Every time you access an ATM, book a ticket or buy online, you are expanding the scope and range of algorithms. And their use is only likely to grow
Updated: Tue, Aug 27 2013. 09 22 AM IST
New Delhi: Every time you hit the search button on Google, the search engine sifts through thousands, if not millions of webpages, to spit out the content you are seeking in a fraction of a second. What makes this possible is the underlying algorithm—a simple set of mathematical rules embedded in the software.
In fact, every time you access an automated teller machine (ATM), enlist for a unique identity number under Aadhaar, book an air or train ticket or buy something online, you are expanding the scope and range of algorithms—a mathematical concept whose roots date back to 600 AD with the invention of the decimal system.
The use of computers, however, has elevated the sophistication and use of algorithms in daily transactions to unprecedented levels.
Every time you use a computer—your laptop, phone, or a mileage calculator in a car—you are using algorithms, says Dilip D’Souza, a Mumbai-based former computer scientist who writes the column A Matter of Numbers for Mint.
“Call them programmes, or software packages, or apps (applications), which they are,” he says.
According to D’Souza, even the Duckworth-Lewis method, which is used to calculate the target score for the team batting second when rain interrupts a limited overs cricket match, is based on an algorithm.
“We should understand that an algorithm is nothing but a set of instructions to be followed. The D/L method is essentially an algorithm, applied to a particular game situation (rain that takes away playing time) to suggest how to proceed,” adds D’Souza.
Although using algorithms to retrieve usable and relevant data from the huge amount of data being collected by agencies, including governments and corporate entities, is not a new phenomenon, the applications of it has expanded to touch deeper into daily lives of ordinary people. Denzil Correa, a PhD scholar at Indraprastha Institute of Information Technology in New Delhi, says that algorithms are not rocket science and have found use in daily lives for a long time now, though their use has become more prominent now.
“When you multiply two and two, then that is also an algorithm because you are following a certain step-by-step procedure. None of it is magic,” Correa said.
“Facebook and Google search are also based on algorithms, though they are more complex than the regular algorithms,” he said.
The Aadhaar card, which will soon evolve as the country’s universal identity card, is based on biometrics, which in turn use an algorithm to store and retrieve fingerprints and iris scans.
“Algorithms are everywhere. Whatever system you develop, first you develop an algorithm to run it. Everyone is trying to develop algorithms that make the comparison process as small as possible so that a person can be identified quickly,” says Karm Veer Arya, a biometrics expert from Indian Institute of Information Technology and Management, Gwalior.
“Biometrics refer to identifying humans by certain physical characteristics, for example fingerprints. Computer scientists have worked out algorithms that can analyze a given fingerprint and match it against a database. This is pretty well understood technology by now and therefore pretty reliable,” adds D’Souza.
Even though algorithms work on what humans define them to be and process large-scale data for certain purposes, not all these purposes might be for good, Correa adds.
For instance, the Indian government has increased the surveillance of its citizens and launched a central monitoring service under which it is mining all possible data, including phone calls, text messages, WhatsApp messages and communication using Skype.
“The government is collecting more data to know more about its citizens. However, biometrics is not a panacea or silver bullet. Traditionally, biometrics has been used to help intelligence agencies,” he adds.
In the West, algorithms are being used to map crime or prevent crime
A report in The Guardian published in July says that the US has already started using algorithms for predictive policing. It says that a team of criminologists and data scientists at the University of Memphis compiled crime statistics from across Memphis over time and overlaid it with other datasets—social housing maps, outside temperatures etc—then instructed algorithms to search for correlations in the data to identify crime “hot spots”. The police then stepped up patrols in those areas.
“Crime prevention is one significant goal of mapping crime. Past crime records which include time and location of a crime, and the criminal behaviour such as the modus operandi and the mobility of the criminal are also analysed. An algorithm is used to predict property crimes like house breaks and patterns are developed over years,” says K. Jaishankar, a criminologist based in Tamil Nadu.
The criminologist has been involved in research on how an algorithm-based GIS (geographic information system) systems can be used for tracing the ethnic, economic and political aspects of a riot. “Hotspots of riots can be mapped and these hotspots can be targeted by the police in a routine manner and this will ensure prevention of riots,” explains Jaishankar.
The use of algorithms is only likely to grow over the next few years as the consumer economy transitions to smart products like an Internet-enabled refrigerators.
But as Correa points out, algorithms are only as good as the use they are put to. “Computers are supposed to (be used to) aid humans. In the current scenario, algorithms are what humans define them to be; they should to be used to assist humans and not replace them.”