7 January 2016

Did data predict the Pathankot attack?

https://www.linkedin.com/pulse/did-data-predict-pathankot-attack-subhash-chandra?trk=pulse-det-nav_art
Jan 5, 2016
Many watchers had predicted a Pathankot kind of attack and so when the attack happened on the 1st of January, this was an 'I told you so' moment for many of these watchers. In a small piece of secondary research done by me and Mihil Bhansali, we attempt to identify the data points that could have predicted a terror attack in Pathankot. 
The fact is that on the basis of a variety of data points/ intelligence, the Govt was able to forewarn the likelihood of such an attack. This led to a large contingent to NSG commandos being transported to Pathankot the night before the attack. In our approach, we used publicly available data (1999-2015) to see if this attack or an attack in general could have been predicted. 

Hypothesis: Likelihood of a terror attack was very high
Result: Yes. Every single meeting between a Indian PM from the BJP and the Pakistani PM/President was followed by a dramatic terror attack in India 

Hypothesis: The Pathankot attack is due to the recent meeting between PM Modi and PM Nawaz Shariff
Result: Past data is unable to link this well. Every terror attack that followed a meeting between a Indian PM from the BJP and a Pakistani leader occurred a minimum 15 days after the meeting between the leaders. This one occurred within 7 days. In fact, if one were to include the hijacking of the Police Jeep, the timeframe would reduce to 5 days. This is not to suggest there is no linkage but the fact that this is unusual when compared with past occurences

Hypothesis: The attack in Pathankot was quite predictable given the fact the previous attack occurred 40 kms away in Gurdaspur. 
Result: No such geographic linkage can be deciphered from past data. Attacks have occurred over wide distances

Hypothesis: Counter missions inside Pakistan will teach the Pakistanis a lesson and will reduce terror attacks in India
Result: The correlation between increased terror attacks in Pakistan and reduced attacks in India is quite low. Instead, international intervention (2001, 2008) has contributed to reduced attacks in subsequent years. 

Summary

1. It is fairly clear that publicly availably data can predict attacks by terrorists based out of Pakistan. However, the predictions can be precise in terms of date and location only with data available from other resources including human and electronics intelligence. Our analysis of publicly available data has yielded more predictions that we have not included in this analysis for ethical reasons.

2. It is advisable for the Govt to focus on strengthening security inside the country in the anticipation of more attacks instead of contemplating macho counter attacks inside Pakistan. Increase in terror attacks in Pakistan have had little impact on reduction in terror attacks in India. Further, we would strongly urge the Govt not to comment on likely future talks with Pakistan at this moment

3. The data suggests a very difficult situation for the Pakistanis - Dramatic terror attacks in India means serious international involvement which has only forced Pakistan to act on terror sponsors within the country thereby reducing terror attacks in India. Low grade terror attacks by terrorists from Pakistan in India would mean that India gets the peace it wants and therefore no real desire to settle Kashmir in a manner favorable to the Pakistanis. It is fairly clear that the Pakistani army must revisit its Strategic thinking process with respect to Kashmir given 50 years of failed efforts (starting from 1965 war). 

PS: Thanks to Mihil Bhansali for the detailed research on terror attacks and Summit meetings

No comments: