IELTS Academic Reading Practice Test 177

By the late 1960s the television set had become the most familiar piece of electronic equipment in the home, and a few engineers began to wonder whether it could do more than receive broadcasts. Computers were still room-sized machines owned by governments and large companies, and the idea that an ordinary family might play games on one seemed absurd. The first practical steps were taken not by a computer company but by a defence contractor, Sanders Associates of New Hampshire, where an engineer named Ralph Baer was allowed to spend a little time on an unusual project.

Baer's team built several prototypes, the last of which was completed in 1968. It was a box of switches and circuits that generated simple shapes on an ordinary television, and from its wooden casing it became known as the Brown Box. Sanders licensed the design to Magnavox, a manufacturer of televisions, and in 1972 the Magnavox Odyssey went on sale as the first video game console offered to the public. It showed black-and-white graphics, produced no sound at all, and its games were chosen by plugging in cards. Sales were modest, but the machine established an idea that would shape the industry: that a console belonged in the living room.

In the same year Nolan Bushnell founded a company called Atari and released Pong, an arcade machine whose two paddles and bouncing ball were understood at once by anyone who saw them. Arcades multiplied across North America. In 1975 Atari turned Pong into a home console and sold it through the department store chain Sears. The machine could do only one thing, but it sold in large numbers and proved that there was a mass market for games played at home.

The next step was to build a console that could play several different games. The Fairchild Channel F, released in 1976, was the first machine to accept interchangeable cartridges, each holding a program on a memory chip; inserting a different cartridge changed the game. Atari followed in 1977 with its Video Computer System, later renamed the Atari 2600, which contained a microprocessor and produced colour pictures and sound. The cartridge was as important commercially as technically. A console could now be sold cheaply, while the games, bought one at a time, generated the profit.

Then the industry collapsed. Too many companies launched consoles, each hoping to repeat Atari's success, and too many poor-quality games appeared. In 1982 Atari released a version of its most popular arcade game, Pac-Man, which players found disappointing, and a game based on the film E.T., which had been made in a hurry and was widely regarded as almost unplayable. Retailers dumped unsold stock at low prices. Revenue in North America fell from about three billion dollars a year to a hundred million, and by 1985 production of consoles in the United States had almost stopped.

The recovery came from Japan. Nintendo, a company founded in the nineteenth century to make playing cards, released its Family Computer, or Famicom, in Japan in 1983. It was an eight-bit machine, cheap and reliable, and sold rapidly. When a version reached North America in 1985 as the Nintendo Entertainment System, Nintendo imposed a strict licensing system: outside developers had to ask permission to publish games, were limited in how many titles they released each year, and had to accept the company's Seal of Quality. The rules annoyed publishers but prevented a repeat of the glut of 1983. The console eventually sold more than sixty million units.

Sega, another Japanese company, challenged Nintendo from 1989 with its Genesis, sold in most of the world as the Mega Drive. The Genesis was a sixteen-bit machine, technically more advanced than the ageing Nintendo console, and an aggressive marketing campaign presented it as the choice of older and cooler players. Nintendo replied in the early 1990s with the Super Nintendo. For several years the two companies fought for the same customers, a contest that filled magazines and playgrounds alike.

The next shift came from a different direction. Sony, an electronics manufacturer, had been working with Nintendo on a machine that would read games from compact discs, but the partnership broke down and Sony built the machine itself. The PlayStation, released in Japan in 1994 and in North America in 1995, stored its games on CD-ROM. Discs were much cheaper to manufacture than cartridges, so games could be produced at a fraction of the old cost. The machine also produced three-dimensional graphics that had been seen until then only in arcades, and it sold more than a hundred million units.

Sony's follow-up, the PlayStation 2, arrived in 2000 and could play DVD films as well as games, which gave families a reason to buy one. It became the best-selling console in history, with sales of more than 155 million. Microsoft, a software company, entered the market in 2001 with the Xbox, which contained the first built-in hard drive on a console and a port for a broadband connection. A year later the company launched an online service, Xbox Live, that allowed players to compete over the internet rather than sitting beside one another.

In 2006 Nintendo took a different route again. The Wii, instead of competing on processing power, used a controller that detected the motion of the player's hand, and its simple games attracted people who had never regarded themselves as players. The other manufacturers concentrated on high-definition pictures and on services that delivered games and films over the internet, turning the console into a general-purpose box beneath the television.

Almost every surface outdoors is a surface that collects dirt. Windows, walls, road signs and solar panels gather soot, dust and the greasy particles that cities produce, and removing them costs money, water and, in tall buildings, a good deal of danger. For several decades researchers have therefore tried to design materials that shed dirt with little or no help. Two strategies dominate. One makes the surface so difficult for water to wet that rain cannot cling to it; the other coats it with a substance that attacks the dirt chemically.

The first strategy takes its inspiration from a plant. The leaves of the lotus, Nelumbo nucifera, an aquatic species found across Asia, stay conspicuously clean even when they grow in muddy water. In the 1990s the botanist Wilhelm Barthlott and his colleagues at the University of Bonn examined the leaf with electron microscopes and found that it is covered not with a smooth waxy skin but with a dense forest of microscopic bumps, each of which is in turn coated with crystals of wax only nanometres across.

The key measurement is the contact angle, the angle at which the edge of a droplet meets a solid. On a surface that water likes, the angle is below 90 degrees and the liquid spreads out; on a highly unwettable surface the droplet sits up like a ball, and an angle above 150 degrees is described as superhydrophobic. Wax alone is not enough. A smooth waxed surface may give an angle well above 90 degrees, yet a drop will still stick when the surface is tilted, because the liquid is in contact with the whole area beneath it. Roughness changes this. When the bumps are close together, a droplet rests only on their tops, with air trapped in the spaces below. The contact area becomes tiny, the droplet is almost spherical, and it rolls away at the slightest tilt, sweeping loose particles with it.

Chemistry offers a second route, and it is the one used in the self-cleaning glass sold since the early 2000s. A transparent film of titanium dioxide, no more than nanometres thick, is applied to the glass during manufacture. Titanium dioxide is a photocatalyst: when ultraviolet light, a component of ordinary daylight, strikes it, the energy is absorbed and, in the presence of oxygen and water, highly reactive fragments are produced that break the molecules of organic dirt into smaller pieces. The same light makes the coating hydrophilic, so rain spreads over the glass in a sheet instead of standing in droplets and carries the fragments away.

Self-cleaning and 'easy-clean' products now include windows for tall buildings, where access for cleaning is costly or dangerous; roof tiles and paints that stay brighter for longer; fabrics treated to resist stains; and ceramic sanitaryware. Solar panels are an important target, since a layer of dust can cut the output of a panel substantially, and in dry regions washing them with water is difficult and expensive.

The limitations, however, are considerable. The lotus effect depends on structures measured in nanometres, and such structures are easily damaged by abrasion, so a surface may lose its properties after repeated wiping. Oils and detergents have a much lower surface tension than water, spread readily over most hydrophobic coatings and leave a film behind. Photocatalytic coatings need light, which puts indoor surfaces at a disadvantage. The technology is also not cheap: a coated window may cost substantially more than an ordinary one, and the savings in cleaning must be weighed against that premium, which for many buildings is never recovered.

The search is now for surfaces that repel oils as well as water. In 2011 a team at Harvard University led by Joanna Aizenberg described a different approach, inspired by the slippery interior of the pitcher plant, a carnivorous species whose trap gives insects no foothold. Instead of trapping air, the researchers made a porous solid and filled its pores with a lubricating liquid held in place by capillary forces. Liquids of almost any kind float on this layer and slide off. Because the lubricant can flow back into scratches, the surface is to some degree self-healing, and tests have shown it resisting ice, condensation and the attachment of bacteria.

Progress is nevertheless slower than advertising suggests. Nobody has yet produced a coating that survives years of weathering and scrubbing without change, and there is no standard way of comparing claims: a surface that withstands a thousand wipes in one laboratory may fail after a hundred in another. Testing has to include abrasion, ultraviolet exposure and repeated wetting and drying, and results are rarely published. The phrase 'easy-clean' is therefore more honest than 'self-cleaning' for most products on the market, since even a good coating needs occasional washing.

None of this makes the research less valuable. The prize is a surface that needs no detergent, little water and no labour, and even a coating that must be renewed every few years would save both money and energy. The lotus, which has had a very long time to perfect the trick, remains a useful teacher.

Police forces have always tried to guess where crime will happen next. Patrols were once plotted on paper maps with pins, and experienced officers carried a mental map of the streets worth watching. What is new is the quantity of data now available and the sophistication of the models applied to it. Predictive policing, as the practice is known, uses records of past offences, together with information about the places where they occurred, to forecast where and when future offences are likely, or which individuals are most likely to be involved in them. The promise is attractive to forces whose budgets are shrinking while the demands on them grow.

Two families of method are used, and the distinction between them matters more than any other in the debate. The first looks at places: it identifies small areas where offences cluster, or streets where a burglary is statistically likely to be followed by another within weeks, and directs patrols there. The second looks at people: it produces a score for individuals, ranking them by their supposed likelihood of being shot or of shooting someone. Both are called predictive policing, yet they raise quite different questions.

The place-based approach grew out of crime mapping. In New York in the 1990s a system called CompStat required each local commander to appear at meetings with maps and figures and to account for the crime in their area; the practice spread to other cities. From about 2010, software began to replace the analyst's pins. One of the best-known systems, used by the Los Angeles police from 2011, was derived from a model of earthquake aftershocks: just as tremors follow a large shock, so certain crimes were said to arrive in clusters that could be predicted from the pattern of earlier ones.

The evidence is more complicated than the marketing. Hot-spot policing, in which officers are sent to small areas where offences are concentrated, has been tested in randomised trials and systematic reviews, and the general finding is that it produces modest reductions in crime. That is not the same as validating the software, since a map of hot spots can be drawn by an analyst with pins and local knowledge. A review published by the RAND Corporation in 2013 found few rigorous evaluations of the commercial systems themselves and warned that the data feeding them were often of poor quality. Confident claims about the size of the reductions that follow a new system should therefore be treated with caution.

The deeper difficulty is that police records are records of policing, not of crime. If officers stop and search more people in one neighbourhood, more offences will be discovered there, and the model will conclude that the neighbourhood needs more officers, who will in turn discover more offences. Researchers, including Rashida Richardson and her colleagues, have described how data produced by unlawful or discriminatory policing can contaminate a prediction, so that the system reproduces the behaviour of the officers who generated the figures. The past is not a neutral guide to the future; it is a record of decisions that were themselves disputed.

Individual risk scores are more troubling still. A list of people judged likely to be involved in violence is built from records of police contact — stops, warnings, arrests, the company a person keeps — rather than from proved offences. Being named by an officer therefore raises a score, and a higher score is treated as a reason for further attention, which may produce another entry. The score becomes evidence of the very thing it was meant to predict. Whatever the intention behind such lists, they also sit awkwardly with the principle that a person should be treated as innocent until an offence is proved.

There is a further problem of accountability. Many systems are sold by private companies that decline to reveal how their models work, and a public body that cannot explain why a street was watched more heavily, or why a young man was visited at home, is in a weak position to defend the decision. In my view, any system used to allocate police resources should be published in outline, tested by independent researchers and open to challenge by the people it affects.

None of this means that the enterprise is doomed. Forecasting where offences concentrate is an extension of what good police work has always done, and there is no reason to reject a well-tested model simply because it is mathematical. The conditions for acceptable use seem to me clear enough: the tool should address places rather than people, the data should be audited, the results should be published, and officers should remain responsible for their own decisions. Several cities have abandoned or restricted these systems in recent years, some because the results disappointed them and others because residents objected.

The argument, then, is not really about mathematics at all. A statistical model can no more be neutral than a police officer can, because both are shaped by the choices of those who built and employ them. What the debate has exposed is how much authority a number acquires once it is printed on a screen, and how rarely that authority is examined. Predictive policing will do little harm, and some good, if it is treated as a map that might be wrong; the danger lies in the authority it lends to familiar assumptions.