Moore's Law states that the number of transistors on a computer circuit will double every two years . Gordon Moore, who was both one of the founders of Fairchild Semiconductor and a former CEO of Intel first observed the phenomenon. Transistors are a type of semiconductor made up of either silicon or germanium that can manipulate the flow of electrons.
In a more abstract interpretation, Moore’s law refers to the more general principle in computing of constant improvement and the unstoppable evolution of technological efficiency and power. The law implies exponential growth. Because Moore’s law is an observation of a historical trend and not a scientific law like the 1st law of thermodynamics, its definition is more easily manipulated and changed in terms of its content and implications. While generally interpreted as a positive trend, Moore's Law and its underlying ethical implications have become a topic of debate, especially as it pertains to the ethical development of emerging technologies within the fields of energy consumption, algorithms, and online privacy.
- 1 Background
- 2 Moore's Law in Other Contexts
- 3 Moore's Law is Dying
- 4 Ethical Implications and Emerging Technologies
- 5 References
Computers are fundamentally a collection of electronic switches. These switches control how electrical charges are routed communicate in order to represent bits, binary digits, that can be translated into understandable information. The first computers, everything from the Model K to the AVIDAC, used vacuum tubes as switches. Inefficiencies in vacuum tubes’ excess heat generation and tendency to break lead to the creation of transistor. Transistors are much smaller, faster and require less power than the vacuum tube. Transistors were invented by John Bardeen, Walter Brattain, and William Shockley. All three men worked for Laboratories and were awarded the Nobel Prize in Physics in 1956.
Creation and Iterations of Moore's Law
Moore's Law was named after Intel's Co-Founder Gordon Moore. In 1965, Gordon Moore put out a short report that would later become what we know today as Moore’s law. In the paper, titled “Cramming more components onto integrated circuits”, he simply referred to a historical trend, observing that, for a set cost, the number of integrated components has increased a rate of 2x per year. This means that the number of transistors within a constant physical space will double every year. In this article, he also predicted that this trend would continue until at least 1975.
At the time, the computer chip industry was capable of integrating tens of transistors on a single silicon die, and Moore was projecting the integration of up to “65,000 components on a single silicon chip”. During the IEEE technology conference in 1975, Moore made a revision to his previous statement. Citing new research and advancements, he said that the doubling of transistor density would continue, but at the pace of 18-24 months for the foreseeable future. Shortly after his revision, a professor at CalTech named Calvin Mead gave Moore’s prediction the name we know it as today.
Moore's Law in Other Contexts
Eroom’s law is another observation of historical trends, but it uses data derived from the research and development of pharmaceutical drugs. It is virtually the opposite of Moore’s Law, and so Eroom's Law is essentially Moore's Law backwards. Originally proposed in the Nature Journal, the phenomenon describes, in the last 60 years, how the number of new drugs approved per billion US dollars has halved roughly every 9 years. While computers are becoming exponentially more efficient, drug discovery is becoming exponentially less efficient. What is significant about this law, is it indicates that certain factors outweigh the technical improvements being made. This means that although progress in commercial drug research is increasing, it is not producing a similar increase in output of solutions, which suggests the underlying problems may not be correctly diagnosed.
Published in a paper titled, “Biology is Technology”, Robert Carlson examined biology as a human technology. In his analysis, he found that, controlling for cost, the performance of technologies including DNA sequencing, DNA synthesis, and a range of other computational tools used in biotechnology has and would continue to double every 2 years.
Andy and Bill's Law
Named after Former Intel and Microsoft CEOs Andy Grove and Bill Gates, the law states that that every time a computer chip is released, software is updated to take full advantage of all of its features. This is closely tied to Moore's Law as Moore's law is what allows the new computer chips to be more powerful, which in turn allow for more and more complex software
Named for Robert Metcalfe, inventor of the Ethernet networking protocol, it states that the a network's impact is the square of its nodes. This help explains the rapid growth in utility of networks such as the telephone and the internet, as well as the growth of business and social networks, such as FedEx, EBay, Facebook, Instagram, and Amazon.
Moore's Law is Dying
Starting in 2010, Moore’s Law began to slow down for two main reasons. The first is electrical leakage. For the second half of the 20th century, the miniaturization of transistors meant that they would be more energy efficient. Recently, they have gotten so small, even as small as 10 nanometers, that the transistor through which the electronic currents flow cannot always contain it. The second reason is based on economic factors. As transistors get smaller and generate more heat as a result of their instability and greater concentration of units, the amount of infrastructure needed to keep the transistors cool also increases, and the cost of cooling large server rooms is getting more and more expensive. This cost of cooling is the main factor holding back transistors from improving as they have been according to Moore's Law. Within the coming years, the general curve that the hardware industry has observed under Moore's Law is no longer applicable as it is simply no longer possible for the trend to continue barring the introduction of completely new technology that would massively change the way transistors are used and constructed. These realities led the CEO of Nvidia to declare Moore’s law to be dead at CES 2019.
Ethical Implications and Emerging Technologies
Moore’s Law, more abstractly, gives a name to the constant, iterative improvement in computational efficiency. This innovation has given us supercomputers in our pockets, the ability to buy household items on an online marketplace and receive them in 2 days, and the means to connect with those who are thousands of miles away. However, as Phillip Brey writes in his paper, Anticipating ethical issues in emerging IT, "current ethics...are insufficiently equipped to address the revolutionary changes that are being brought about with new and emerging technologies." When the pace of technological advancement, underpinned by Moore's Law, is far greater than that of its complementary practices of ethical scrutiny, emerging technology continues to progress unchecked, and the negative, non-moral values internalized by the technologies only come to light as the result of disaster and tragedy. The troublesome aspect of this rapid development is the concerning slow regulation of it. A potential solution to this is to make the matter more publicized and less concentrated within the realm of academia.
One of the nagging, unforeseen consequences of Moore’s Law is energy consumption. As computer chips become more powerful, the amount of power they consume also increases, as does the ratio of overall human energy consumption that is devoted to computational practices. The ethical implications for this eventuality come from the means by which the energy is being produced. Events like the 3 Mile Island and the Exxon Valdez oil spill are not likely to stop as long as energy demands are increasing. One could argue that those events allowed us to learn from our mistakes, but, as Daniel Dennet writes in Virtues of Ignorance, “can we say, with confidence better than a coin flip, whether 3 Mile Island was one of the good things that have happened or one of the bad?”.
Side-effects of disruptive technology can present themselves in unintended ways. Technological advancement facilitated by Moore’s Law has lead to developments in automated algorithms, but this rapid growth often goes unchecked and unregulated. Safiya Noble discusses instances of algorithmic search engines reinforcing racism in her book, "Algorithms of Oppression". Frances Godzinsky delves into the mechanisms behind technological bias in her paper titled, “The ethics of designing artificial agents”. She argues that the burden of responsibility for the actions of these artificial agents falls on the human designers. Both Godzinsky and Noble see that there is an external prompter for every piece of technology and that technology often internalizes the biases of its creators. For Godzinsky and Noble, the solution for this issue is to slow the pace of progress and more cautiously consider what the ethical consequences of these advancements might be. This emphasis on careful thought preceding technological innovation is at odds with the principles that underlie Moore’s Law.
Moore’s law also speaks to increasing computational efficiencies being coupled with increasing monetary efficiencies. As it becomes more cost effective to aggregate and store massive amounts of information, the tendency to do so increases. One implication of this is that information on the internet doesn’t really get deleted. A lack of control over personal information’s availability online has raised questions of online privacy. Leading ICT ethicist Luciano Floridi writes that the right to privacy is a right to a renewable identity, an idea that doesn't mesh well with the ease of keeping information online that Moore's Law provides.
Decline of Moore's Law
In 2016, along with its second quarter earnings, Intel also revealed that the previously set at 2 years doubling period had most likely become 2.5 years or roughly 30 months. Intel pushed back its next generation transistor technology release by a year to 2017, while suggesting that silicon transistors could only perhaps continue shrinking for another 5 years, likely leading to wide repercussions in the technology sector, especially for phones and handheld devices.