IELTS Academic Reading Practice Test 170

A non-stick frying pan seems a modest piece of household equipment, yet the surface inside it is the product of one of the twentieth century's more surprising discoveries in industrial chemistry. The coating is almost always polytetrafluoroethylene, usually shortened to PTFE, a plastic that resists heat, water, oils and almost every chemical it meets. Eggs slide out of a pan coated with it because its surface energy is exceptionally low: nothing readily bonds to it, including food. The story of how this slippery white solid came to be sprayed on kitchen equipment involves a laboratory accident, a wartime emergency and a fisherman with a clever wife.

On 6 April 1938, a young chemist named Roy Plunkett was working at the DuPont research laboratories in New Jersey. He was experimenting with refrigerants, the gases used in cooling systems, and had filled a cylinder with tetrafluoroethylene, a compound of carbon and fluorine. When he opened the valve, no gas came out, although the cylinder still weighed the same. He sawed it open and found a white, waxy powder inside. The gas had polymerised — its molecules had joined into long chains — and the result was PTFE, a substance that no one had set out to make.

What made PTFE remarkable was the strength of the bond between carbon and fluorine. Fluorine is a highly reactive element, but once attached to carbon it holds on tightly, forming a shield around the carbon backbone of the chain. The resulting material is chemically inert: it does not react with acids, alkalis or solvents, and it can withstand temperatures from far below freezing to around 260°C in continuous use, while its melting point is about 327°C. It also has one of the lowest coefficients of friction of any solid, which is why it has been used in bearings and seals as well as in pans.

DuPont registered the trademark Teflon in 1945, but the first users were not cooks. During the Second World War PTFE was needed for gaskets and seals in equipment handling uranium compounds for the American atomic programme, and for insulation in radar and aircraft. It was expensive and difficult to process, and production was controlled. Only after the war did manufacturers begin to look seriously for civilian markets, and even then the material remained a specialist product of the chemical industry rather than something the public could buy.

The person who changed that was Marc Grégoire, a French engineer and keen fisherman. In the early 1950s he was looking for a way to stop his fishing line from tangling and his tackle from corroding, and he experimented with coating them in PTFE. His wife, Colette, suggested that the same treatment might stop food from sticking to her cooking pans. Grégoire developed a method of bonding the plastic to aluminium, which he patented in 1954, and in 1956 he founded a company whose name, Tefal, combined the words Teflon and aluminium. The first non-stick pans went on sale in France that same year.

American cooks were introduced to the idea a few years later, and the pans became a consumer sensation in the early 1960s. The earliest coatings were thin and easily scratched, and manufacturers advised owners to use wooden or plastic utensils, to wash the pan gently and never to heat it when empty. A scratched pan lost much of its value: once the metal beneath was exposed, food stuck to that spot and the coating around it began to lift.

Successive generations of coatings addressed these weaknesses. Manufacturers applied the plastic in several layers, added a primer that gripped the metal more firmly, and embedded hard particles such as ceramic or minerals in the top coat to resist scratching. Aluminium pans were pre-treated so that the coating had something to hold on to. Many pans today carry three or more layers, and the more expensive ones have a textured surface or a sprayed ceramic finish. Each of these refinements lengthened the life of a pan, though none made the coating permanent.

A more serious problem emerged in the 2000s, and it concerned not the PTFE itself but a chemical used to make it. Perfluorooctanoic acid, or PFOA, was employed as a processing aid in the manufacture of PTFE and some other fluoropolymers. It persists in the environment and has been found in the blood of people and wildlife around the world, and studies of exposed communities have linked it to health problems.

Because the chemical does not break down readily, it has spread far from the factories where it was used. Under pressure from the United States Environmental Protection Agency, major manufacturers agreed to eliminate PFOA from their processes, and they reported that they had done so by 2015; some of the replacements, such as GenX, have themselves since come under scrutiny.

Today the non-stick market is broader than PTFE alone. Ceramic coatings, which are typically based on silicon dioxide and applied as a gel, tolerate higher temperatures but often lose their release properties sooner. Cast iron and stainless steel, properly seasoned, remain the choice of many professional cooks, who value durability over convenience. PTFE itself has found other uses: expanded PTFE, a form riddled with tiny pores, is the basis of waterproof fabrics that still allow moisture to escape. The non-stick pan, in other words, is a compromise like most technologies, and its slippery surface is a triumph of chemistry with a complicated history behind it.

In April 1965 the magazine Electronics published a short article by Gordon Moore, then director of research at Fairchild Semiconductor. Moore had been invited to predict how the components of the integrated circuit would develop over the following decade. Looking at the few data points available, he noticed that the number of components that could be fitted onto a single silicon chip had been roughly doubling every year, and he suggested that the trend would continue for at least ten years. The observation was later named Moore's law. It was never a law of nature; it was a forecast that an industry decided to treat as a schedule.

To understand why this mattered, it helps to know what a transistor does. A transistor is, in essence, an electrically controlled switch. Current flows through a channel of silicon between two terminals known as the source and the drain, and a voltage applied to a third terminal, the gate, determines whether the channel conducts. The gate is separated from the channel by an extremely thin layer of insulator. Switching a transistor requires a small electric charge, and making the device smaller reduces the amount of charge needed, so each operation takes less time and consumes less energy. This is the physical reason why miniaturisation has delivered such consistent gains in computing power.

Moore revised his forecast in 1975, slowing the expected rate of doubling to about once every two years, and it is this version that most people now have in mind. The name itself was popularised by Carver Mead, a physicist at the California Institute of Technology. More important than the wording was the use the industry made of the idea. From the 1990s, collaborative roadmaps set out, technology by technology, the dates by which each advance would be ready. A prediction had turned into a target, and thousands of engineers were employed to meet it.

For three decades the shrinking was accompanied by another piece of good fortune. In 1974 Robert Dennard and his colleagues at IBM showed that if a transistor's dimensions and its operating voltage were reduced together, the power consumed by each square millimetre of silicon would stay roughly the same. This principle, known as Dennard scaling, meant that each new generation of chips could run faster without overheating. It held until about the middle of the 2000s, when voltages could no longer be lowered without excessive leakage of current. Clock speeds then stalled at a few gigahertz. Further gains came instead from putting more processor cores on a chip, and from designing specialised circuits for particular tasks.

The numbers involved are hard to grasp. The Intel 4004, the first microprocessor sold as a single chip, appeared in 1971 with about 2,300 transistors. A modern processor contains tens of billions, and the largest designs, assembled by joining several pieces of silicon together, exceed a hundred billion. The labels attached to each generation are also misleading. When a manufacturer advertises a three-nanometre process, the figure no longer refers to a single physical dimension on the chip; it is closer to a marketing name for a family of manufacturing techniques. Since a silicon atom is roughly 0.2 nanometres across, the features involved are only a few dozen atoms wide.

Patterns are drawn on silicon by photolithography. A wafer is coated with a light-sensitive resist, light is shone through a mask carrying the design, and the areas exposed are washed away, leaving a stencil through which material can be added or removed. Finer patterns require shorter wavelengths, and the industry's latest machines use extreme ultraviolet light with a wavelength of 13.5 nanometres, generated by firing a laser at tiny droplets of tin. Only one company, ASML of the Netherlands, builds these machines, and each costs more than a hundred million dollars.

The cost of the factories has become a constraint in its own right. A plant that can produce leading-edge chips now costs several billion dollars, and the most advanced exceed ten billion, a sum that few companies can raise. As a result, only a handful of firms still manufacture the most advanced processors. Economists point out that this limit is of a different kind from those described by physics: it is a limit of capital.

Physical limits are also drawing closer. Insulating layers only a few atoms thick allow current to tunnel through them, and the wires that connect transistors become relatively slower as everything else shrinks. The industry has responded by changing the shape of the transistor itself. In the early 2010s designers began to raise the channel into a fin so that the gate could grip it on three sides, and more recently they have used structures in which the gate surrounds the channel completely. Another response is to build upwards: stacking layers of circuitry on top of one another, or joining separately manufactured chips into a single package, allows more transistors in a given area without making each one smaller.

Moore's law, then, is best understood as a description of an economic and engineering programme rather than a rule of physics. The doubling has slowed, and some observers pronounce the law dead. But the accumulated effect of six decades of compounding is so large that even a much slower rate of improvement will change what computers can do. What has ended is not progress but the particular rhythm to which the industry became accustomed.

Few disputes in computing are argued with as much moral conviction as the one between open-source and proprietary software. On one side are those who regard secret code as a restriction on the users of a product they have paid for; on the other are those who see software as an engineered product like any other, in which a company's investment deserves protection. The argument is usually conducted as a contest between freedom and profit. I want to suggest that this framing obscures the question that matters, which is not whether code is free but who can inspect it, repair it and leave.

The distinction is straightforward in principle. Proprietary software is distributed without its source code — the human-readable instructions from which the program is built — and its licence forbids copying, modification or redistribution except on the owner's terms; revenue comes from licences, subscriptions and support. Open-source software is published under a licence that grants those freedoms, subject in some cases to conditions. The practice has a history. Richard Stallman launched the GNU project in 1983, founded the Free Software Foundation two years later, and wrote a licence, the General Public License, which requires anyone distributing a modified version to release it under the same terms. The term open source was adopted in 1998 by people who preferred a business-friendly label for much the same idea.

Open-source programs are no longer a marginal enthusiasm. The Linux kernel, begun by Linus Torvalds in 1991, runs most of the world's web servers and, through Android, most of its smartphones. Much of the software that directs internet traffic, stores data and runs scientific computing is open source, and companies that once regarded it as a threat now contribute to it. IBM paid about 34 billion dollars for Red Hat, a company whose business is selling support for open-source software, and Microsoft, which in 2001 had described Linux as a cancer, later became one of its largest contributors.

The case for openness rests on three claims. The first is scrutiny: if anybody may read the code, defects and hidden behaviour can be found by anyone, not only by the vendor. The second is durability: a program whose source is available can be maintained after its author loses interest or goes out of business. The third is competition: a customer who can switch suppliers, because the code can be moved to another firm, is not captive. The first claim, however, is weaker than its supporters suggest. In 2014 a serious flaw in OpenSSL, an open-source library used to encrypt a large share of internet traffic, was found to have gone unnoticed for about two years.

The case for proprietary software is not merely the case for profit. Software that governs a hospital's records or an aircraft's controls must be maintained, documented and supported by someone who can be held responsible, and customers pay for that assurance as much as for the code. Companies also argue, reasonably, that the years of work needed to make a complicated product usable and secure will not be funded by goodwill. The dichotomy itself has blurred: most commercial products contain open-source components, while the largest open-source projects are developed mainly by engineers paid by corporations. The Linux kernel is a corporate artefact as much as a communal one.

Where, then, should the line be drawn? My view is that the argument matters most where software performs a public function. When a government holds personal data, runs an election or operates a water network, the public has an interest in the code being open to inspection, and in the state not being permanently dependent on a single supplier. Several governments have adopted open formats for documents for exactly this reason, and some cities have tried to migrate their offices to open-source systems, though the results have been mixed and the retreats as instructive as the successes. Public money should not buy secrecy about the handling of public business.

None of this makes openness a virtue in itself. A licence can be open and the project badly run; an open project can be captured by a company that changes its terms once the community has made it valuable, which is why the freedom to copy the code and continue it separately — to fork it — matters more than any particular licence. Nor is proprietary software inherently dishonest. Much of what people value in modern computing, from video editing to accounting, has been built by firms that keep their code. The relevant question is whether customers have real alternatives and whether they understand the bargain they are accepting.

In practice, the most useful reforms are modest. Institutions should check licences before they sign; they should insist on open formats so that their own information can be moved; and they should help pay for the maintenance of critical open-source components. The last point is easily forgotten: in late 2021 a severe flaw was found in a widely used logging library maintained by a small number of volunteers, a reminder that a great deal of the world's digital plumbing rests on unpaid labour.

The open-source debate is therefore at its most useful when it abandons the language of morality and asks about power. Code that cannot be examined, moved or replaced concentrates control in the hands of those who supply it. That is the real argument for openness — not that it is free, but that it leaves the user with somewhere to go.