Saturday, March 31, 2012

The world entered the age in which computers take control

One of the most important inventions of the twentieth century, the computer has brought immense benefits to humanity, enabling more efficient task. If the computer facilitate the achievement of the last century decisions of people, today it is becoming increasingly responsible for taking decisions on our behalf, taking the slightly-more easily control all aspects of our lives.

In recent decades, computing has been dominated by so-called "Moore's Law". This is sometimes called co-founder of Intel, Gordon Moore, who in 1965 observed that since the integrated circuit was invented in 1958, the number of transistors on a silicon chip applied to doubling every 12 months. Moore predicted that this trend will continue for "at least 10 years". His prediction was confirmed over time, remained valid until today (even if the time necessary to double the computing capacity is about 18 months). Moore's Law has long been adopted by companies aimed at producing microchips, and ensuring validity for a long time from now.

This exponential growth of processing capacity chips started a revolution, allowing the emergence of personal computers, the Internet and giant companies that dominate Earth today, including Google, Apple or Microsoft.


One of the most important inventions of the twentieth century, the computer has brought immense benefits to humanity, enabling more efficient task. If the computer facilitate the achievement of the last century decisions of people, today it is becoming increasingly responsible for taking decisions on our behalf, taking the slightly-more easily control all aspects of our lives.

In recent decades, computing has been dominated by so-called "Moore's Law". This is sometimes called co-founder of Intel, Gordon Moore, who in 1965 observed that since the integrated circuit was invented in 1958, the number of transistors on a silicon chip applied to doubling every 12 months. Moore predicted that this trend will continue for "at least 10 years". His prediction was confirmed over time, remained valid until today (even if the time necessary to double the computing capacity is about 18 months). Moore's Law has long been adopted by companies aimed at producing microchips, and ensuring validity for a long time from now.

This exponential growth of processing capacity chips started a revolution, allowing the emergence of personal computers, the Internet and giant companies that dominate Earth today, including Google, Apple or Microsoft.

Recently, however, a new trend was reported by a group of specialists in science and technology in a report prepared for the White House. They compared the time required to optimize production tasks. A 1988 computer, using software algorithms existing at the time, would have solved this problem in 82 years. Instead, a computer in 2003 that would have used existing software algorithms that year would have solved the same problem in just one minute, which is an improvement by a factor of 43 million.

The gain in productivity contribution consists of processors multiplied by the algorithms. Researchers were surprised that more efficient processors contributed only by a factor of 1,000 improvement observed and improved algorithms to a factor of 43,000.

Thus, the increase in performance software algorithms have an upward curve even higher than the computational ability, ingenuity specialists have developed these algorithms accelerating rate of progress in many areas.

What
are algorithms?


Algorithms are a set of mathematical instructions that are designed to solve a problem. The evolution of programming languages ​​and developing more efficient algorithms have led to more efficient computer processing capacity.

If a few decades ago requiring sophisticated computer algorithms strong (and also expensive) that allowed them only large companies, military and weather organizations, progress in recent years made ​​it possible to implement them in surprising places - for example, in vacuum . Roomba autonomous vacuum cleaner robot is guided through the room by an algorithm that chooses the best way for a perfect cleaning.


Without realizing it, the algorithms controlling much of our everyday life. From the use of search engines to give us the most relevant results and social networks to show us our friends' latest activities, to the algorithms used by GPS devices to choose the shortest route or retailers to predict what products will be required, software algorithms come to increasingly dominate many areas.

Financial markets, controlled by software algorithms

An important area that has come to be controlled largely by the algorithms is that of financial markets.

Thanks to powerful algorithms and computers, in one second can be processed thousands of sales and purchase orders, people making it impossible to match such a performance.

London Stock Exchange can process an order of microseconds to 124 since its launch. For comparison, the fastest human reaction to auditory stimuli occurs, it took place in about 140-160 milliseconds. Any human response faster than 140 milliseconds is actually a prediction. Therefore, at sporting events, when an athlete starts the race in 0.1 seconds (100 milliseconds) after the acoustic signal, the situation is considered "false start" because it would not have had time to react to sound, but tried to predict its time.

It is estimated that Wall Street, in a typical day, up to 70% of transactions have a computer at home, not an order placed by a person.



Because only a few milliseconds algorithms react to any tendency of market transactions by them tend to be a growing fraction of total sales and purchases.
A recent report published by Reuters provides evidence of the importance of algorithms in world markets. News agency reported that in August this year, hedge funds controlled by computers were able to profit in a volatile market, while funds controlled by people with negative results.
Because algorithms can perform thousands of transactions over a single minute, they can take advantage of moments of market instability, rapid response capability giving them a huge advantage over human brokers.
Such a moment occurred in mid-August, when one of the most popular U.S. indexes, Dow Jones fluctuated heavily over a period of half past one. In such moments, most transactions are made of algorithms, according to Tabb Group, a company specialized in researching and analyzing financial markets. Ted Weisberg, one of the brokers on Wall Street, Los Angeles Times commented on the stock market changes hallucinating: "I think it's great if you are smart enough that they take advantage of these fluctuations. But in reality, the market movements are not caused ordinary investors. All these fluctuations are caused by algorithms ".
Because the algorithms react quickly to market movements, there were instances in which trained heavy snowfall in exchange for no apparent reason. The best known example is the so-called "Flash Crash" of May 6, 2010. Then, the Dow Jones fell 900 points, equivalent to 9% of the total, the eight largest companies shares reaching the value of one cent.
The market recovered quickly, the same day, but the financial world was shocked by the sudden collapse of its actions, and exchange officials have imposed trading limits, never to allow such algorithms to produce a fall.




Recently, a new innovation appeared in the computer world of Wall Street: algorithms that try to trick other algorithms. How does this practice, which is called "algo-sniffing"? Electronic signature system detects a computer algorithm that intends to buy shares of a company, purchase those shares milliseconds before it, then sell them something more expensive, getting a healthy profit in less than a few seconds.

World Finance is only the start to take control algorithms. Soon, cars may be driven by algorithms, which would act as driver.







DARPA, the Pentagon wing is to develop new technologies for the U.S. military, launched in 2004 challenging the best research departments of American universities. Mission? Designing a vehicle to drive one thanks sensors, hardware and sophisticated software algorithms, applying all knowledge of robotics and artificial intelligence.

Again, progress was remarkable. In the first year of the contest organized by DARPA, 2004, no vehicle was able to walk without problems more than 10 km. In the competition organized in 2007, 5 cars (each belonging to other universities) have managed to finish the race that stretched about 130 kilometers in an urban environment.

All cars were able to successfully complete complex maneuvers such as entering a busy road or providing proper priority at an intersection where there is the 4-way road marking "STOP".

Google has hired teams that were ranked first and second places to start the project "Google driverless car". In 2011 Google designed car has traveled 225,000 kilometers without being led by a man is driven exclusively by software algorithms that analyzed data from the cameras and sensors installed on the machine.

Those who argue in favor of cars that drive themselves say that they never tire and never is never drunk so that their use would save many lives. Engineer Sebastian Thrun, project leader Google driverless car, the car presented in this year's TED conference, saying that "future generations will look back and say" how ridiculous it was when people drive cars. '"



European Union project designed SARTRE (Safe Road Trains for the Environment), which aims to develop technologies that will allow cars to form long convoys on the route. Thus, only the first vehicle will be actively driven by a driver while the cars behind it would lead single, chasing her on the front automatically. The vehicles will communicate with each other on a special frequency so that all can respond timely to any decision taken by the driver first car.
EU claims that this technology will improve traffic, will provide more comfort for drivers and lead to lower consumption, while reducing the cost of travel and pollutant emissions. In January 2011, Volvo made its first test of technology, which was a great success:
Those involved in the EU project announces that this technology can go into production within a few years. "The more you take, however, acceptance by the public," they stated.
According to European statistics, the human factor behind at least 80% of road accidents. As the cars took control, road trips will become more reliable and secure wireless communication between machines will reduce traffic jams.
Remains to be seen, however, whether drivers will accept to cede control of a software algorithm.
"Cruise Control" leads to oblivion
As we accept to give up control of your computer, you have to be prepared for the consequences of this decision. A warning in this regard comes from aviation, where technological progress has made the autopilot to be so used, that pilots have lost touch.
A study by the Federal Aviation Administration (FAA) in the U.S. shows that the pilots "delegated too much responsibility automated systems", and a large number of security experts say that the number of cases in which pilots do not know how to react when confronted with disabling autopilot is increasing.
Rory Kay, airline pilot and co-chairperson of the FAA pilot education, told the Associated Press that he "observed a new type of accident in these sophisticated aircraft - that the pilots forget how to fly".


Skill loss caused hundreds of deaths in aviation accidents in the last five years, experts say. During this period there were 51 accidents in which planes came into tailspin in flight and pilots could not intervene to fix it. According to the International Air Transport Association (IATA), this type of accident is the most common air.
According to FAA study, pilots were increasingly fewer opportunities to steer manually to maintain skills, because many airlines and regulatory organizations discourage or even prohibit pilots off autopilot.
"Airline industry suffers from a reliance on automation," said Kay. Another member of the FAA and airline pilot also, Bob Coffman said that "more and more airlines report incidents in which pilots wasting time trying to restart autopilot, you should take control and fly the plane".
Under applicable law, pilots are required to use autopilot altitude between 8900 meters and 12,500 meters, ie the portion that planes spend most time. Coffman said that most airlines advise pilots to activate the autopilot at about 90 seconds after takeoff, the plane rises about 300 meters, and keep it turned up to 90 seconds before landing.
Even so, to strive for greater automation. In the U.S., the FAA intends to replace air traffic control systems that rely on radar technology with more accurate, based on GPS. This change will allow the substitution of old used to landing.
If now the speed aircraft must descend for landing, handling time and fuel consuming, thanks to GPS technology aircraft will be able to make steep descents, planând. This new technique will allow more frequent and more frequent landings, even in adverse weather conditions, the pilots will know the precise location of other aircraft and ground objects. The only downside of the new methods of land disposal is one that requires the onboard computer control. "These new procedures should be performed on autopilot, because one sneeze is enough to ruin everything," said Bill Voss, president of Flight Safety Foundation, an organization that promotes aviation safety.
Thus, "the pilots' aircraft will continue to guide aircraft to the destination, but their role is more like that of a programmer than the image that we all have in mind, that the aircraft with its hydrostatic drive.
Cars and planes are not the only elements of our lives driven algorithms. Recently, they have become a much more intimate guide our life.


Algorithms and love handles?

In recent years, more people turn to Internet to find a partner. If 10 years ago half of job seekers on the Internet were not seen very well, today's "virtual matchmaker" is accepted by societies worldwide. A survey in 2010 for BBC World Service in 19 countries across the world shows that one third of Internet users consider it a good place to find half.

Today, industry "online dating" is worth only U.S. $ 1.9 billion and the largest site in the world, match.com, has 1.8 million paying customers and has been used to establish 75 million.


Manages this site as a large number of people and how he recommend them right partner? Of course, it is an algorithm. "Synapse" of the Match algorithm takes into account a number of factors to find the ideal partner for each. It does not take into account only the explicit choices of each user, but tries to "guess" what they want members of the site depending on the actions they will undertake on site.

One of the engineers who worked on the design algorithm for the Financial Times explained how it works: "The algorithm learns the same way that human brains do. When you give a stimulus, it creates certain neural pathways. When we May like something, they disappear. Thus, learning along the way. "He added: "Before, potential partners that they present users were chosen according to preferences set by them. But I found that people are always interested in what you say they are interested".

Change algoritmulului to enable it to learn from user behavior led to a doubling of the number of interactions between site members with those recommended by the manufacturer, and now more than half of the messages they send members of the site were addressed to people suggested by the algorithm. As the site builds more and more terabytes of data, the algorithm gets better.


What does the future hold?

So sophisticated algorithms that process large amounts of data already managed to control the essential aspects of our lives, from how we find life partner to decisions on financial exchanges that decide the fate of the global economy.

Every day, sophisticated algorithms are emerging in new areas, with the promise that they will produce radical changes in them. In medicine, Medical Predictive Technologies company announced this summer that has developed an algorithm which, when connected to devices monitoring a patient's health, they can analyze data generated to predict up to 24 hours in advance of events such as heart attacks or respiratory crisis. This will allow health professionals to work by adjusting the treatment to prevent them.


What next? Take control algorithms in many aspects of small, due to the exponential growth of available data.

Due to lower production costs, all sensors can be integrated into more of the world around us. All these sensors produce data being processed by more sophisticated algorithms. For example, company Thames Water, which manages water resources in London, began to install sensors on water pipes, and analyze data generated by these algorithms to predict when there is a high probability that the pipes to crack. Thus, the computer can be programmed repairs even before failures occur!

As sensors become ubiquitous, the algorithms required for processing data generated by them will control more and more aspects of everyday life.

If the majority of science fiction writing was the computers took over this unpleasant consequences for humanity, in reality these algorithms seem to bring greater efficiency, lower consumption of resources and improved quality of life. However, their integration into our lives to be realized, taking into account the warnings offered by the unexpected collapse of Wall Street and pilots of aircraft position.

One thing is certain: the inexorable march of progress can not be stopped. Depends only on us but if we know how to manage its consequences.

No comments:

Post a Comment

Powered By Blogger