• Twitter Social Icon
  • Instagram Social Icon

"The good life is one inspired by love and guided by knowledge."

-Bertrand Russell

Chapter 5.

“The Forest Fire Problem”


Foresters and ecologists monitoring Yellowstone National Park started to realize that there was something different about the 1988 fire season on July 2. A fire had ignited in Storm Creek in Gallatin National Forest several weeks earlier. Standard fire management practice is to let naturally-caused fires take their own course. This is the procedure park managers had initially followed with the fire in Gallatin. But on this particular Thursday, United States Forest Service supervisor John Burns told Yellowstone Superintendent Robert Barbee that they didn’t want the Gallitin fire spreading into the Targhee National Forest. The park managers began an effort to suppress the fire, an action more aggressive than their usual policy. 

In a typical fire season, Yellowstone will see a few dozen wildfires. Most of these are minor, covering less than an acre each. Before the summers months of 1988 there were no indications that it would be anything other than a typical fire season. So when the first few fires broke out in late June, park managers let them burn. Usually a forest fire loses its inclination to spread only shortly after its ignited, whether due to the discouragement of rain or a lack of fuel. The park managers expected to the rains to return in July, as they usually do. Eleven of the eighteen initial fires burned themselves out as hoped. But the July rains never came. 

By July 12, about 6,000 acres of Yellowstone had been burned by the wild fires. It had been almost two months without rain. Superintendent Barbee announced on July 15 that they would begin suppressing all new fires, regardless of the cause. Barbee and the other park managed put together a map summarizing the damage done by the fires. The map of Yellowstone—which is shaped like a rectangle centered in northwest Wyoming, reaching into Montana in the north and Idaho in the west—showed that the fire dotted the landscape across the park in small flecks. At this point the fires were widespread but still manageable. 


By July 21, fire acreage doubled in just over a week. It covered 17,000 acres now. The park managers issued a statement saying that they were dedicating the full force of their resources to fighting the fires. It was late July and still no rain. Conditions were bad, worse than even what the professional fire fighters were used to. One technique that fire fighters often use is called a burnout fire, where fire fighters light a small, controlled fire in the path of an existing fire in an attempt to preemptively exhaust its fuel supply. But standard approaches like lighting burnouts weren’t working. Because of high winds, the fires were spotting—embers would leap into the air and travel up to a mile and half ahead of the primary flame, jumping rivers, canyons, fire barriers, and overcoming obstacles that would normally prevent the fire from progressing.

In late July, a group of fire experts convened for two days in West Yellowstone to try and get a handle on things—how bad was the damage, and how much worse was it going to get? It was the first time such an extensive forecast had been attempted for Yellowstone. They looked at historical weather patterns, how the fire had behaved so far, the ages of the forests that were nearby the fire, and the different vegetation zones, each with different burning properties, that were in the fire’s proximity. On August 2, the committee made its decisions about how best to allocate crews and equipment. According to their best estimates, 150,000 acres had already been burned. They stated that “as much as another 100,00 acres” more could be added, but that the worst was over “because of shortages of fuel.”

Two hundred and fifty individuals fires started in Yellowstone between June and August, but only a handful of them did the majority of the damage. There was the Mink fire, ignited by lightning on July 11 in the Bridger-Teton National Forest. There was the Hellroaring fire, ignited August 15 in the Gallatin National Forest when embers from an untended campfire crept onto nearby trees. More acutely devastating was the Huck fire, ignited August 20 when a tree fell on a power line outside of Flagg Ranch and spreading at the urging of high winds. Ashes fell on Billings, Montana, 60 miles north east of the flames. Airplanes and helicopters were grounded. There was nothing that the fire fighters could do. The winds pushed the fire across 150,000 acres in a single day. It was more than had burned in the one hundred years of Yellowstone’s recorded fires combined, a day that fire ecologists now refer to as “Black Saturday.”

The largest fire of the summer began at North Fork. It started on July 22 in Idaho’s Caribou-Targhee National Forest when a man who was out cutting timber thoughtlessly flicked his cigarette onto the ground. It smoldered there for an innocuous moment and then caught on an adjacent pile of kindling. There was no stopping the spread once it got going. The fire continued north and east from Idaho into Wyoming. In early September it was approaching the famous geyser Old Faithful. Air tankers flew overhead and dropped an all out aerial bombardment of fire retardant to protect the historic area. It was no use. The fire struck Old Faithful in early afternoon. Upwards of a thousand firefighters cleared brush and dug holes, but the fire couldn’t be stopped. The flames engulfed Old Faithful. Over the course of the summer, that fire originating in Caribou-Targhee covered the largest area. 

In mid-August, the Boise Interagency Fire Centered requested assistance from the Department of Defence. Many of the 3,500 fire fighters in Yellowstone had been working fourteen hours per day for weeks now. The military began to put soldiers through a two-day firefighting course. On August 23, two army battalions arrived in Yellowstone with eight helicopters. The military deployed over 11,000 personnel, from the army, navy, air force, marines, and the Wyoming national guard. 

Even with the fire fighting effort, the fires grew exponentially throughout the summer, doubling in size every week or so. By the end of August, it was clear that nothing could be done to stop the fires. Only rain or snow could quell the spread. The fire fighters persisted working to prevent the fire from damaging property or endangering lives. 

On September 11, quarter of an inch of snow fell, and the spread of the fire halted. By the final week of September, only a handful of fires were still burning. The last coals smoldered until the first week of November. The 1988 Yellowstone fire season was over. At final tally, the fire claimed 783,880 acres of Yellowstone, over one third of the Park’s the total area. The Greater Yellowstone Ecosystem—including Targhee and Gallatin National Forests—saw almost 1.5 million acres burned in total. More than 25,000 people were involved in combating the fire. Fire fighters dug more than 665 miles of firebreaks by hand, and 137 miles by bulldozer. They used more than 100 fire engines and 100 aircraft, including 77 helicopters which carried more than 10 million gallons of water. Aircraft dropped 1.4 million gallons of fire retardant in the park, and 18,000 flight hours were logged in the park. All said and done, control efforts cost the government $120 million. It was the largest fire fighting effort in the United States to date. 

They say that where there is smoke, there is fire. But it is also true that where there is fire, there is someone asking what caused it. It is a relatively straightforward question about the Yellowstone fires of 1988 which is fiendishly difficult to answer: What caused them to be so devastating?




When we observe another person engaged in a particular activity—tossing a cigarette into the woods, eating a twelve egg omelette, devoting their life to the consumption of heroin—we naturally want to know why. Their behavior, like a forest fire, must have been caused by something, and we instinctively seek recourse to assign blame to these causal factors. Psychologists call this process of blame assignment attribution

There are two kinds of factors to which behavior can be attributed. The first kind is dispositional factors. Suppose you’re at a restaurant with your friends and Jane orders a piece of chocolate cake. Why? Because she likes chocolate cake. That’s an explanation via disposition. Then suppose Sarah orders the cheese cake. Why? Well, one possibility is that she also ordered that kind of cake because she likes it. But another possibility is that Jane ordered the last piece of chocolate cake, which Sarah would also have preferred. In this case, Sarah’s behavior isn’t a reflection of her internal dispositions—she wanted chocolate cake just as ardently as Jane—but rather of external factors dictated by the situation in which she found herself. This the second kind, which we’ll call situational factors.

Dispositional factors appeal to causes within the person. Situational factors appeal to causes outside the person. Invoking a dispositional factor to explain someone’s behavior suggests that a different person in the same situation might have made a different decision. Given the choice between cheese cake and chocolate cake, Jane chose chocolate. But suppose your other friend, Mary, prefers cheese cake to chocolate cake. Her disposition will cause her to make a different choice than Jane. Different person, same situation: different outcome. Situational factors imply the opposite. In Sarah’s position, given no choice at at all, each of the girls would make the same decision. Their disposition doesn’t play into their behavior. Same person, different situation: different outcome. 

In 1984, the social psychologist Joan Miller ran an experiment to study how people use these two kinds of factors to explain behavior. Miller started by bringing a sample of forty middle class American adults into her lab at the University of Chicago. She asked each person to tell a story about a time when someone they knew behaved inappropriately. She called this deviant behavior. Then after they told the story, she asked them to give an explanation for why the person did what they did. 

Here’s an example from one of her participants.

Story: "A neighbor of mine—she and her husband, they talk to us about a great number of things. She recently told me with a certain amount of even pride how she itemized her taxes to get back even more from the government—really outright cheating. On giving to charity she declared the maximum. And she knows and we know—'cause she tells us—that they didn't give anything to charity at all.” 

Explanation: "That's just the type of person she is. She's very competitive.”

Her neighbor cheated on her taxes. Why? Because she’s competitive. That’s just what she’s like. This is an explanation via dispositional factors. 

Here’s another example.

Story: ”This involved one of the teachers I work with at school. It was a process of scheduling—something to do with scheduling. I came up with an innovative idea of organizing the scheduling, of what we should do. I talked to some of the other faculty members about it, and this first teacher picked it up and quickly went to the principal and presented it as if it were his own idea.” 

Explanation: "He was just a very self-absorbed person. He was interested only in himself.”

Again, it’s an explanation via dispositional factors. 

Here’s one more.

Story: “This involved a scholar in some other department, and she has got her PhD now. She wanted to publish four or five papers from her thesis. She produced some papers, but the thing is, her advisor, he put his name as first author and this young scholar as the second author. She was very hurt because that means usually the credit goes to the first author.”

Explanation: "She was his student. She would not have the power to do it (publish it) by herself."

This time it isn’t about what kind of person the student is. It’s about her relationship to her advisor, and her position in the academic community—which is to say, subordinate and low ranking. This is an explanation via situational factors. 


What Miller found is that her participants were three times more likely to attribute deviant behavior to dispositional factors than situational factors. Then, in a different version of the experiment, Miller asked her participants for the same kinds of stories and explanations but with one adjustment. Instead of asking them to tell a story about an instance of deviant behavior, she asked them to tell a story about a time when someone they knew did something good. She called this prosocial behavior. Her question was, do we attribute differently based on whether the behavior is good or bad?

I recently went a local coffee shop I like. It was busy, and the staff were occupied making drinks and food. All of the tables were taken. Just before I was about to leave I saw someone get up to go. He grabbed his coat and walked away, leaving all of his dirty dishes and mugs at the table. I took the table, but I had to bus his dirty dishes myself. Why did he leave his plates there instead of cleaning up after himself? I’ll let you make your own judgment, but my mind went to the intuitive explanation that he was just too lazy or inconsiderate to clean up his own mess. That’s a case of deviant behavior attributed to dispositional factors.

After I bused the dishes and purchased my coffee, I sat down at the table. Shortly after that the person next to me also got up to go. Now she actually cleaned up her dishes. What’s the difference between her and the first guy? Did I assume that she, in opposition to the first guy, was a good person? Well, no. Busing one’s own dishes is laudable, I suppose, but it’s also just good etiquette. It’s what people should be doing. Putting away your own dishes doesn’t make you a good person any more than does not running over pedestrians with your car. It’s just part of your duty to society. That’s a case of prosocial behavior, this time attributed to situational factors. 

This is essentially what Miller found. When someone does something inappropriate, attribute it to their intrapersonal faults. But when someone does something that accords with what we expect, we don’t attribute it to any especially valiant character but simple adherence to society’s ground rules. It’s much easier to convince people that you’re a bad person than a good one. Miller’s participants were twice as likely to use dispositional factors to explain deviant behaviors than than they were to explain prosocial behaviors. Though in both situations people relied on dispositional factors more often than situational factors. 

The results of this study, I think, are intuitive. If we read the stories that Miller’s participants told, we could, more often than not, independently come up with the same explanation that they did. We would also find ourselves attributing people’s behavior more often than not to their individual dispositions. And we’d be more likely to do so for the times when they behaved poorly than when they behaved nicely. 

But just because we agree doesn’t mean that we’re right. This is, after all, just our intuition, and not a scientific investigation into the causes of behavior. What actually causes behavior is more complicated, just as what actually causes a forest fire is also more complicated than what we might initially imagine.  




Let’s look at a simpler case of fire. Suppose you went camping with your family, and you wanted to light a campfire. What would you need to start the fire? You would need a primary fuel source to sustain the fire, most likely dry logs. You would need something that’s highly flammable, like small sticks and newspaper, for kindling and tinder. And you would need a match or a lighter to ignite the flame. To start a campfire, you would light the match which would in turn ignite the newspaper, then the smaller sticks, then the larger logs. 

So, what caused the campfire to start?

Well, each element is necessary to start and maintain the fire. If you didn’t have the match, the newspaper aren’t going to combust spontaneously. If you didn’t have the newspaper, then it would be difficult to directly light the kindling with a match. If you didn’t have logs, then you need quite a bit of kindling to sustain the fire. But it doesn’t seem right to say that the matches or the kindling or the logs cause the fire. You could crumple up some paper and build a teepee of sticks and logs on top of it and let it sit there all day. It isn’t going to ignite. If you hadn’t have gathered the supplies, set it up, struck the match, and nurtured the flame, then there would be no fire. In short, what really caused the fire is you. 

That much seems straightforward. There are several things that are instrumentally helpful for starting a fire, but at the end of the day the most important ingredient is the intent to do so. Our tendency as humans is to use this intuition about something small, like a campfire, and use it to extrapolate up to something big, like a forest fire. We isolate the cause of a forest fire in a single ignition event—a single lightning strike or a single cigarette tossed into the woods.We use terms like human-caused, or naturally-caused, or lightning-ignited. These explanations are handy because they’re compact. And to some extent these explanations make sense, too. What would have happened if the lightning hadn’t struck that tree? The fire would never have even gotten started. Just as if I hadn’t built the campfire and struck the match, then it would have never ignited. It’s a matter of counterfactuals. If things had happened differently than they actually did, then there would have been a different outcome. 


But forest fires aren’t like campfires. What defines a forest fire is not that it ignites, but how far it spreads and how severe it gets. Sure, campfires can vary in size. But the difference between a small campfire and a large campfire is the difference between a modest smolder and flames as tall as a person. That difference in magnitude doesn’t hold for forest fires. About 75% of forest fires are small. They burnout on their own in less than a quarter of an acre. The difference between a small forest fire and a large forest fire, then, is the difference between a cluster of a few burning trees and a charred landscape covering an entire third of Yellowstone National Park. 

The counterfactual account of igniting the fire with lighting doesn’t explain why the fires were so severe in 1988, but not in 1987 or 1986 or any other year. It doesn’t even explain why some fires in 1988 started then quickly burned out and why some went on to burn swathes of the park. In the summer of 1988, Yellowstone had two thousand lightning strikes per day. Of those, only forty-two started fires. If I were trying to build a campfire, and it took me two thousand attempts to get one fire going, then we would probably come to a different conclusion about my ability to cause fire. When it comes to campfires, if you can cause one then you can cause them all. But that’s not true of forest fires. Getting the fire going isn’t the hard part. It’s keeping it going. 

Ignition events are proximate causes. They explain why the fire started at that particular moment. But they don’t necessarily explain why one fire spreads and another doesn’t. In other words, if that particular lightning strike hadn’t started the fire, then another one would come along shortly and done so. The 1988 fires, in a sense, were blazes just waiting to ignite. It didn’t matter which lightning bolt got it going. In order to really understand a forest fire, you have to look at its ultimate causes.




I told you about one of Miller’s subjects, and how she told the following story: 

“This involved a scholar in some other department, and she has got her PhD now. She wanted to publish four or five papers from her thesis. She produced some papers, but the thing is, her advisor, he put his name as first author and this young scholar as the second author. She was very hurt because that means usually the credit goes to the first author.” 

Then instead of calling the guy a jerk, the partipant gives a situational explanation. She was his student, and that’s just how things go in that relationship. When you first heard it, did that sound a little off to you? It certainly seems to me like there are fair grounds on which to critize that professor’s behavior. The difference between that story and the ones that preceded it is that it was told to Miller not by an American, but by someone from Mysore, a city in southern India. There is a longstanding observation that Westerners tend to focus on people as individuals, whereas Easterners tend to view people according to the societal roles they play. Miller ran her studies on two different populations, one of Americans and another of Hindus, to see how this cultural difference played out in the way people attributed behavior to dispositional versus situational factors. 


Here’s another story from one of her Hindu participants, translated from Kannada, the original language in which the interview was conducted.

Story: ”I had to construct a house, and for that I had given advance money for Agent A to do that construction work. Agent A had promised—he had given in writing—that he would do that particular work. I gave him an advance of 1,500 rupees. He utilized it for his personal purposes, and then he never did that work or returned the money. That man, he deceived me up to the extent of 1,500 rupees. That's a great injustice. But I can go to the court of law. I have the documents, everything.”

Explanation: "The man is unemployed. He is not in a position to give that money.”

It’s another explanation via situational factors. The construction worker cheats the man out of his money and the victim chalks it up to the fact that the guy is poor. Everyone knows that you can’t get reimbursed with money someone doesn’t have. That’s just how things work. 

Again, I’m not inclined to make the same attribution. I would probably want to say that the guy is a lowlife, and that’s why he cheated me out of my money. What Miller found was that Hindus were more likely to attribute behavior to situational factors than dispositional factors, the opposite of American sensibilities. It’s the difference between calling someone lazy for leaving their dishes versus acknowledging that that’s often how it works when you go out to eat—someone else cleans up after you. 

Here’s another example from Miller’s study. 

Story: "This concerns a motorcycle accident. The back wheel burst on the motorcycle. The passenger sitting in the rear jumped. The moment the passenger fell, he struck his head on the pavement. The driver of the motorcycle—who is an attorney—as he was on his way to court for some work, just took the passenger to a local hospital and went on and attended to his court work. I personally feel the motorcycle driver did a wrong thing. The driver left the passenger there without consulting the doctor concerning the seriousness of the injury—the gravity of the situation—whether the passenger should be shifted immediately—and he went on to the court. So ultimately the passenger died.”

Explanation: ”It was the driver's duty to be in court for the client whom he's representing; secondly, the driver might have gotten nervous or confused; and thirdly, the passenger might not have looked as serious as he was.”

Again, the Hindu participant favors dispositional factors. It’s the driver’s duty to be in court. He was hurried and might not have noticed how serious the injury was. Only the remark about getting nervous and confused hints at a dispositional explanation.

Based on these studies, Miller wondered if Hindus and Americans really attributed different causes to similar situations. What if, instead, Hindus and Americans tended to pick different situations to tell their stories about? What if Hindus just chose stories that lended themselves to situational explanations and Americans chose stories that were well explained in dispositional terms? So she showed Hindu stories to Americans. Here’s an American’s explanation for the story above:  ”The driver is obviously irresponsible; the driver was in a state of shock; the driver is aggressive in pursuing career success.” Americans still favored disposition, regardless of the story they were presented. 

So, who is right? Is it the American who wants to explain behavior in terms of disposition? Or is it the Hindu who say the that it’s the situation that matters? 




The seminal text of the field of user-centered design is a book published in 1988, the same year as the Yellowstone Fires, called The Design of Everyday Things, written by the cognitive scientist Don Norman. He begins the book with a story—one which I think will be a familiar experience to anyone living in the modern world:

A friend told me of the time he got trapped in the doorway of a post office in a European city. The entrance was an imposing row of six glass swinging doors, followed immediately by a second, identical row. That’s a standard design: it helps reduce the airflow and thus maintain the indoor temperature of the building. There was no visible hardware: obviously the doors could swing in either direction: all a person had to do was push the side of the door and enter. 

My friend pushed on one of the outer doors. It swung inward, and he entered the building. Then, before he could get to the next row of doors, he was distracted and turned around for an instant. He didn’t realize it at the time, but he had moved slightly to the right. So when he came to the next door and pushed it, nothing happened. ‘Hmm,’ he thought, ‘must be locked.’ So he pushed the side of the adjacent door. Nothing. Puzzled, my friend decided to go outside again. He turned around and pushed against the side of a door. Nothing. He pushed the adjacent door. Nothing. The door he had just entered no longer worked. He turned around once more and tried the inside doors again. Nothing. Concern, then mild panic. He was trapped! Just then, a group of people on the other side of the entranceway (to my friend’s right) passed easily through both sets of doors. My friend hurried over to follow their path.

How could such a thing happen? A swinging door has two sides. One contains the supporting pillar and the hinge, the other is unsupported. To open the door, you must push or pull on the unsupported edge. If you push on the hinge side, nothing happens. In my friend’s case, he was in a building where the designer aimed for beauty, not utility. No distracting lines, no visible pillars, no visible hinges. So how can the ordinary user know which side to push on? While distracted, my friend had moved toward the (invisible) supporting pillar, so he was pushing the doors on the hinged side. No wonder nothing happened. Attractive doors. Stylish. Probably won a design prize.

Opening doors should be easy. Yet, like Norman’s friend, we’ve all had the experience of failing to get them right. Sometimes we push doors that want to be pulled. Sometimes we pull doors that want to be pushed. And, as Norman says, sometimes we “walk into doors that neither pull nor push, but slide.” Product design doesn’t get much simpler than a door. There are only two things that a door can do. It can be either open or closed, and all anyone has to do to use a door successfully is to make it go from one of those stated to the other. 

Why, then, do we have consistent trouble with something so simple? 

The explanation from user-centered design is a concept called an “affordance.” An affordance is what a product can do. For example, doors afford opening. The product is the door, the affordance is being able to open and close. Chairs afford sitting. The product is the chair, the affordance is being able to sit on it. The term comes from psychologist James Gibson, in his 1979 book The Ecological Approach to Visual Perception. Gibson writes, “The affordances of the environment are what it offers the animal, what it provides or furnishes, either for good or ill. The verb to afford is found in the dictionary, the noun affordance is not. I have made it up. I mean by it something that refers to both the environment and the animal in a way that no existing term does. It implies the complementarity of the animal and the environment.” An affordance, in other words, isn’t a property. It’s a relationship. An affordance relates, in Gibson’s case, the animal and the environment. For designers, it relates the person and the product.

It is only partially accurate to say that a door affords opening. A door might afford opening, but only if you know where to push. Norman’s friend didn’t, and he got stuck. The problem was that Norman’s friend couldn’t tell what the affordances were. The door didn’t have what designers call a “signifier.” When all goes according to plan, how do you know whether to push or pull on a door? In one case, it could simply say push or pull. That’s one kind of signifier. If you walk up to a door and it has a push sign on it, then it’s clear what you’re supposed to do. But there are also more subtle, more ingenious kinds of signifiers. For example, what if you walk up to a door and instead of a label, it just has a flat metal plate on the side of it? Clearly, it’s a push. There would be no way to pull it, even if you wanted to. That metal plate is a signifier. Likewise, if there’s a bar that extends from the door and runs across waist height, then the door wants to be pulled. But on some doors that bar runs across both sides. Which one is pull, and which one is push? The bar is a signifier. But it might be signifying the wrong thing. In short, people have trouble with doors when the signifiers don’t match the affordances. 

The power of the idea of affordances and signifiers—and the basis of user-centered design more generally—is that the person and the product form a system. And the system doesn’t function as intended unless each part of it works in concert. Designing a product well isn’t just about building in functionality. It’s about designing a system which encourages a person to behave in a way that unlocks that built in functionality. So when we ask about who was at fault with a misuse of a product such as pushing the pull door, we can’t just look for “human error” and chalk it up to that. We have to look at the system as a whole. 




One thing that is easy for the public to misunderstand about forest fires is that they are not necessarily detrimental to the ecosystem. We see hundreds of thousands of acres burning in Yellowstone and think that something very valuable and very special has been irrevocably lost, as if our own neighborhood was on fire. But fires are a natural part of the ecosystem, especially minor fires. Minor, lightning-ignited fires burn out the highly flammable undergrowth of the forest as well as provide other benefits critical to the forest’s ecology.

Foresters didn’t understand this until the seventies. From 1945-1971, Yellowstone’s park managers practiced pretty stringent fire suppression. The procedure was to put a fire out as soon as it was identified, regardless of the cause. So when the fires of 1988 were still raging, the park managers’ reflexive response was to point the finger at earlier suppression policies. These decades of unnatural fire suppression had built up a dense undergrowth of highly flammable material, and not allowed the ecosystem to thin it out in its usual way. But that wasn’t quite right. For one thing, the forests that were on fire had trees that were hundreds of years old. Thirty or so years of fire suppression wouldn’t dramatically change the structure of the vegetation. And for that matter, all types of vegetation with different fuel loads had what are called crowning fires, where the flames reach all the way to the top of the tree, shooting hundreds of feet into the air, instead of remaining close to the ground, as they would had they relied purely on undergrowth. 

On the other side of the argument, mostly from those outside the park, were those who said that the more “natural” policy of letting fires didn’t do enough. They maintained that the fires got out of control because park officials let them burn too long without taking action. That contention doesn’t seem quite right, either. For one thing, the usual methods for combating fires proved ineffective. “In late August,” writes one account, “one flank of the North Fork fire broke over a containment line that had held for a month. Most of the hundreds of miles of fireline dug in greater Yellowstone in 1988 could not halt fires that were capable of hurdling the Grand Canyon of the Yellowstone, unvegetated geyser basins, highways, and parking lots. When a fire reaches that intensity, it is not only impossible to contain, but foolhardy to try, for a fire crew in front of the advancing flames can easily be overrun or trapped between fires.” Ultimately, any policy-based explanation, whether too much or too little, probably overestimates the amount of impact human activity had on the course of the fires. 

The whole of the American west had heavy fires in 1988. Over 3.7 million acres burned in the lower 48 states. Several million more burned in Alaska. By August, more than 15,000 firefighters were at work across the country. A lot of it came down to weather, which was unpredictably severe in 1988. The year before had ended in mild drought. For example, Yellowstone’s winter snowpack was only 31% of the long-term average. But it was nothing to raise the alarm about.  Then the spring of 1988 turned out to be especially wet, with precipitation 55% above average in April and 81% above average in May. Wet spring means increased undergrowth, the kind of easily ignitable fuel that fires feed on. Rainfall abruptly plummeted in June, July, and August during the fire season. Precipitation was only 36% of the long-term average. The summer of 1988 turned out to be the driest of the 112 years on park records. 

Fire ecologists use a measure of flammability called 1,000-hour fuel moisture. It measures the amount of moisture in a fuel source, like a log, that would dry out within 42 days (1000 hours is 41.67 days to be exact). Kiln-dried lumber, for instance, has a 1,000-hour fuel moisture of 12%. The average 1,000-hour fuel in the park during a typical fire season is usually between 14-18% moisture. Fires usually don’t start above 18%. In the 14-18% range, fires will ignite but they aren’t likely to spread quickly. If a tree in that range gets struck by lightning, it’s more likely to smoke than to burn. The threshold at which wood becomes susceptible to combustion is 13%. An extreme drought would be around 5%. Much of the summer of 1988 was below 18% moisture, frequently as low as 10%. A record low of 6% was recorded at Tower Falls on August 22. And that’s for the trees that are still standing. Much of the forest would have been littered with fallen trees that would drop as low as 5% moisture. Another measure of moisture, Palmer Drought Severity Index, marks extreme drought at -5 and extreme wetness at +5. Yellowstone’s rating in 1988 was -6, the lowest year of any on record. 

To make matters worse, dry storms were frequent in July and August. These rainless thunderstorms increased lightning strikes by twice as much as usual. They brought winds up to sixty miles per hour. 

Then there were even more insidious causes than weather. Outbreaks of the mountain pine beetle Dendroctonus ponderosae in the 25 years leading up to the 1988 fires killed many older trees, creating the kind of heterogenous mix of forest which fires loves to feed on. Another less obvious contributor to fire behavior is what geographers call aspect, or the position of a hill’s slope—its steepness and the direction it faces. South-facing slopes get more sun, leading to earlier snowmelt. More sun in the summer also means more heat, increasing fire risk. North-facing slopes tend to have a higher moisture content and therefore have a thicker undergrowth with a higher fuel load. Fires tend to spread more quickly upslope, because heat rises and flames jump upward. The fires of 1988 generally went from southwest to northeast, generally going down-slope on north-facing hills. That is, they moved slowly. The combination of slower movement and lusher fuel load likely made for a heartier spread of the fire. 

Ignition events, then, are merely proximate causes of forest fires. They explain why a particular fire started at a particular moment. But they do very little to explain the actual consequences of how the fire behaved. Unfortunately, these proximate causes are much easier to quantify and point a finger at in blame. If we want to understand the ultimate causes of the 1988 Yellowstone fires, then we have to look at the ecosystem at a whole, a tangle of interconnected considerations within a vast and complex system, not just at a single ignition event. This is the Forest Fire Problem. 




Imagine going to your local supermarket and picking up a package of Swiss cheese, the presliced kind. When you’re back home in the kitchen, take the cheese out of the package and place each slice out in a row in front of you. One by one, take a slice at random from the row and make a stack. Grab a pencil from the drawer. Now, try to stick the pencil into one of the holes of the top slice in the stack so that it all the way through the other slices to the bottom of the pile. For any one slice, it’s easy to stick the pencil through. But the more layers you add, the more difficult it is to get through. This is the idea behind the Swiss cheese model of human error. 

One of Norman’s colleagues, James Reason, uses the Swiss cheese model to explain how to design a system that is robust against human error. Each part of the system protecting against human error is like a layer of cheese. “In an ideal world,” writes Reason, “each defensive layer would be intact. In reality, they are more like slices of Swiss cheese, having many holes—although, unlike in the cheese, these holes are continually opening, shutting, and shifting their location. The presence of holes in any one “slice” does not normally cause a bad outcome. Usually this can happen only when the holes in many layers momentarily line up to permit a trajectory of accident opportunity.” The Swiss cheese model helps to explain why it is when something goes wrong in a complex system, there isn’t one single cause. Something has to go wrong at each slice, allowing the holes to line up. If you changed one of the slices so that the hole was in a different position, things might have turned out differently. 

Each part of the Yellowstone ecosystem acted like a slice of Swiss cheese. There was the weather, the beetles, the aspect of the hills, the fire suppression policies, the drought, the wind, the lightning. The fires were so devastating because, unlikely though it may have been, each of the holes in these slices aligned to allow the pencil to go all the way through the bottom. If you had shifted one of these slices, perhaps it would have prevented the whole thing from getting as bad as it did. But you can’t single out any one of the slices as the culprit. You have to consider the stack as a whole. 

If we take the Swiss cheese model seriously, then there are three ways to prevent things from going wrong in a system. This is true of man made systems, like a nuclear power plant, moreso than natural ecosystems, though some of the principles may still hold. The first way is to add more layers. With each layer you add—assuming you’re drawing the slices randomly, and not in the order that they came in the package—you’re making it less and less likely that the holes will line up. The more defense mechanisms you have, the more probable it is that one of them will catch an issue and prevent the pencil from going all the way through. The second way is to reduce the number of holes. If you imagine a slice that is more holes than cheese, it would be easy to get the holes to line up. But the fewer the holes, the less likely it is they will line up. The third way is to have a preemptive alert when the holes are in alignment. If the alarm sounds whenever a single hole can go all the way through, then the stack can be rearranged before someone puts the pencil through. 

Each hole in the cheese represents an affordance. It affords the opportunity for something to go wrong with the system. Each of the factors of the Yellowstone fires can be thought of an affordance. Combustible forests, built up by a wet spring and dried by a cloudless summer, afford ignition to stray lightning strikes or misplaced cigarettes, in much the same way that a door affords opening to a human that wants to pass through it. The affordances of a system—a relationship between the product and the person, the environment and the animal—determine the ultimate causes of takes place. 




The implication of the Forest Fire Problem on human behavior lies in what psychologists call the fundamental attribution error. It is an asymmetry in the way we attribute the causes of our own behavior versus the behavior of others. When someone else does something wrong—Miller’s deviant behavior—we attribute it to dispositional factors. They’re just not a good person. But when I do something wrong, it is because I had no other choice. There weren’t any better options. We cheritably attribute our own deviant behavior to situational factors. My favorite example of the fundamental attribution error at play is driving. When I cut someone off in traffic, it’s because I had no other choice. I need to get where I’m going and making a quick lane change was the only way to do that. But when someone else cuts me off in traffic, it’s because they’re an asshole. Simple as that. But if we realize—when we know the whole story, as with our own behavior—that situational factors constitute a valid explanation of our own behavior, then what reason do we have to believe that they do not also constistute a valid explanation for the behavior of others?

In the same way that it is convenient but inaccurate to attribute the causes of a forest fire to proximate concerns, such as ignition events, we are inclined to attribute the causes of human behavior to dispositional factors. These are mere proximate causes, explaining why the behavior occurred at that particular time and place, but not giving a full accounting of why it happened at all. The proximate causes of human behavior are dispositional. But the ultimate causes are situational.

The implication is that if you want to understand the ultimate causes of someone’s behavior you have to look, so to speak, at the ecosystem in which they take part. In order to understand a person, you have to understand their context. This may include a culture, an institution, a nation, a job, a relationship, anything that situates the person in a context larger than themselves. 

This is the primary difference between anthropologists and psychologists. Psychologists seek to understand people by diving further and further into their heads. The premise of psychology is that if you can catalogue and account for every disposition that could occur in the human mind, you’d have a complete picture of human behavior. Anthropology, on the other hand, leaves consideration of individual people alone. Anthropology instead seeks to understand the context. 

Psychology is fundamentally about trying to find the lowest common denominator among all humans—what are the core mental faculties and experience that we all share, regardless of where and when we were born? This is both is its most admirable quality and its biggest short coming. For when we try to reduce the human experience to something in common to everyone, we lose the important specificity that characterizes it. We may very well all share the same basic psychology, but it is our context that informs what our personal psychology develops into. This, of course, is not to discount completely psychological research or its insight. Instead, like empathy and perspective-taking, it is a single tool in an array of implements for understanding others. And also like empathy and perspective-taking it can be a tool that works best on those in the same faction as ourselves, rather than for putting a useful interpretation to the things that make us different. If we want to understand those from other factions, it is crucial that, like anthropologists, we understand their context. 




In 1992, a group of Dutch psychologists led by Carolien Martijn did a study on how we diagnose  morality. That is, how do we know whether or not someone is a good person? Martijn and her colleagues wanted to know how the way we make this diagnosis differs from how we diagnose someone’s ability. How do we know when someone is talented? 


The idea in both cases is that there is some action that we can observe—immoral behavior, winning championships, acts of kindness, prowess on the gridiron—and we use these observations to diagnose a character trait which is itself unobservable. You can’t see whether someone is a good or bad person. You can only infer it based on what you observe them doing. Likewise, talent is not something that we perceive directly. It’s the sum of relevant characteristics that we can observe, such as strength, a winning record, and one’s batting average.

The experiment was very simple. They gave participants sentences like, “Imagine a person who ___” where the blank is a behavior. Then, “How likely is it that this person is ___” where the blank is a trait. In one case the behavior was about morality. For example, “Imagine a person who volunteers at a homeless shelter every Sunday. How likely is it that this person is moral?” The behavior is volunteering, the trait is morality. The contrasted this with a case where they asked about ability. For example, “Imagine a person who makes 90% of their free throws. How likely is it that this person a talented basketball player?” Unsurprisingly, for sentences that have positive behaviors, volunteering on Sunday and making free throws, participants infer that the person described must have the corresponding trait, being moral or good at basketball. 

The surprising thing that Martijn and her colleagues found was how their participants interpreted the opposite behavior. Think of someone who you consider to be an honest person. What if you found out that they had just recently been fired from their job because they had lied to their boss? You might reevaluate what you thought about them. Perhaps they weren’t that honest after all. Now, think of someone who you remember from college as being really good at chemistry. What if you found out that person had actually received a D on important exam in a chemistry course? You wouldn’t write them off as a poor chemistry student. You would, instead, probably suppose that they had been drinking the night before and were hung over on the day of the test, or that they had had a family emergency that week and didn’t have time to study. In both these situations we have an idea about what a person is like—they’re honest (a moral trait), or they’re good at chemistry (an ability). Then you observe something that contradicts that initial idea. What Martijn found was that the contradiction was damning when it came to honesty but not when it came to chemistry. 

Good people don’t do bad things. If they do, then they probably weren’t all that good of a person in the first place. But talented people can have an off day. It doesn’t mean they aren’t talented. It was just an off day, and they’ll be back at it tomorrow. This is, at least, our intuition. 

In this chapter I’ve argued that situational factors are the ultimate causes of behavior. This should make us suspicious of our inclination to make categorical judgments about someone’s character based on the behavior we observe. This is, I understand, dramatically different from the way we usually think. Especially as Americans, we have taught to think about the causes of behavior in dispositional terms. In order to make this point further, I want to look in the next chapter at a case study. This is the story of a man from Rwanda committing—to put it mildly—deviant, immoral behavior. Throughout the chapter we will consider the ultimate causes of his behavior. You can decide for yourself. 

Are they dispositional, or are they situational?

draft 08/26/2018

Next: Chapter 6. As with a Harvest.