The Wrong Impression
from Irrationality by Stuart Sutherland

The leading part in the film Jaws was a man-eating shark. The screening of the film caused a sharp drop in the number of swimmers off the coast of California, where the occasional shark is to be found near the beaches. It has been calculated that the risk of swimmers being snapped up by a shark is very much less than the risk of their being killed in a road accident while on the way to the coast. People do not take account of the true facts they go upon what makes the deepest impression, or on what first comes to mind.
As a further illustration, consider these two questions: 'Are there more words with "r" as the first letter than with "r" in the third position?' 'Are there more words beginning with "k" than with "k" as the third letter?' Unless you sense that there is some sort of trick being played, you are likely to answer yes to both questions. But you would be wrong -there are more words with 'r' or 'k' in the third position than there are words beginning with each letter. The mistake is made because words, both in dictionaries and in our minds, are arranged by their initial letter. It is easy to retrieve from memory words beginning with 'r', like 'roar', 'rusty' and 'ribald', but much more difficult to recover words like 'street', 'care' and 'borrow', despite their greater frequency. Lest you think this experiment is unfair because nobody could know the answer without counting up words in a dictionary, here is a variant on it in which knowledge is not involved. The question is, 'Are there more words ending in "-ing" than ending in "-n-" (that is, having "n" as the penultimate letter)?' Most people think that 'ing' endings are more common, but in fact '-n-' must be more frequent since all '-ing' endings have 'n' as the penultimate letter in addition to many other words (like 'fine'). People can recall words ending in '-ing' more readily than those ending in '-n-' and they do not stop to go through the simple argument outlined.
Judging by the first thing that comes to mind is called the 'availability error'. I have made it the first error to be described because it permeates all reasoning and, as we shall see throughout the rest of the book, many other specific errors are in reality just further instances of it. Suppose you are thinking of buying a car and you mention it to a friend. He gives you a glowing account of his own car. Deeply impressed, you rush out to buy the same model, only to find that it is totally unreliable and has an outrageous thirst for petrol. The immediacy and salience (availability) of his description have made you forget all the statistics to be found in consumer magazines. You have also committed a second common fallacy, which will be discussed later: no matter how good your friend's car is, it may or may not be representative of that model in general. No two cars of the same type perform equally well and he may simply have been lucky with his.
There are dozens of experiments demonstrating faulty reasoning caused by the availability error. In one extreme case, subjects first had to learn a list of words (a task much beloved by psychologists). The words were the same for all subjects, except that those for one group included four terms of praise -'adventurous', 'selfconfident', 'independent' and 'persistent'. For a second group, they included four disparaging words -'reckless', 'conceited', 'aloof' and 'stubborn'. After learning the
words, all the subjects read a short story about a young man who had several dangerous hobbies, thought well of his abilities, had few friends, and rarely changed his mind once it was made up. Finally they were asked to evaluate him. Although it was made clear that the previously presented list of words had no connection whatsoever with the man in the story, those who had learned the favourable adjectives thought much better of him than those who had learned the unfavourable ones. The words were in their minds (available) at the time they read the story and hence had coloured their interpretation of it. If items such as the words learned in this experiment can affect subjects' interpretation of something completely unconnected, how much more strongly are people likely to be affected by aspects of a situation that are highly salient and are closely connected with whatever is being judged?
In order to expound the next experiment it is necessary to explain a fiendish game, known as 'The Prisoner's Dilemma'. It is based on the following scenario. Two people are in gaol for a crime they are thought to have committed together. The governor tells them that the length of their sentence will depend, in a rather complicated way, on whether or not they confess to the crime. The sentences are as follows:

1. If one confesses and the other does not, the one who confesses goes free, while the other gets twenty years in gaol.
2. If neither confesses, they will each be given a two-year sentence.
3. If both confess, they will each spend five years in gaol.

The dilemma facing the prisoners is whether or not to confess -they are in separate cells and neither knows what the other will do. The best outcome is if neither confesses, because the combined length of time they spend in gaol is only four years. But not confessing is dangerous since if the other prisoner confesses, the one who does not spends twenty years in gaol.
The game is not as far removed from real life as might at first appear. In the long run it would clearly be beneficial for all countries to reduce the emissions of carbon dioxide, the principal cause of the greenhouse effect, the results of which may prove disastrous. On the other hand, such reduction is expensive: it entails using fewer fossil fuels to produce energy or using less energy. If all countries agree to a reduction, all will benefit. But if a few refuse (as the USA is at the moment doing) but most agree, those refusing will benefit both by saving the costs of reducing the emissions and by the reduction in the size of the greenhouse effect caused by other countries reducing emissions. To give a more mundane example, people have to decide whether surreptitiously to water their gardens in a drought. If everyone did so, the water supply might be exhausted and the effects would be disastrous for all. On the other hand if just a few people act in this antisocial way, they will benefit at the expense of a small loss to the community. These situations exactly parallel the Prisoner's Dilemma, a game which is often used by psychologists to measure people's willingness to collaborate. Making the choice that if made by both leads to the minimal loss to the two combined is known as 'collaborating', the other choice is 'defecting'-if made by one person, it leads to a large loss for the other if he collaborates.
The game has produced endless speculation among philosophers because it is unclear what is the rational thing to do. Until recently the puzzle has remained insoluble. Even if your opponent has collaborated for some time, you never know when he will defect, thus landing you with a substantial penalty if you are collaborating. Interestingly, there is now an indication of the best way to play the game. A strategy is any rule adopted by a player for a number of rounds against the same opponent, for example, 'Always defect', or 'Defect randomly half the time, collaborate on the other half'. In a recent study, a large number of different strategies were proposed by mathematicians and others. These strategies were tested against one another on a computer. The best strategy, that is the one that maximised the player's gains against any of the other strategies tested, turned out to be 'Collaborate on the first round and then copy whatever the opponent did on his last move'. This strategy punishes the opponent for defecting and rewards him for cooperating. Its success is particularly interesting since it suggests that behaving altruistically (on some occasions) can secure the highest possible gain for the person doing so: altruistic behaviour, whose existence has long puzzled evolutionary theorists, can therefore help to achieve one's own ends, and hence survival. Although in real life the Prisoner's Dilemma rarely occurs in the same form more than once, it occurs over and over again in different forms. Hence the strategy outlined is still likely to be the best.
In experiments, monetary rewards or penalties are substituted for gaol sentences, if only in order to make it easier to find volunteers as subjects. The 'prisoners' are usually faced with two buttons -call them 'C' and 'D' for 'co-operating' and 'defecting'. They are given rules such as:

If they both press C, they receive £5 each.
If one presses C and the other D, the first is fined £10 and the second receives £10.
If both press D, they are each fined £1.

In the experiment with which we are concerned here, one group of subjects was exposed to a touching radio programme about someone who had given a kidney to a complete stranger in need of a transplant, while the other group listened to an account of some particularly vile human behaviour, to wit an urban atrocity. Pairs of subjects were then asked to play Prisoner's Dilemma. Those who had listened to the touching story of the kidney transplant collaborated much more than did those who had heard about the atrocity, although the stories had absolutely nothing to do with the game they were playing. Once again, recent experiences even when they are irrelevant can make people behave more or less selfishly.
Here is a rather different but equally irrational example of faulty judgement directly caused by the availability error. Subjects were read lists of names of men and women, some of them fictitious, some the names of famous people. All were called by both their Christian name and surname, so the sex of each was obvious. Each list contained approximately half female and half male names, and the subjects had to judge whether there were more men's or women's names. When the men were all famous, like Winston Churchill or John Kennedy, and the women were not well known, subjects thought there were more men than women, and vice versa when the women were famous and the men unknown. Names of important people made more impact (were more available) than those of unknown people and the judgements were based on this factor rather than on the actual frequency of men and women in each list.
Before discussing what makes material available, it is worth considering some examples of the availability error being craftily put to use in real life. The organisers of lotteries give maximum publicity to past winners, and of course say nothing about the great majority who have won no prizes. By publicising winners, they make winfling foremost in the minds of potential buyers of tickets and hence make them believe that they are more likely to win than they really are. Similarly the rattling of coins disgorged from a fruit machine is intended to call people's attention to the possibility of winning money: the machine maintains a steadfast silence at other times.
People's tendency to base their judgements on what is available is manipulated by shopkeepers throughout the world and also by otherwise respectable publishers. Would you be more likely to buy a book costing £5.95 or one costing £6.00? The important figure is the number of pounds: it is therefore more available than the pence and people seize on it, disregarding the fact that in the case quoted there is only a difference in price of 5p.

One may ask what makes something 'available'. The experiments already cited show that recently presented material is available, but it has also been found that anything that produces strong emotion, that is dramatic, that leads to the formation of images and that is concrete rather than abstract is also readily available. A murder committed by a Muslim or Japanese will receive far more coverage in the newspapers than one committed by John Smith: it is more dramatic, less of an everyday occurrence and hence more available. Moreover, people are likely to have stronger emotional feelings about Muslims and Japanese than about John Smith.
A massive amount of work has been done on images, which affect all aspects of our mental life. If someone has to learn to pair one word with another, for example to say 'car' when given the word 'dog', he learns much faster if he is told to form an image connecting the members of each pair, for instance, by imagining a dog sitting in a car. Moreover, people have an amazing capacity to remember pictures. After being shown 10,000 photographs just once they can correctly recognise almost all of them a week later. This is in marked contrast to the very poor memory for isolated words. Later in the chapter, I shall illustrate the power of images to evoke irrational responses by their use in advertising.
There have been several experiments showing that concrete material is more available than abstract material. One was again based on the Prisoner's Dilemma. The subject's partner in the game was not a real partner. The moves made by the 'partner' were in reality made by the experimenter and consisted of a prearranged proportion of co-operating and defecting moves. In one condition, subjects learned what move their partner had made by watching which of two lights flashed. In the other, they were passed a handwritten note through a slit. One might think that this would make little difference to how they thought of their partner, but in fact it made a great deal. When notes were passed the subjects saw their partners' moves as being made much more deliberately, that is they saw them as intending to cooperate or defect. Moreover, they showed more trust in a partner who made co-operative moves when notes were passed than when communication was made by the lights, for they made more co-operative moves themselves in the former case than in the latter. Similarly, they distrusted a defecting partner more in the condition with notes than in that with lights. It is extraordinary that whether a light is flashed or a note passed should make such a difference to people's behaviour: the note is a concrete reminder that they are dealing with a real person, who is more or less trustworthy.

The availability error is responsible for a large number of irrational judgements in real life. Do you regard fairgrounds as dangerous? Certainly most people do. There is the Big Wheel with its carriages turning precariously in the air, the roller-coaster with its frightening bends and changes of speed, the Octopus subjecting you to massive centrifugal force while rocking you violently to and fro, and many other machines moving in a variety of contortions. Yet most people (including myself until I learned the facts) are wrong. According to a report of the British Health and Safety Committee, if you cycle on main roads for an hour you are forty times as likely to be killed as if you spend the same length of time riding fairground machines, and you are seven times as safe on them as when driving a car. Fairground accidents are of course dramatic and well publicised: they are 'available'. It is also known that people grossly overestimate the chances of dying a violent death, for example, in an air crash or in street riots. In one study it was found that people think they are twice as likely to die of an accident as from a stroke; in fact forty times as many people die of strokes as from accidents. The reason for this false belief is that, although most people die in their beds, air crashes and violence are constantly reported in the media and are highly dramatic: they are therefore 'available'.
Not only do people hold irrational beliefs about the frequency of violence, but they are driven by their beliefs to wholly irrational actions. In 1986 the number of Americans visiting Europe as tourists showed a sharp drop. They had been scared away as a result of a few muchpublicised plane hijackings and possibly by the American bombing of Libya. But they had failed to take account of the less-publicised prevalence of violent crime in the US: in fact Americans living in cities put themselves at greater risk of meeting a violent death by staying at home. Exactly the same irrational refusal to fly occurred during the Gulf War.
Sometimes the availability error does seemingly drive people to act rationally. In California the number of insurance policies taken out against earthquakes increases steeply after a quake, but then drops gradually until the next one. But even this behaviour is not really rational since whether you take out insurance should depend not on when the last earthquake was but on the probability of earthquakes in the future. Again, after Mrs Ford and Mrs Rockefeller developed breast cancer, large numbers of American women rushed to the hospitals to have diagnostic tests. They had hitherto been completely unmoved by government warnings that they should have tests at regular intervals.
There is a more everyday example of the effects of availability, which is familiar to everyone who drives. A driver who has just passed an accident almost invariably slows down. The accident makes available the possibility that he too will crash: unfortunately the effect wears off within a few miles. The sight of a police car has the same result.
The availability error is just as common in the professions as in everyday life. It is known that a doctor who has recently seen a number of cases of a particular disease becomes more prone to diagnose that illness in patients who have not got it. This would of course make sense in the case of contagious diseases, but the mistaken diagnoses occur even in non-contagious ones such as appendicitis. The same kind of error afflicts stockbrokers who on seeing the market going up recommend their clients to buy, and to sell when it goes down. Statistically, there is little or no connection between rises and falls one day and the next or even one week and the next, but the mere fact that shares have risen prompts people to buy them. The correct strategy is the opposite of that commonly used, namely, to buy at troughs and sell at peaks, though it is not easy to implement. Nor can high-level managers be exempted. They are likely to be influenced by a conversation they had at lunch or by a stray item read in a newspaper rather than using all the evidence at their disposal and even better seeking new information when it is needed.
Statistics are abstract and pallid. For this reason most people ignore them. The knowledge that smoking increases the risk of lung cancer tenfold has little effect. People who give up smoking usually do so only when an isolated dramatic event occurs, for example, if they develop pneumonia and are told by their doctor that it could have been caused by cancer, or if a close friend dies of lung cancer. It might be thought that the reason why smoking has decreased more in doctors than in the general population is that they are intelligent people and that they know the figures on deaths caused by smoking: moreover, they presumably want to set a good example to their patients. A large-scale survey of doctors showed that this picture is considerably idealised. Smoking has dropped most dramatically in those doctors who have been most exposed to the effects of smoking, for example, chest physicians and radiologists. It has dropped considerably less in consultants in other specialities and in general practitioners. Even for doctors the statistics on cigarette smoking do not have the same immediacy as the sight of someone dying as a result of the habit.

It is often said that first impressions are the most important. The saying would appear to conflict with the 'availability error', since that would suggest that what happened last would be uppermost in the mind and therefore most important. Before resolving this paradox, some of the evidence for the importance of first impressions should be examined.
One of the first experiments on the topic was run in the USA by Solomon Asch. He asked subjects to evaluate a person simply on the basis of a list of six adjectives describing him. They might be told that he was 'intelligent, industrious, impulsive, critical, stubborn and envious'. Other subjects were given exactly the same six words but in the opposite order, 'envious, stubborn, critical, impulsive, industrious and intelligent'. All subjects were then asked to fill in a rating sheet in order to evaluate the person. For example, they had to indicate how happy they thought he was, how sociable he was, and so on. The subjects who heard the first list, which began with favourable adjectives, evaluated the person considerably more highly than did those given the list beginning with the derogatory words. This effect -being more heavily influenced by early than by late items -is called the 'primacy error'. There are two possible explanations for it.
First, in the Asch experiment, when subjects heard the first words, perhaps they began to build a mental picture of the person. They then tried to make subsequent words fit in with this picture. A subject who had heard that the person was intelligent and industrious might think of 'impulsive' as meaning spontaneous and regard it as quite a good thing, whereas someone who had first heard the adverse words 'envious' and 'stubborn' might think 'impulsive' meant acting rashly and without thought.
The alternative explanation is that while people are absorbing material, their attention may begin to wander and therefore they are more influenced by the first items they receive than by later ones. An ingenious experiment suggests that this explanation cannot be correct. Subjects watched a stooge attempting to solve thirty anagrams in succession. The stooge, who knew the answers, always solved exactly half the anagrams, but he would deliberately solve many anagrams early on and very few later or he would solve many of the later ones but few of the early ones. Subjects were subsequently asked how many anagrams had been solved by the person they had been watching. When more anagrams had been solved early on, they thought that more had been solved in total than when more had been solved towards the end. This is just another example of the importance of early impressions, but the clever part of the experiment was that the subjects were told to guess after each anagram whether the person they were watching would solve the next one. They must have been paying attention throughout, because their guesses changed as the number of anagrams solved changed: at times when many anagrams were being solved, they tended to think that the next one would be, but when few anagrams were being solved they would guess that the next one would not be. Although they were therefore clearly paying attention throughout, the subjects still thought that more anagrams in total were solved if the soluble ones came mainly at the beginfling than if they came towards the end. Hence, lack of attention to later items is not the cause of the primacy error.
These experiments and many others suggest that beliefs are formed by first impressions: later evidence is interpreted in the light of these beliefs. There is, however, no conflict between the primacy error and the effect of recency on availability. The primacy error occurs because when connected material (such as a newspaper article or lecture) is presented, the interpretation of the later material is coloured by the earlier. The recency effect, on the other hand, occurs when material is not connected; under these circumstances, we tend to be influenced by what we have seen or heard most recently.
The primacy error can be regarded as one form of the availability error: the early items are immediately available in our minds when we encounter the rest. In making a judgement it is not the actual items that matter so much as the meaning we attach to them and that meaning can be altered by the first material we encounter, particularly if it is relevant to the remainder. This error is itself related to another bias in thinking that will be discussed in a later chapter, which will show that for a variety of reasons people cling tenaciously to their existing beliefs and make every effort to avoid discovering that they might be wrong.
The primacy error has important effects in everyday life. If you first encounter someone when he happens to be in a bad mood, you are likely to be prejudiced against him, even if he subsequently behaves more pleasantly. It has been demonstrated that interviewers form an impression of the candidate within the first minute or so and spend the rest of the interview trying to confirm that impression. If you are writing a book, make sure that the beginning is really good: it may be remarked parenthetically that few people ever finish a book so for the most part it would make little difference if the last chapter were mere gobbledegook. If you are writing examination papers, be sure to produce a really good first paragraph. And if you are a doctor diagnosing a patient, make every effort to take as much account of the symptoms you discover last as of those you discover first.

Also related to the availability error is the halo effect. If a person has one salient (available) good trait, his other characteristics are likely to be judged by others as better than they really are. Handsome men and women tend to be rated highly on intelligence, athletic prowess, sense of humour and so on. In fact physical appearance has little to do with such other characteristics: there is a small correlation between being handsome and being intelligent but it is not enough to account for the mistakes people make in their judgements. There is, incidentally, the opposite effect, which is known as the devil effect. The presence in an individual of one salient bad trait, like selfishness, can lower people's opinion of all his other traits: he tends to be seen as more dishonest and less intelligent than he really is. An extreme example occurred when I was serving on a jury, trying a case of rape of a minor. One of my fellow jurymen began the proceedings by remarking about the defendant, 'I don't like the look of him. We should find him guilty.' People influenced by the halo effect are completely unaware that they are biased by it.
One of the most extraordinary consequences of the halo effect occurs in blackjack. If in a casino the dealer's first upward card is an ace, any player may 'insure', that is he may make a side bet of up to half his original stake: if the dealer gets a blackjack, the player receives twice
the value of his side bet, otherwise he loses all of it. A simple calculation shows that (unless he has counted cards) the player will lose on average 7.7 per cent of the money staked as 'insurance'. Yet Willem Wagenaar showed that in a Dutch casino most players sometimes insure and more than 12 per cent always insure. He concludes that the only explanation for this irrational behaviour is that they were deceived by the name 'insurance' into thinking that this was the most prudent course of action.
The halo effect has other pernicious consequences. In one study, the same examination scripts were rewritten twice, once in good handwriting and once in bad handwriting. They were then given to two sets of examiners: each saw all the scripts with half written badly and half well. They were all told to disregard the handwriting and to mark purely on the content. On average the scripts in good handwriting received considerably higher marks than those in poor writing. A similar experiment had an even more horrifying result. When the same essay was shown to examiners bearing a surname and either a male or female Christian name, it received higher marks when the examiner thought it was written by a man.
The halo effect has been put to good (or bad, depending on your point of view) use by the advertising industry for many years. A can of orangeade bearing the name 'Sunblessed' conjures up visions of oranges ripening under a Mediterranean sun and the effect may be enhanced by depicting luscious trees bearing large brightly coloured oranges. And why not throw in a beach for good measure? To the potential buyer the attributes suggested by the name and the picture rub off on the contents of the can and he has every expectation that it will taste delicious whether or not it does. As a matter of fact it is likely to taste better than it really is because he brings to it a set of expectations -juicy, ripe oranges and a holiday atmosphere -which will influence the way it actually tastes. For most products, however, the name and the packaging are irrelevant, except in so far as they signify that the maker has enough sense to choose a good advertising or packaging agency.
Although the halo effect has been known for seventy years, it is remarkable how little notice is taken of it. Only recently have most university examination papers been submitted by number rather than by name, a device that university administrators render worthless, since they usually number the papers in alphabetical order, presumably under the misapprehension that the examiners can't count. One of the most damaging ways in which the halo effect is ignored is the almost universal prevalence of the interview as a means of selection, whether for hospital staff, university students, army officers, the police, civil servants, or whomsoever. I will demonstrate later that the great majority of selection interviews are useless, and may indeed lower the chances of selecting the right candidate. Part of the reason is the halo effect: the interviewers are too influenced by comparatively trivial but salient aspects of the interviewee, which affect their judgement of his or her other characteristics.
One might think that in a scientific subject, the halo effect would not occur. Unfortunately this is not so. When a scientist or more usually several scientists submit a paper to a learned journal, a decision has to be taken on whether to accept it. Normally, the paper is sent off to two or three referees, who are chosen as experts in the usually rather narrow field covered by the paper. The editor decides whether to publish it on the basis of their reports. In 1982, two psychologists published an account of a revealing trick. They selected from each of twelve well-known journals of psychology one published article that had been written by members of one of the ten most prestigious psychology departments in the US, such as
Harvard or Princeton: in consequence, the authors were mostly eminent psychologists. Next, they changed the authors' names to fictitious ones and their affiliations to those of some imaginary university, such as the Tri-Valley Centre for Human Potential. They then went through the articles carefully and whenever they found a passage that might provide a clue to the real authors, they altered it slightly, while leaving the basic contents unchanged. Each article was then typed and submitted under the imaginary names and affiliations to the very same journal that had originally published it.
Of the twelve journals, only three spotted that they had already published the article. This was a grave lapse of memory on the part of the editors and their referees, but then memory is fallible; however, worse was to come. Eight out of the remaining nine articles, all of which had been previously published, were rejected. Moreover, of the sixteen referees and eight editors who looked at these eight papers, every single one stated that the paper they examined did not merit publication. This is surely a startling instance of the availability error. It suggests that in deciding whether an article should be published, referees and editors pay more attention to the authors' names and to the standing of the institution to which they belong than they do to the scientific work reported. You might think that such bias by referees could not occur in a really rigorous subject like physics. But a review of bias, based on 619 articles published in journals of physics, concludes that 'access to publication may sometimes be easier' if you are 'part of the current in-group of well-known physicists' -surely a most delicate way of putting the point.
There are a number of explanations for the lapses over the psychology articles. I shall assume that they should all have been published, but it does not affect the argument; as far as human irrationality goes, the editors had either made a mistake by agreeing to publish originally or they made a mistake by not agreeing to publish subsequently.
The referees and editors could have acted as they did for either or both of two rational reasons. First, the research reported might already have been published by other workers in the two years since the original papers came out. Scrutiny of the referees' reports revealed that this was not the reason: none of them rejected the papers because the findings were not new. Second, it could be argued that workers from a good institution would be more careful in collecting their data and less prone to fraud than those from an unknown institution. This is an implausible reason, if only because some psychologists at good institutions owe their eminence to successful fraud. It is unlikely to have caused the rejection of the papers by unknown authors, since the referees made detailed criticisms of various points in the papers, many of which appear to have been valid. They criticised the statistics used and made remarks like 'The theoretical organisation seems loose and filled with... undocumented conclusions' or 'It is all very confusing.'
The most likely explanation is that both the original acceptance and the subsequent rejection occurred for irrational reasons. The first words a referee or an editor sees on reading an article are the authors' names and the name of their institution. If these are prestigious, he will be biased towards interpreting the paper in the best light possible; if they are not, he is probably going to look for flaws and to be more sensitive to what is wrong than to what is right. Here, then, is a dramatic demonstration of the availability error combined with the primacy and halo effects.
Everyone is irrational some of the time and in particular everyone is susceptible to the availability error. I give a final striking example, this time to do with publishers. In 1969, Jerzy Kosinsky's novel Steps won the American National Book Award for fiction. Eight years later some joker had it retyped and sent the manuscript with no title and under a false name to fourteen major publishers and thirteen literary agents in the US, including Random House, the firm that had originally published it. Of the twenty-seven people to whom it was submitted, not one recognised that it had already been published. Moreover, all twenty-seven rejected it. All it lacked was Jerzy Kosinsky's name to create the halo effect: without the name, it was seen as an indifferent book. Once again, the publishing industry is no more irrational than any other and despite the Kosinsky episode probably does not deserve Colin Haycraft's remark, 'If you can't live, write; if you can't write, be a publisher; if you can't be a publisher, be a literary agent; and if you can't be a literary agent, God help you.'

MORAL

1. Never base a judgement or decision on a single case, no matter how striking.
2. In forming an impression of a person (or object) try to break your judgement down into his (or its) separate qualities without letting any strikingly good or bad qualities influence your opinion about the remainder. This may seem cold, but it is important in situations, such as interviews or medical diagnoses based on a range of symptoms, where the judgement may seriously affect the person being judged.
3. When exposed to connected material, suspend judgement until the end: try to give as much weight to the last item as the first.
4. Try to avoid obtaining information that would bias you: for example, in judging whether an article or a book should be published, remain ignorant of the author's name until you have formed your own opinion of the work.
5. If you happen to be a publisher, check your back list on receipt of an MS: you don't want to publish the same book twice.

© The Estate of Stuart Sutherland. All rights reserved.

go to Irrationality page