November 29, 2009

Artificial intelligence

Srinath Raghavan

http://www.asianage.com/presentation/leftnavigation/opinion/opinion/artificial-intelligence.aspx

Nov.27 : On the first anniversary of the Mumbai terror attacks, the Indian security apparatus has come under some scrutiny. However, the ongoing discussions have overlooked a key component of this establishment - the intelligence agencies. This is surprising, for these agencies had drawn the ire of the media and pundits in the immediate aftermath of the attacks. Indeed, the attacks were attributed to an "intelligence failure". The agencies tried to fend off these accusations by a series of leaks to the press. The government naturally sought to reconcile these conflicting claims. The Union home minister stated that there had been problems of coordination between the numerous agencies and their subsidiaries, and that these gaps had been plugged.

To be sure, there were problems of coordination in the run up to the attacks. For instance, a crucial warning from the agencies was passed on to the Coast Guard; but the Navy and the Maharashtra police apparently did not receive this information. Nonetheless, in focusing excessively on issues of coordination, the government might be overlooking more deep-seated issues that need to be addressed as well.

It is now clear that the intelligence agencies did provide important inputs. Towards the end of September 2008, the Intelligence Bureau (IB) issued warnings that the Taj Mahal Hotel was among the high-profile targets shortlisted by the Lashker-e-Tayyaba (LeT). The Research & Analysis Wing (RAW) also gathered from communications intelligence that the LeT had reconnoitred several targets including the Leela Kempinski. On November 18, RAW intercepted a satellite phone conversation, which was traced to a location about 60 km off the coast of Karachi. Subsequently, the terrorists abandoned this vessel after hijacking an Indian fishing trawler.

Why, then, were the dots not joined? The problem was that these pieces of pointed information were part of a wider stream of more generic warnings through the year. This seems to have led to what might be called a "crying the wolf" syndrome. It is easy to accuse the Coast Guard, the Navy and the Maharashtra police for not taking the warnings seriously. But the last, especially, had a serious problem of capacity. It is equally easy to suggest that a "worst-case scenario" approach should have been adopted for every warning. Such an approach, however, would have required far more resources and caused much more inconvenience to the public. In fact, after the RAW warning on Leela Kempinski, hotels like the Oberoi did introduce restrictions but eased them just a week before the attacks. Notwithstanding all the beefing up of security forces over the past year, "a worst-case scenario" approach will remain unviable.

Furthermore, these warnings were issued by the agencies and were not accompanied by a threat analysis based on all available inputs. This problem stems in part from the fact that intelligence reports are usually inconclusive. By the time they become conclusive the event is already upon us. Reforming this system might be desirable; but it will also result in additional delays. Besides, increasing the number and frequency of intelligence assessments might be counterproductive. Top decision-makers seldom have the time to work their way through a pile of intelligence.

These innate problems of intelligence analysis, warning and action are not specific to India. These can be observed in many instances of intelligence failure across countries. Comparative studies also suggest that these are usually intractable. Organisational restructuring is certainly an inadequate remedy. Intelligence failures, then, may be unavoidable.

Yet there are some ways of mitigating these problems. The first step would be to identify and analyse recurrent patterns in intelligence failures. This would entail a comparative historical examination of these instances of failure. The agencies could conduct in-house studies, but also allow competent outsiders to give an informed assessment based on full access to records. The British intelligence agencies have been the pioneers in this regard. An "authorised history" of MI5 written by the Cambridge historian, Christopher Andrews, has recently been published. Similar histories of the MI6 and the Joint Intelligence Committee are due to be released next year. Such exercises would also give us reasonably good idea of the ratio of success to failure, and hence the "batting average" of different agencies. Apart from history, it is import to sensitise intelligence professionals and consumers to the methodological and cognitive pitfalls that usually accompany intelligence failures.

The Mumbai attacks have also underscored the need for a qualitative leap in our capacity for intelligence gathering. The recent investigations into the attacks by the Italian police demonstrate the sophistication of the threats that now confront us. The attackers and their handlers in Pakistan used a US-based Voice Over Internet Protocol (VOIP) for real time communication. The VOIP number was owned by a Belgian company, which in turn had leased it to an American telecommunications firm. The VOIP account was activated by money transfers through a franchisee of Western Union. And this transaction was carried out under a false name and identification by a duo based in the northern Italian city of Brescia.

Anticipating such threats will require an entirely different order of capabilities and skills. Given India's large base of talent in information technology and related areas, developing the necessary technological capabilities should not be too difficult. But our agencies also need to be able to attract people with the requisite skills. The existing policy of relying largely on Indian Police Service officers on deputation needs to be reconsidered. Like many of their Western counterparts, our agencies should be able to compete openly in the marketplace of colleges and universities for the best talent. They also need to acquire better area specialists - people who have a strong grounding in the language and culture, history and politics of different regions.

Finally, they should reach out to a wider community of specialists and experts. The Canadian intelligence, for instance, has a designated "academic outreach" programme, which regularly organises conferences. These may not be very pertinent for immediate security concerns. But they are useful for understanding the bigger, longer-term strategic picture.

All of this would require our agencies to be more transparent and open to outside influence. As the Director-General of MI5 rightly notes, such openness to the society is in itself a strategic advantage in confronting contemporary threats.

Srinath Raghavan is a Senior Fellow at the Centre for Policy Research, New Delhi

No comments: