World Bulletin / News Desk

Algorithms are a crucial cog in the mechanics of our digital world, but also a nosy minder of our personal lives and a subtle, even insidious influence on our behaviour.

They have also come to symbolise the risks of a computerised world conditioned by commercial factors.

 A gift from a Persian scientist 

 Long before they were associated with Google searches, Facebook pages and Amazon suggestions, algorithms were the brainchild of a Persian scientist.

The term is a combination of mediaeval Latin and the name of a ninth century mathematician and astronomer, Al-Khwarizmi, considered the father of algebra.

A bit like a kitchen recipe, an algorithm is a series of instructions that allows you to obtain a desired result, according to sociologist Dominique Cardon, who wrote "What Algorithms Dream Of".

Initially known mainly to mathematicians, the term spread as computers developed. 

The brains of computer programmes are algorithms, and are thus a central cog in the internet machine.

 Where are algorithms found?  

"We are literally surrounded by algorithms," says Olivier Ertzscheid, a French professor of information technology and communication.

"Every time you consult Facebook, Google or Twitter you are exposed to choices" that algorithms calculate for us, and we are also sometimes influenced by them, he told AFP.

They reign in the finance sector, one example being high frequency trading programmes, which can execute trades in milliseconds driven by algorithms that analyze a range of market and economic factors. Their speed and rule-based nature means they can make markets volatile and have triggered so-called flash crashes in the foreign exchange and stock markets. 

Police forces increasingly use algorithms to predict where and when crimes are most likely to be committed. Predpol, a software programme, claims to have contributed to double-digit drops in burglaries, robberies and vehicle theft in several US states and is also used in Kent, southern England.

Satellite tracking and surveillance would not have reached the point they are at today without sophisticated algorithms.

  How Google began 

 In the 1990s, PageRank (PR) was created in Stanford, California by Larry Page and Sergey Brin, Google's co-founders.

PR made it possible to class web pages by order of popularity. It became the heart of the Google research engine, which responds to key words within a fraction of a second. In addition to PR, Google uses "a dozen algorithms... to deal with spam, detect copyright infractions" and handle other crucial tasks, Ertzscheid explains.

 Facebook and the 'filter bubble' 

 Facebook uses sophisticated algorithms to offer its more than 1.8 billion users worldwide personalised content, in particular on its News Feed service which compiles messages from "friends", and shares articles selected according to each users social media contacts.

One risk posed by such a system is that of "The Filter Bubble" according to Eli Pariser, who developed the concept in a book of the same name. Being surrounded by information filtered by algorithms based on one's friends, tastes and previous digital searches and choices, someone surfing the internet can be plunged unwittingly into a "cognitive bubble" that just reinforces their convictions and perspective on the world.

 Algorithms and the truth  

Another risk was exposed during the last US presidential election -- the prevalence of so-called fake news or hoaxes on Facebook and other social media. Facebook's algorithms were not designed to distinguish true from false -- a feat that is difficult even for artificial intelligence -- but the popularity of information.

Facebook chief Mark Zuckerberg has sought to deflect criticism that it had been used to fuel the spread of misinformation that may have impacted the presidential race, but the company responded to growing criticism by saying new tools would be provided so users could call attention to controversial content.

Market crises link?

They are agnostic on market direction, but ubiquitous in markets. Firms like Goldman Sachs and Morgan Stanley may get the headlines, but algorithms are the real force today on Wall Street.

The mysterious programs were at the center of the so-called "Flash Crash" in May 2010, when the US stock market plunged more than nine percent in a matter of minutes for no discernible reason.

The swoon, which was reversed within a matter of minutes, was sparked by a computerized order to sell a large quantity of S&P 500 futures in a short period. That, in turn, triggered a chain reaction of other computerized orders by high-frequency traders.

The quickfire crash and recovery underscored the pivotal role of algorithms, which execute automated trades after taking instantaneous readings of inputs that can include news events, economic data and stock price movements.

The era of automated trading began with the 1971 founding of the Nasdaq market, which introduced computers to Wall Street in a significant way.

Today, algorithms are the 900-pound gorilla in markets, playing a role in the vast majority of securities transactions.

"I would say it is 90 percent algorithmic, but there is not a great way to quantify that," said Valerie Bogard, an equity analyst at Tabb Group, a research and consulting firm.

"Even though (people) may send an order through a sales desk, it is possible it is going through an algo."

The Securities and Exchange Commission in 2014 levied its first fine for such bogus trades. Since then, US regulators have pumped more resources into enforcing the rules on electronic trading.

The SEC now requires firms to identify those who design and operate algorithms and can subpoena documents from firms employing the technology.

There has also been a preference among some investors to rein in the machines. The New York Stock Exchange has announced plans to slow down transactions in one of its platforms, following the introduction of a similar mechanism by IEX Group, which delays transactions by microseconds.

These moves were taken in response to criticism that high-frequency traders were stiffing investors by driving up transaction costs. 

But market insiders say there is no evidence of a clear trend as far as whether high-frequency trading makes the market more prone to crises than with conventional trading. After all, markets were also prone to violent gyrations prior to when the machines took over.

"When the crash in 1987 happened, it was before algos," said Noll. "It is a somewhat harsh reality but true."