IELTS Academic Reading Practice Test 189

By the middle of the nineteenth century the streets of central London were crowded with horse-drawn omnibuses, carriages and pedestrians, and the railway had already taught engineers how to control moving vehicles. Railway signalling relied on semaphore arms and coloured lamps, which told drivers when to proceed and when to stop. Applying the same principles to the road was an obvious next step, although it took several decades to complete.

The first attempt was made in 1868. John Peake Knight, a railway manager who had designed signalling arrangements for the South Eastern Railway, proposed a device for the junction of Bridge Street and Great George Street, close to the Houses of Parliament. Installed on 9 December 1868, it consisted of a pillar carrying two semaphore arms, which were raised horizontally to mean stop and lowered to mean proceed. At night the arms were replaced by red and green gas lamps, the colours already established on the railways as signals of danger and safety. A police constable operated the device by hand.

Knight's signal did not last long. Within weeks a leak in the gas supply caused an explosion which injured the constable working the equipment, and the device was dismantled. The idea then lay dormant for more than forty years. Traffic was slow, and outside a few large cities light enough to be managed by police officers on foot; nor did many municipalities have the money or the administrative capacity to install and maintain such machinery.

The electric traffic signal appeared only when motor vehicles began to fill the streets. In 1912 Lester Wire, a police officer in Salt Lake City, Utah, built a device in a diamond-shaped wooden housing with red and green bulbs, which could be switched from a small booth at the side of the road. Wire's invention was not manufactured in quantity, but it influenced the systems that followed. Two years later the American Traffic Signal Company installed a signal designed by James Hoge at the corner of Euclid Avenue and East 105th Street in Cleveland, Ohio.

The Cleveland installation began operating in August 1914. It displayed red and green lights, and it added a feature that has since disappeared: a buzzer, sounded by the officer in charge, which warned drivers that the lights were about to change. The signal was connected to the city's fire stations so that it could be held on red while fire engines passed. An officer watching the traffic from a booth decided when to change the lights. Similar equipment was soon installed in other American cities.

The three-colour signal, with the amber light that drivers now take for granted, is generally credited to William Potts, a Detroit police officer who designed a system controlling four approaches at once. Potts used amber to give drivers a warning interval between red and green, and his design was installed in Detroit in 1920. The first version was mounted on a tower in the middle of the junction, with an officer inside it.

Patents followed. In 1923 Garrett Morgan, an inventor in Cleveland, obtained a patent for a T-shaped signal with three positions, the third of which stopped traffic in every direction and allowed pedestrians to cross. The rights to the patent were later sold to the General Electric Company. Morgan's device was never widely manufactured, but the idea of a phase in which all vehicles halt survives in the pedestrian crossings of many cities.

Britain came late to automatic operation. The first automatic signal in the country was installed in Wolverhampton in 1927, and others followed during the 1930s. By then the fixed-time cycle was already being challenged by equipment that detected approaching vehicles, using a pneumatic tube laid across the road, and adjusted the lights in response.

Congestion, rather than safety, drove the next stage. Engineers noticed that a series of signals set to a common cycle allowed vehicles to travel through a town without stopping, a technique known as the green wave, and by the 1950s many city centres were controlled in this way. In 1963 Toronto became the first city to place its signals under the control of a computer, which adjusted the timing of individual junctions according to the traffic it detected.

Computer control made possible the adaptive systems in use today. Developed at Britain's Transport Research Laboratory in the 1980s, SCOOT — Split Cycle Offset Optimisation Technique — collects data from detectors buried in the road, models the queues forming at each junction and adjusts green times continuously. Comparable systems operate in Australia and elsewhere. Modern signals also use light-emitting diodes, which replaced incandescent bulbs from the 1990s onwards: LEDs are brighter in sunlight, last far longer and consume a fraction of the electricity.

Colour, meanwhile, remains a problem for some road users. Red and green are precisely the hues that people with the commonest form of colour vision deficiency find hardest to distinguish, and roughly one man in twelve is affected to some degree. Several countries now supplement the lights with a position code, red at the top and green at the bottom, or with distinct shapes, so that the signal can be read without relying on colour alone.

Most countries measure air quality with instruments known as reference monitors. These sit in climate-controlled shelters, cost tens of thousands of pounds each and are operated by trained technicians who calibrate them at regular intervals. The figures they produce are trusted by regulators, courts and doctors. In Europe and North America the law requires that the air people breathe is measured to a defined standard, and only instruments that meet it may be used for the purpose. The difficulty is that there are so few of them: a large city may have half a dozen official stations, so that a map of its air is drawn from a handful of points, even though pollution can vary enormously from one street to the next.

Low-cost sensors, which began to appear in large numbers around 2010, promise to fill in the gaps. The commonest type is the optical particle counter. A small fan draws air through a chamber, where a beam of light — usually from a laser, sometimes from a light-emitting diode — crosses the airflow. Particles passing through the beam scatter light in all directions, and some of it falls on a photodiode, a light-sensitive detector. The instrument counts and measures the flashes, and software inside the device converts them into a mass concentration, typically in micrograms per cubic metre of PM2.5, the fraction of particulate matter small enough to reach the deep lung.

Gases are measured differently. An electrochemical cell contains electrodes separated by a thin layer of electrolyte; the gas of interest diffuses through a membrane to one of them and is either oxidised or reduced, and the resulting current is proportional to its concentration. Cells of this kind, small enough to fit in a box the size of a paperback, are used for nitrogen dioxide, ozone and carbon monoxide. Each cell responds to one gas, so a device that reports several pollutants needs several cells. Metal-oxide sensors, which change their electrical resistance when certain gases are present, are cheaper still but less specific.

The attraction of such devices is partly price — they cost from a few tens to a few hundred pounds — and partly scale. A school, a residents' group or a university department can afford dozens or hundreds of them and hang them on lamp posts; some are mounted on bicycles, which allows a single instrument to cover a whole neighbourhood in an afternoon. Networks of this kind have pinpointed hotspots that official maps missed: a street beside a bus depot, the playground of a school next to a main road, a district where coal is still burned for heat. In parts of Africa and Asia, where official monitoring is scarce, cheap sensors have provided the only continuous record available.

Accuracy, however, is a different matter. The biggest single influence on an optical counter is humidity. In damp air water condenses around particles and forms droplets large enough to scatter light strongly, so a sensor that counts them as solid particles will report a sharp rise on a foggy morning when the air is in fact no dirtier. Some instruments dry or heat the sample before it reaches the chamber. Particle composition matters too: black soot absorbs light while pale sulphate scatters it, so a sensor calibrated in the factory for a typical urban mixture will misread smoke from a wildfire.

Sensors also drift. Fans become clogged with dust, lamps dim and detectors lose sensitivity, so that an instrument's response changes over months in ways its owner cannot see. Electrochemical cells decline in much the same way, and are cross-sensitive to other gases. The remedy is regular comparison with a reference monitor — co-location, in the trade's vocabulary — followed by correction of the raw signal. Because few cities have enough reference instruments to go round, researchers are now training machine-learning algorithms on the comparisons that do exist, so that a correction derived in one city can be transferred to another.

Getting the readings out of the sensor is a further design problem. A monitor fixed to a lamp post has no cable to the internet, so most devices send their data over the mobile phone network, or over a low-power radio link to a local receiver. Power is equally awkward: a unit that must be wired into the electricity supply costs far more to install than one that runs from a small solar panel and a battery. Much of the engineering effort now goes into instruments that can be fixed to a wall in minutes.

Standards are slowly emerging. The United States Environmental Protection Agency has published performance targets and testing protocols against which sensors can be judged, and several universities run open comparisons in which dozens of devices are exposed to the same air for weeks. The results are sobering: few cheap sensors meet the targets comfortably, and the cost of a reliable instrument, though falling, remains far above that of the cheapest consumer devices. The most promising research aims not at replacing the reference monitor but at combining many imperfect readings with weather data and models of how pollution disperses.

For legal purposes the distinction between the two kinds of instrument remains firm. A reading from a cheap sensor will not usually stand up in a prosecution, because its uncertainty cannot be stated with confidence. What such sensors do best is to suggest where the expensive instruments should go, and to show whether a policy is having any visible effect. They do not answer the old question — how clean is the air? — so much as allow a new one: whose air is dirty, and when?

The internet is often described as a network of networks, and the description matters for the argument that follows. When you send a message it is broken into packets, each addressed to its destination, and routers pass them on without enquiring what they contain. For most of the internet's history this has meant 'best effort': packets travel as quickly as the machinery allows. Nothing in that design prevents the company that sells you your connection from examining those packets and treating some of them differently from others. A packet carrying a video call and a packet carrying an email are indistinguishable to a router, but not to a company that chooses to look more closely.

The phrase 'net neutrality' was popularised in 2003 by Tim Wu, then a law professor, to name a principle rather than a technology: that a broadband provider should not favour or obstruct particular applications or content. The concern was not abstract. In 2005 a small American telephone company agreed to stop blocking its customers from using an internet calling service that competed with its own telephones, after regulators intervened. Three years later a large cable operator was found to have slowed traffic associated with a popular file-sharing protocol, and was ordered to change its practices.

The case for legal protection rests on incentives. A provider that owns the pipes and also sells television, telephony or streaming has both the means and the incentive to make rivals' services unreliable. A start-up cannot negotiate a fast lane, and the knowledge that it might one day need to do so discourages investment in exactly the kind of experiment that has produced most of the internet's successes. Supporters of the principle describe it as protecting 'permissionless innovation': the freedom to launch a service without asking anyone's consent.

The counter-arguments deserve a hearing. Not all discrimination is sinister. A surgeon operating a robot remotely, or a vehicle that must brake within milliseconds, has a genuine need for a path through the network with minimal delay, and networks must manage congestion somehow if heavy users are not to spoil the service for everybody else. Priority for ambulances and emergency calls, defenders point out, is a form of discrimination that nobody would want to forbid. Providers also warn that close regulation deters investment in faster infrastructure. And, critics add, where several competing networks reach most homes, a provider that blocked popular services would soon lose customers.

A more recent dispute concerns zero-rating, the practice of exempting certain apps from a customer's data allowance. It looks generous — free music, free social media — but it alters the market. A small rival whose service consumes the customer's allowance is competing not only with a larger company's product but with its price advantage, and where most people buy a bundle of data that advantage can be decisive. The customer pays for a quantity of data, and the operator decides which data count. Regulators in Europe and elsewhere have taken different views of the same practice, which is itself revealing.

There is also a technical objection to the way the debate is usually framed. Net neutrality rules have generally governed the connection between a customer and their provider, yet much of the internet's traffic passes through interconnection points where large networks exchange data. Disputes between content distributors and access providers have on several occasions left customers unable to reach a service, and rules aimed at the last mile did nothing about them. If the aim is to stop gatekeepers extracting tolls, the argument about where the gate actually stands ought to be settled first.

Enforcement has proved just as awkward. A rule that bans 'unreasonable' discrimination depends on somebody deciding what is reasonable, and the regulators who must decide rarely employ enough engineers. Provisions for 'reasonable network management' leave room for argument about whether a limit protects the network or protects the operator's other businesses. Meanwhile the rules have swung with politics: the United States adopted a strong version in 2015 and repealed it two years later, the European Union has legislated through a regulation on open internet access, and each change costs money and creates uncertainty for everyone building a service. Certainty, in this area, is worth almost as much as the rule itself.

My own view is that the argument is strongest where it is narrowest. Clear prohibitions on blocking lawful content, on throttling particular services and on charging content providers for priority would command wide assent, and a breach would be easy to detect. The case for treating internet access as a public utility, subject to price controls and detailed oversight of engineering decisions, is much weaker: it assumes that regulators know better than network engineers how a network should be run, and the evidence for that assumption is thin.

The debate is often presented as a contest between freedom and control, which is misleading. Every arrangement grants somebody the power to decide whose traffic goes first. The real question is whether that power should lie with the company that sells the connection, answerable to its shareholders, or with rules that apply to everyone. On that question I have no doubt: a network that treats all comers alike is worth the small loss of flexibility that it costs.