28 February 2002: Dawn. Munaf and Amjad, Muslim shop-owners of Kabadi Market-2 in Ahmedabad awoke to horrifying news. A train had been burned in Godhra town the day before, ostensibly by a Muslim crowd, and ‘40 to 50’ Hindu passengers were thought to be killed in the fire. The two shop-owners had witnessed violence in 1969 and 1985 and knew that this was a trigger of astounding magnitude. Wary of reprisals, Munaf, along with other shop-owners of Market-2, hastily visited their shops around 8 am, instructed the three Muslim watchmen there to keep guard from within the market's colossal iron gates, and left for their homes. At 9 am, a crowd of Hindu attackers killed two of the watchmen, and looted the shops. At 10 am, the second market, Kabadi Market-3—half a kilometer away—was targeted in an equally potent attack. Puzzlingly, the outcome of the attack made on the second market, which bore goods worth three times more than the first, turned out to be different—no killings or looting occurred there.

Within the same electoral constituency, what explains different levels of violence across streets and neighbourhoods? When I first started fieldwork in neighbourhoods of Ahmedabad in August 2010, my primary objective was to test the hypothesis of political incentives. This was the most obvious next step following from Chapter 3 and Chapter 4. It was necessary to explain why patterns of violence should vary at the micro level when political incentives and politicized networks of attackers prevailed across an entire constituency. This is where the behaviour of people became the focal point of the explanation.