IELTS Academic Reading Practice Test 178

The first coin-operated amusement machines appeared in the late nineteenth century. Early devices measured the strength of a player's grip or delivered a printed card predicting the future; others showed a short sequence of flickering photographs for a penny. Entrepreneurs in American cities gathered dozens of such machines into what became known as penny arcades, often attached to saloons, amusement parks and railway stations. The business model was established then and has barely changed: the customer pays a small sum, receives a brief experience, and is invited to pay again.

Pinball, the machine that dominated the arcade for decades, grew out of bagatelle, a table game popular in France from the eighteenth century in which a player used a cue to propel balls into numbered pockets. American manufacturers shrank it, added a spring plunger and a sloping table, and by the early 1930s had produced countertop versions that paid out in coins or free plays. The first to sell in large numbers was Baffle Ball, launched in Chicago in 1931 by David Gottlieb, which cost one cent to play. Within a year or two rival firms, including the company that became Bally, had introduced machines standing on legs.

Electricity transformed the machines. It powered lamps, electromagnetic coils that kicked the ball back into play and ringing bells, and it allowed scores to be registered and displayed automatically. It also made the games attractive to gamblers, and during the 1930s many American cities restricted or banned them. The best-known prohibition was imposed in New York in 1942 by Mayor Fiorello La Guardia, who held a public event at which seized machines were smashed with sledgehammers and tipped into the harbour. The ban remained in force for more than three decades.

The feature that turned pinball into a game of skill arrived in 1947, when a designer named Harry Mabs added flippers to a machine called Humpty Dumpty, built by Gottlieb. Players could now keep a ball alive, aim it and learn to control it. Flippers also gave the industry an argument when New York reconsidered its ban. In 1976 the city council agreed to hear evidence, and a skilled player demonstrated to the assembled officials that a good player could keep a ball in play for a long time. The ban was lifted.

Meanwhile the electronics that would eventually displace mechanical games were becoming cheap enough to use. In 1971 a small Californian firm called Nutting Associates released Computer Space, a coin-operated video game, but it was too complicated for most players. The following year two of its designers, Nolan Bushnell and Ted Dabney, founded Atari and produced Pong, a simplified game of table tennis in which two white rectangles knocked a square back and forth. A prototype placed in a bar proved so popular that its coin box had to be emptied more often than expected, and the commercial version sold in large numbers.

The arcade industry expanded rapidly after the release of Space Invaders in 1978. Designed by Tomohiro Nishikado for the Japanese company Taito, it gave players a fleet of aliens to destroy, and it became so popular in Japan that a story circulated about a shortage of the hundred-yen coins needed to play it. Pac-Man, created for Namco by Toru Iwatani in 1980, attracted players who had little interest in shooting, and colour monitors, microprocessors and dedicated sound chips allowed designers to produce games of growing sophistication. By the early 1980s arcades stood on many city streets, and the period is often called the golden age of the video game.

The boom ended abruptly. In 1983 the North American market collapsed, overwhelmed by too many poorly made games competing for the same players, and many manufacturers closed. Home consoles, which allowed people to play similar games in their own living rooms, took over much of the audience during the following decade, and in Britain and the United States the number of arcades fell sharply. Japan was the exception: large, multi-storey amusement centres continued to flourish in its cities, and Japanese firms developed games, such as the rhythm and dancing machines of the late 1990s, that depended on equipment only an arcade could provide.

Arcades survive today in several distinct forms. Family entertainment centres combine rides, bowling and redemption games, in which players win tickets that can be exchanged for small prizes; the restaurant chain Chuck E. Cheese, founded in 1977 by Nolan Bushnell, was an early example of the formula. Elsewhere, bars equipped with older machines attract adults who remember the games from childhood. Claw machines, which reward persistence rather than skill, remain among the most profitable machines in any arcade.

Preservation has become a problem. The cathode-ray tubes, custom circuit boards and controls of 1980s machines are difficult to repair, and few manufacturers hold spares. Collectors restore original cabinets, while software projects such as MAME, begun in 1997, have catalogued and emulated thousands of titles so that they can still be run on modern computers. Museums in several countries now hold arcade collections, treating the machines as industrial design rather than disposable entertainment. For a form of entertainment designed to be forgotten once the coins ran out, that is an unexpected afterlife.

Few sights are as familiar to the modern driver as a queue at traffic lights that appear to be changing for nobody. In most cities the timing of those lights was worked out years ago, using counts of vehicles made at a handful of peak periods, and it has not been revised since. The cost is not merely irritation. Stopping and starting wastes fuel, adds to emissions and accounts for a large share of the delay that vehicles experience in urban areas. Artificial intelligence is now being applied to the problem, not by replacing the lights but by changing the way their timing is decided.

A conventional junction is controlled by a signal plan. Engineers divide traffic into stages, decide how long each stage should last within a repeating cycle, and choose how much green time each approach receives. Where junctions lie close together they also choose an offset, the difference in timing between neighbours, so that a vehicle travelling at the intended speed meets a succession of green lights. Together these choices amount to a timetable for the junction. Plans are calculated in advance from surveys, and the simplest equipment runs them to the clock, whatever is happening on the road.

The first improvement was to let a junction respond to what it can detect. Loops of wire buried in the road surface register the presence of a metal vehicle, so a signal can extend a green light while traffic is still arriving, or skip a stage when nobody is waiting. Adaptive systems go much further. SCOOT, developed at Britain's Transport Research Laboratory, and SCATS, developed by the traffic authority of New South Wales, both use continuous measurements to adjust the green splits, the cycle length and the offsets every few seconds. The result is a control loop rather than a timetable: the system predicts where vehicles will be a few moments later and alters its plan accordingly.

The newest systems add machine learning. Cameras on poles feed images to computer-vision models that count vehicles, estimate their speed and classify them as cars, buses, lorries or bicycles. The counts become the input to a model which has learned, from months of examples, how traffic in that particular city responds to a change in timing. Some research groups go further and train the controller itself, using a technique called reinforcement learning, in which a program tries millions of sequences of decisions in a simulated city and is rewarded whenever it reduces total waiting time.

The quality of any such system depends on its data. Inductive loops are cheap and reliable but are damaged by roadworks and say nothing about the size of a vehicle. Radar measures speed directly; cameras see almost everything but struggle in fog, low sun or heavy rain. Some cities match the identifiers that phones broadcast over Wi-Fi or Bluetooth, or track the satellite-positioning units fitted to buses and taxis, to follow individual journeys across a network. Each source has gaps, and much of the engineering lies in combining them.

Adaptive control is now used in hundreds of cities. Its most visible application is priority for particular vehicles: a bus running late can be given a longer green light, and an ambulance approaching a junction can turn the signals red in its path. Similar logic governs the ramps that admit vehicles to fast motorways, where the rate of entry is adjusted to keep the main carriageway moving, and the variable speed limits displayed above busy stretches of road. The aim in each case is to prevent a small disturbance from becoming a long queue.

AI traffic management also has limits. A model is only as good as the data used to train it, and traffic is full of events it has rarely seen: a burst water main, a stadium emptying, a snowstorm. A controller tuned in a simulator may behave unpredictably when it meets the real world, which is why almost every deployed system keeps a conventional fallback and a human engineer who can override it. Objectives are a further difficulty. A system told to minimise vehicle delay will lengthen the time pedestrians must wait, and an efficient network, by making driving easier, can attract more cars onto the roads it was meant to relieve.

Cameras and phone scanning raise questions that engineers cannot answer alone. Number plates and device identifiers can be personal data, and some jurisdictions limit how long they may be kept. A signal tuned by a proprietary algorithm may be impossible for a city to audit, which matters when a decision to hold one approach at red affects the safety of everyone crossing. Contracts increasingly require suppliers to explain the logic of their controllers rather than merely to demonstrate that they work.

Research now looks beyond the individual junction. Connected vehicles can tell the infrastructure where they are and what they intend to do, allowing a controller to plan a few seconds ahead with far more certainty than a camera permits. Trials of cooperative merging, in which vehicles adjust their speed to join a motorway without forcing others to brake, suggest that small changes in behaviour can produce large improvements in flow. Whether such systems will work when only a fraction of vehicles are connected, and drivers of the rest behave as they always have, remains the central question.

In Britain it has been illegal to hold a phone while driving since 2003, and the rule was tightened in 2022 to cover almost any use of a handheld device, including taking a photograph or scrolling through a playlist. Using a phone hands-free, however, remains lawful, whether the driver is speaking into a dashboard system, a wireless earpiece or a unit clipped to the sun visor. The distinction rests on an assumption that is rarely stated and seldom examined: that the danger of a phone lies in the hand that holds it, and that once both hands are on the wheel the driver is safe. A large body of research suggests that this assumption is wrong.

Psychologists have been testing the question since the 1990s, mostly in driving simulators. In experiments at the University of Utah, David Strayer and his colleagues asked volunteers to drive a simulated route while holding a conversation, sometimes on a handheld phone and sometimes on a hands-free one. Both groups missed more traffic signals, braked later and drifted from their lane more than drivers who were not talking. The impairment was, in most measures, much the same whether or not the phone was held. The researchers concluded that the cost of the conversation came not from the hands but from the attention it consumed.

Other work has compared phone use with alcohol. In one often-cited simulator study, drivers who were talking on a phone were more likely to collide than drivers whose blood alcohol had been raised to the legal limit for driving, although the drunk drivers were more erratic in their control of the car. Comparisons of this kind should be treated cautiously, since a simulator is not a road and a short experiment is not a lifetime of driving. Yet the direction of the finding has been repeated often enough to be difficult to dismiss.

Why should a conversation be so demanding? A passenger can see the road. When traffic thickens or a child steps off the pavement, the passenger stops talking, and the driver's silence is understood. Someone at the other end of a telephone connection has no idea what is happening in the car, and keeps talking. There is a further burden that has been studied in its own right: a driver listening to a description of a journey or a meeting tends to form a mental picture of it, and the brain areas used for that imagining are the ones needed to watch the road. In one well-known experiment, people walking across a campus while talking on a phone were far less likely than others to notice a clown on a unicycle passing nearby.

Against all this, three arguments are usually made. The first is that hands-free equipment at least keeps the driver's hands where they should be, which must be better than fumbling with a small screen at seventy miles an hour. The second is that a conversation is not obviously more dangerous than the many other things drivers do: listening to the radio, arguing with a child in the back seat, or simply thinking about the day ahead. The third is practical. A police officer standing on a pavement cannot see whether the driver of a passing car is speaking to a passenger in the vehicle or to a client in another city, and a prohibition that cannot be enforced may bring the law itself into disrepute.

Voice-activated systems complicate the picture further. Manufacturers present them as a safer way to send a message or change a destination, because the driver's eyes and hands stay where they belong. Studies of the mental effort they require tell a different story. When a voice assistant mishears a name or a number, the driver repeats it, corrects it and becomes irritated, and each of these responses takes attention from the road. Research by the AAA Foundation for Traffic Safety in the United States, using a scale that rates the demands of different tasks, found that some speech-based systems placed a heavier load on drivers than a conversation on a phone.

Simulator findings would matter little if the effect vanished on real roads. It does not appear to. Naturalistic studies, which fit cars with cameras and sensors and observe drivers going about their ordinary lives for months, have consistently linked phone use to a markedly higher risk of crashing, and the risk is not confined to the seconds in which a number is dialled. What such studies cannot easily do is separate the conversation from the kind of driver who chooses to have one at the wheel.

There is also a question of consistency. If the objection to a phone is that it occupies the mind, then the law is oddly selective. It permits a driver to listen to a heated radio debate or to supervise a quarrel on the back seat, and it does nothing about the daydreaming that occupies far more drivers than any device. Critics of restriction argue, reasonably, that a state which polices attention will struggle to know where to stop. The answer, I think, is that the law already penalises drivers who are too tired or too ill to concentrate, even though neither condition can be seen from the pavement.

My conclusion is that the present rule is drawn in the wrong place. Keeping the ban on handheld devices, the law should also restrict the conversations that make hands-free use attractive, while leaving alone the brief, one-way exchanges, such as a navigation instruction or a warning about traffic, that impose little load and can be ended at once. Employers whose staff drive for a living could do much of the work through company policies, and insurers could reward those who switch the phone off. None of this is straightforward. But the law should stop implying that a conversation is safe merely because the driver's hands are free.