IELTS Academic Reading Practice Test 107

For engineers in the middle of the nineteenth century, the streets of a large city presented a problem that seemed to be getting worse. Horse-drawn omnibuses, carts and carriages moved at little more than walking pace, and anything that promised to speed the delivery of messages was certain to attract interest. The electric telegraph could already carry words along a wire, but the physical letter — the invoice, the contract, the printed circular — still had to be taken across town by a boy on foot. Several inventors therefore proposed to send the letters themselves through tubes buried beneath the roads, driven by compressed air.

The principle was simple. A cast-iron pipe, generally between six and twenty centimetres across, was laid between two stations. Inside it travelled a cylindrical container, closed at both ends and fitted with rings of felt or bristle so that it filled the bore of the tube. A steam-driven pump or fan at one end pushed air in behind the container, while at the other end a valve was opened to let out the air in front of it. Because the container ran on a cushion of air rather than on wheels, there was very little friction, and the pressure required was surprisingly modest — a small fraction of the pressure inside a car tyre. Letters were loaded through an airtight hatch, and the operator signalled to the engine house to start the pump.

The engineer most closely associated with the idea in Britain was Josiah Latimer Clark. In 1853 the Electric Telegraph Company, for which he worked, installed a short tube in London to carry telegrams between its central station and the Stock Exchange, and the line worked well. Clark then proposed a much larger network to carry ordinary mail, but the Post Office received the scheme coolly. Officials questioned whether the volume of mail could justify the cost of construction, and the plan was set aside.

In 1859 a group of businessmen formed the London Pneumatic Despatch Company and began to build on their own account. The first section opened in 1863, running from the London and North Western Railway's station at Euston to a nearby district post office, a distance of a little over half a kilometre, and it was intended to carry parcels and mail bags between the railway and the Post Office. The company was confident that the Post Office would adopt the system once its advantages had been demonstrated to officials.

The difficulties soon appeared. Keeping a long iron tube airtight proved far harder than the company had expected: joints worked loose as the ground settled, hatches failed to close properly, and the pumps wasted power. The line was extended, but traffic never reached the levels the promoters had assumed and the company ran short of money. After a few years of intermittent operation the undertaking collapsed, and the Post Office, which had watched the experiment without committing itself, declined to take the system over. The London tube was abandoned, and no pneumatic mail service was ever established in the capital.

Paris succeeded where London failed. The French postal administration opened its first pneumatic line in 1866, and over the following decades the network was extended until it ran beneath much of the city, linking branch post offices with sorting centres and main railway stations. At its height the system consisted of several hundred kilometres of tube, and a letter posted in a suburban office could reach a distant district in a fraction of the time a van required. Generations of Parisians sent their telegrams and express letters through it. The last section remained in service until 1984.

New York produced the most theatrical episode in the history of pneumatic transport. In 1870 Alfred Ely Beach, the publisher of a scientific magazine, opened a tunnel 95 metres long beneath Broadway and invited the public to ride in a single carriage propelled by air. Beach had excavated the tunnel secretly, having failed to obtain permission from the city authorities, and the money and political support needed to extend it never appeared; the ride closed within a few years. Postal tubes proved more durable. A network connecting post offices and railway termini opened in 1897, and the last of it continued to carry mail until 1953.

Elsewhere in Europe, pneumatic post became a routine part of city life. Berlin opened a network in 1876 which eventually reached most districts of the city, and Vienna and Prague followed. Prague's system, opened in 1889, was still working a century later. These networks carried telegrams, which had to reach a telegraph office quickly, as well as the small packets of documents that businesses sent in large numbers, often several times a day.

Cheap telephones, motor vans and eventually electronic mail removed most of the reasons for these systems. Prague's tubes were damaged by floods in 2002 and have not been restored, and the Paris network dwindled to a few lines before it closed. Yet the principle has never disappeared. Pneumatic tubes still carry blood samples and prescriptions in hospitals, cash from supermarket tills to a secure room, and documents between the counters of banks. Visitors to some drive-in banks still send their payments to the cashier through a tube — an everyday survival of a technology that Victorian engineers believed would transform the post.

Every driver constantly performs a task that turns out to be remarkably difficult to reproduce in a machine: reading the road. The human eye and brain detect a child stepping off a kerb, a cyclist wobbling in the rain and a temporary sign half-hidden by a lorry, and they do so without apparent effort. Engineers describe vehicle automation in six levels, numbered 0 to 5, from a car that merely warns the driver to one that needs no human being at all; at the middle levels the vehicle handles some of the driving while a person remains responsible for watching the road. Whatever the level, a car that is to take any decision for itself must first build a picture of its surroundings, and that is the work of its sensors.

The cheapest and most familiar sensor is the camera. A modern car may carry half a dozen of them, behind the windscreen, in the bumpers and on the boot lid, and each provides a dense, colourful image of the scene at high resolution. Because they record something close to what a human would see, cameras are particularly good at reading road signs, distinguishing traffic lights and following lane markings. Their weakness is that the image must be interpreted by software, and that interpretation is difficult: a shadow, a puddle or a poster of a road can look alarmingly like something solid. Low sun directly ahead, heavy rain or a dirty lens can spoil the picture, and a camera on its own gives no direct measurement of distance.

Radar avoids several of these problems. A transmitter sends out radio waves and measures the reflections, and because radio waves pass through fog, dust and darkness, radar keeps working when a camera cannot. It also measures speed directly: the change in frequency of the returning wave, known as the Doppler effect, reveals how fast an object is approaching or receding. Radar has been used for years in adaptive cruise control, which adjusts a car's speed to the vehicle in front. What it lacks is detail. The reflected signal gives the range and the movement of an object but says little about its shape, so radar may struggle to distinguish a parked van from a bridge support.

Lidar — light detection and ranging — sits between the two. It fires rapid pulses of laser light and times their return, calculating the distance to each point that reflects them. By firing in many directions, a lidar unit builds up a point cloud: a three-dimensional map of the surfaces around the car, accurate to a few centimetres. Early units were bulky and very expensive, and many of them were mounted on the roof, where they rotated to sweep the road in every direction. They typically use invisible infrared light at wavelengths of about 905 or 1,550 nanometres. Solid-state designs, which steer the beam electronically and contain no moving parts, have cut both the size and the price, although a lidar unit is still far costlier than a camera.

Ultrasonic sensors complete the set. They emit sound waves far above the range of human hearing and time the echo, and they are cheap enough to be fitted all round a bumper. Their range is only a few metres, which makes them ideal for parking and for detecting a bollard or a child close to the car, but they are of no use at speed.

None of these devices is sufficient without a way of knowing where the car is. Satellite navigation, more precisely known as a global navigation satellite system, provides an absolute position, and detailed digital maps record the line of lanes, kerbs and traffic signals to within a few centimetres. The weakness of satellite positioning is that its signals are faint. Between tall buildings or beneath dense trees, and in tunnels, the receiver may lose its fix altogether. Engineers therefore add an inertial measurement unit, which senses acceleration and rotation many times a second and can carry the estimate forward while satellite signals are missing, together with readings from the wheels.

The final stage is to combine all this information, a process known as sensor fusion. A statistical technique called the Kalman filter, developed in the 1960s for spacecraft navigation, is widely used: it maintains an estimate of where objects are and how fast they are moving, and updates that estimate as each new measurement arrives, giving more weight to the sensors that are currently most trustworthy. The point of fusion is redundancy. A camera may be blinded by the sun, a lidar return may be lost in fog, a radar may mistake a metal drain cover for an obstacle, but the three together rarely fail in the same way at the same moment.

What remains is the long tail of unusual events. A system that has seen a million ordinary road scenes may still be baffled by a sofa falling from a lorry, a horse-drawn carriage, a traffic officer waving cars past a broken light, or a road resurfaced in a way the map does not expect. Engineers call these edge cases, and there is no way to collect them all in advance; they must be found by driving, by simulation and by replaying recorded incidents. The result is that the sensing hardware is, in many respects, already good enough, while the software that makes sense of what it reports is still the limiting factor. The problem is not seeing the road, but understanding it.

A woman opens a refrigerator, takes out a bottle of beer and sets it on the counter with the label facing the camera. Nobody in the audience is surprised, and most people will not consciously notice. Product placement — the paid insertion of branded goods into films and television programmes — is by now so common that it is easier to name the productions that avoid it than those that use it. Nor is it new. Branded goods appeared in films almost from the beginning, and the practice was well established long before anybody thought to study it. What has changed is the money involved and the sophistication with which the deals are made.

The moment the industry likes to date its conversion to is 1982, when a film about a boy who befriends a stranded alien used a particular brand of peanut-butter sweets to lure the creature out of hiding. A larger confectionery company had reportedly turned the offer down, and watched as sales of the rival product rose sharply. The episode became a lesson: a single scene could move more goods than a television advertisement, and studios began to employ specialists to sell such scenes in advance. Within a decade, placement had become a routine item on the budget of a studio film.

The economics are straightforward. A film with a large budget must recover its costs, and a placement fee, or the free supply of cars, computers and clothing, reduces the sum that has to be found. For a low-budget production the loan of a vehicle or the use of a location can be the difference between shooting and not shooting. Studios defend the practice on these grounds, and point out that the alternative — a cast dressed in anonymous clothing and driving unmarked cars — is not obviously more honest.

In some films the result is genuinely convincing. A character who opens a kitchen cupboard or drives down a motorway will encounter branded objects, and a set that has been stripped of every logo can look oddly artificial, as though the action were taking place in a world where manufacture had ceased. When a brand is chosen because it suits the character — the cheap lager drunk by a student, the expensive watch worn by a financier — the audience is being told something true about the world.

The trouble is that the fit is rarely the point. The brand that pays is the brand that appears, whether or not it suits the part, and writers are asked to build a scene around a product rather than a product around a scene. Audiences are quick to detect this. Research on placement has generally found high levels of recall for prominent examples but much weaker evidence that they change what people buy, and there is a clear difference between noticing a brand and being persuaded by it. What the research does show consistently is irritation: when a placement is obvious, viewers report that it broke their attention and lowered their opinion of the film.

Rules differ from country to country. In Britain, product placement in television programmes made by broadcasters was prohibited for decades, but the rules were relaxed in 2011 to allow paid placement on commercial television, subject to restrictions: it remains forbidden in children's programmes, news and current affairs, and programmes about religion. In the United States, television broadcasters must tell viewers when a payment has been made. Cinema, by contrast, is barely regulated at all, on the assumption that a paying audience knows what it is watching.

The exception, and the strongest argument for regulation, concerns children. Advertising aimed at young viewers is restricted in many countries for a good reason: children are less able to recognise persuasion and to defend themselves against it. A placement, however, does not look like advertising. It arrives inside the story, in the company of characters the child likes, and no announcement interrupts it. A cartoon in which the hero eats a particular cereal is not obviously different from one in which he eats a bowl of cereal, and that is exactly the problem.

Technology may make the question harder still. A film can now be altered after it has been released, and in principle a product visible on a table could be changed for different markets or different years, so that the version a viewer sees in five years' time contains a brand that did not exist when the film was shot. Whether audiences would accept such substitutions is another matter, and no studio has yet made a habit of it.

None of this argues for a ban. Placement is a legitimate way of paying for expensive work, and at its best it adds a texture that artificial sets lack. What it needs is transparency. A viewer who knows that a brand has paid to be there can decide what to do with that knowledge, and a parent can at least choose what a child watches. The case for regulation, in my view, rests not on the sinfulness of commerce but on a simple principle: persuasion should be visible as persuasion.