanwhile, Ward, the Army robot expert, continued to talk up the Negotiator. "He kept asking us whether we had taken a look at this robot, that it was a damn good piece of kit," the executive says. Ward denies that he asked any executive to team up with Ahed. That would have been against the rules, and anyway he didn't have to. "Eventually, you figure out for yourself where things lie," the executive says. In mid-June, he flew out to meet with Ahed. A month later, when the xBot contract was formally announced, he went back for a personal demo in Ahed's basement workshop.

At first, the executive was horrified at this "dungeon." Then he gave the robot a try. Ahed laid a quarter on a table. The executive put his hand into the puppet controller, pinched his fingers, picked up the coin with the Negotiator's claw, and dropped the quarter into his shirt pocket. No other robot on the market was this easy to handle or this agile. "A first-time user — picking up a goddamn quarter!" the executive shouts. "That's when I knew: I wanted that robot."

He proposed a deal. Ahed would move out of the basement and into a facility befitting a big-time military supplier. If he won the xBot competition, the executive's firm would buy Robotic FX and use its own, larger assembly lines to crank out robots by the hundreds. And iRobot wouldn't be able to underbid, because the Negotiator was so cheap to make. It would be a triple win: The government got bomb spotters cheap enough to outfit the whole infantry, Ahed got rich, and the defense company got a giant new market. Ahed agreed. How could he not?

Their arrangement might seem a little underhanded. But in the world of defense contracting, it was business as usual — and depending on your perspective, kind of admirable. Pentagon officials often write requirements for military gear with a particular company in mind. (iRobot itself likely benefited from such tuned specs in earlier contracts.) If anything, the rigid performance thresholds and reverse-auction format of the xBot deal made it fairer than many Pentagon awards. "This is a case of trying to get the best product for the best price," says Scott Amey, general counsel for the Project on Government Oversight, a nonprofit, nongovernmental organization that regularly investigates suspicious Pentagon deals. "Tell me what's wrong here."

Of course, there is the not insignificant matter of the Negotiator's treads — the ones that look so much like the PackBot's. The executive and Ahed knew iRobot would come after them for patent infringement. But they had a plan. One line of their defense went like this: Because iRobot had developed the PackBot in part with money from the US military, its design could arguably be considered public property. At the very least, debating such a claim could tie up a lawsuit in court for years. "We were prepared to spend hundreds of thousands to defend ourselves," the executive says. And while the lawyers squabbled, the Army was ready to push ahead. "This [is] a business issue between two companies and the government [is] not intending to become involved," Ward emailed the two robot makers.

It almost worked. "Jameel thought it was over. He thought he could get this $300 million contract and there was nothing anyone else could do about it. He thought his dreams were about to come true," the executive says of Ahed. "He was very naive about what the other players in the market would do."

When Ahed got the email from iRobot's attorneys announcing the lawsuit, it couldn't have been a surprise — he had received the cease-and-desist letter months earlier. He and his partner discussed a media strategy in which Ahed would be portrayed as "the aggrieved party ... in a David vs. Goliath situation," according to one recovered email.

Yet when marshals showed up at Ahed's door, he called the executive in a frenzy. "What should I do?" he shrieked. The man answered: Cooperate, no matter what. Tell them absolutely everything. Of course, Ahed responded. Of course. But he had already destroyed evidence, giving iRobot the ammunition it needed to undermine Ahed's credibility and get the deal scuttled.

iRobot has delivered the first few dozen machines to Alabama, and the company expects to fulfill its initial obligation of 101 robots this month. In all, under the terms of the contract, iRobot could provide up to 3,000 infantry robots over the next five years. Ward, the xBot manager, has retired from the military. The executive who hooked up with Ahed is still looking for a robot to crack the infantry market.

In December, iRobot settled with Ahed. Under the terms of the agreement, he is barred from "competitive activities" in the robotics industry for five years and his venture was officially dissolved. Its Web site now reads: "Robotic FX is no longer in business. Certain residual assets are now owned by iRobot Corporation, the leader in small unmanned ground robots." In other words, iRobot can now use some of Ahed's Negotiator designs. And the thousands of robots that are supposed to accompany America's future ground forces will almost certainly be built by iRobot.

Contributing editor Noah Shachtman wrote about network-centric warfare in issue 15.12.