What if we could close our eyes for a moment and be magically transported to Israel in the year 2025? What would we see – a continuation of the current status quo, two peoples locked in a bloody struggle? Would the Islamic State be sitting right on Israel’s borders? Or would the term ISIS be relegated to history? Predictions on these matters are not only an intriguing curiosity; they can be critical to the future of countries and their citizens. Computers and algorithms may be leveraged in order to achieve credible predictions of the future of such matters.
One way to use a computer to predict the future is to feed it numerous examples of events and their outcomes, and then feed it a new event and elicit a prediction on the outcome based on the previously seen examples.
For example, if we want to know if a particular flight will be delayed on a certain day, we’ll feed the computer all the past data about this flight, including delays, add data about other flights and data about external factors such as the weather. At this point we will “train” the computer until it can show good forecasting ability on flights for which the delay status is available. Once we’ve gained enough confidence in the computer’s prediction ability, we’ll move on to predicting delays for flights that have not yet occurred. The advantage computers have over people is their ability to process tremendous amounts of data using a range of algorithms in order to make predictions.
Similarly, to predict the future of the Israeli-Palestinian conflict, we could feed the computer data on a number of “parallel universes,” each with a different chronology of historical developments. In one universe, for example, the Oslo Accords and disengagement from Gaza would occur as they did in reality; in another, Israel would annex and apply Israeli law in the territories; and in a third, Yitzhak Rabin would sign a permanent peace treaty with the Palestinians. With the aid of the computer, we would observe the implications of each of these universes and try to find patterns. Unfortunately, current technology is not yet able to perform such an experiment, but it is progressing rapidly in that direction.
The fear/trust index
One way to simplify complicated situations with multiple variables, such as the Israeli-Palestinian conflict – and thus “make it easier” for computers to formulate predictions – is to deliberately refer to only some variables while excluding others. In many cases, it is possible to do so without affecting the quality of the prediction. This is a very intuitive idea that human beings make use of regularly.
Weather forecasting, for instance, is a complex business, but you don’t need a supercomputer to know that you should take an umbrella when the sky fills with gray clouds. In reality, scientists are working very hard to identify the special group of variables relevant to this problem, while ruling out the significance of others. This is one of the most fascinating challenges in computer science, and many approaches to solving it have been developed over the years.
One example of such a significant variable is the fear/trust index of Israelis in the context of the Israeli-Palestinian conflict. This is a hugely influential factor in terms of the evolution of the conflict and its future. Using fear indices for the purposes of prediction is not a new idea. The VIX index – a measure of investor anxiety that gives a real-time picture of risk – is regularly used on the New York Stock Exchange.
Suppose we had a fear/trust index of Israeli society, which could reflect at any given moment the fears of the majority of Jewish Israelis in regard to the Israeli-Palestinian conflict. The more worried we were about our personal security due to developments in Israel and the region, the more the index would drop. Such an index would perhaps decline if estimations of the Palestinian willingness to accept a neighboring state with a Jewish majority were low; or when the Palestinians were not showing the will or desire to impose discipline on the armed militias; or if Israelis lost trust in the international community due to a sense that Israel was not receiving fair treatment relative to other countries in the world. If we could credibly monitor such an index in real time, we could fare much better at predicting how the conflict would develop.
Were it available throughout the course of Israeli history, such an index would have certainly recorded many low points. However, a number of historical events would surely have advanced the index upward. Two memorable possibilities are Egyptian President Anwar Sadat’s visit to Israel in 1977, and King Hussein of Jordan’s visit 20 years later, following the massacre at Naharayim. In both cases, the leaders chose to make direct contact with the Israeli public – at personal and public risk. In both instances, the leaders convinced Israelis that their intentions were genuine and raised the level of trust, insofar as it existed then.
Sadat did not use computers to predict the outcome of his visit, of course, but he did calculate what the future could bring. In his autobiography, he wrote, “My calculation was that my trip to Jerusalem would break the vicious cycle in which we’d been trapped for so many years. My calculations turned out be quite accurate. For just as the reception from my people was fantastic, even incredible, the Israeli people – women and children and the elderly – were just as incredibly receptive. Even the members of Israel’s special forces and paratroopers assigned to guard me literally danced with joy.”
Sadat was right about the outcome of his visit, but he could have been wrong – for human history is largely the story of people’s right and wrong predictions. Sadat succeeded because he formed accurate insights about the mood among the majority of citizens on both sides, among Egyptians and Israelis. It’s amazing to see how much of the Israeli-Palestinian conflict is shaped by a collection of predictions that each side holds regarding the other. For example, today the prevailing belief among most Israelis is that most Palestinians are not ready for long-term peace that is based on two states for two peoples and entails forgoing the right of return for refugees. And the Palestinians have their own predictions about Israelis. The story of the conflict is the story of these predictions. Day in and day out, the future of the conflict is shaped by the form that these predictions take. But are these predictions necessarily correct?
Measuring the public mood
For computers to be able to predict the future of the conflict, they have to “know” its situation at any given moment. How can a conflict be measured? One prevailing management theory in the business world today is that it’s much easier to manage what can be measured. Therefore, many companies invest substantial resources into various types of measurements, identifying the areas where an improvement in productivity is possible. If conflict management were treated in the same way, we might develop a series of such measures, including real-time indices of the public mood.
Israeli intelligence is brimming with resources and talent, but does not focus on methodically gauging the mood of the Israeli or the Palestinian public, but rather is focused on identifying potential threats to Israelis. Therefore, even Israeli intelligence is not in an ideal position to predict future developments in the conflict. That leaves the gauging of the public mood to the independent pollsters, who generally do this at the request of different media outlets or other organizations. The use of polls is problematic, however – they are not conducted continuously and are limited in the number of questions and answers that can be included each time. For this reason, many information technology companies no longer rely on polls as a tool for improvement, instead looking to other kinds of data that can be continuously collected and measured. Technology giants like Amazon, Google and Microsoft, as well as data-rich industries such as hospitals, banks and municipalities, all pursue this type of collection and analysis.
The technology that makes mood analysis possible has made huge strides in recent years. The combination of the Internet in every home, social networks, natural language processing (NLP) and sentiment analysis, along with segmentation and profiling, has made it possible to mathematically analyze the general mood with incredible accuracy. Imagine if all of these technologies were brought to bear to better understand the conflict and manage it in real time.
It’s surprising to see what kind of things can be predicted by means of “mood measurement.” Such research usually begins with data collection, and then progresses to statistical experiments and a hunt for connections. Researchers have found they can predict with great accuracy the death rate from heart disease in different U.S. states just by analyzing the choice of words used on Twitter in those locales: The more positive the tone of the tweets and more positive their subject matter, the lower the death rate. The research also found that forecasting by Twitter analysis provided more accurate results than any other index – even more than analyzing the number of smokers or diabetics, and more than analyzing demographic and other data. One study managed to predict with 86-percent accuracy the daily rise or fall of the Dow Jones Industrial Average index, just by analyzing the general mood on Twitter that day.
The Power Law
In order to “measure the conflict,” different variables and connections must be carefully examined. Here, too, we are sure to discover connections that are counterintuitive and seem hard to explain. For example, earlier this year the journal Nature featured two articles on forecasting the incidence of crime and terrorism. Both articles mentioned an international database of terrorist incidents from the 1960s to the present. American computer scientists analyzed the data with the aim of discovering any sort of pattern that could point to something. One thing they looked for was any connection between the number of people killed in terrorist attacks and their frequency. Intuitively, one would not expect to find a link between one event and another, but the researchers found that terror attacks follow a certain mathematical pattern. It transpired that a similar pattern is also found in the intensity of earthquakes, the size of craters on the moon, the frequency of words in a language, the duration of power outages in a particular state, and collapses of the world capital market. This pattern is known as the Power Law, and currently there is much discussion about it and the way it affects human behavior.
In a world of such advanced technology, it’s disappointing to see that all attempts to understand the conflict and predict its future are based on outdated tools and very partial information. In his book “The Wars of the Jews,” the first century C.E. historian Flavius Josephus quotes King Agrippa: “When the vessel is still in the harbor, my friends, it is wise to foresee the approaching storm and not to sail out into the middle of the hurricane to sure destruction. For those on whom disaster falls out of the blue are at least entitled to pity, but a man who plunges into destruction with his eyes open earns only contempt.”
We don’t know what storms are approaching, but we do know that our future here is too important to all of us to just let it happen.
The writer is a researcher in the field of computer science and data mining, and a high-tech entrepreneur.
Want to enjoy 'Zen' reading - with no ads and just the article? Subscribe todaySubscribe now