Working Paper no. 70 - Cities and Fragile States - THE OPEN CITY: SOCIAL NETWORKS AND VIOLENCE IN KARACHI Azmat Ali Budhani, Haris Gazdar, Sobia Ahmad Kaker, Hussain Bux Mallah Collective for Social Science Research March 2010 Crisis States Working Papers Series No.2 ISSN 1749-1797 (print) ISSN 1749-1800 (online) Copyright © A.Ali Budhani, H. Gazdar, S. Ahmad Kaker, H. Bux Mallah, 2010 24 Crisis States Research Centre The Open City: Social Networks and Violence in Karachi Azmat Ali Budhani, Haris Gazdar, Sobia Ahmad Kaker and Hussain Bux Mallah Collective for Social Science Research, Karachi Karachi is not only the largest metropolis of Pakistan and its commercial hub, it is also known as a ‘mini-Pakistan’. This is a reference to the ethnic and religious diversity of Karachi’s population. It has been a city of migrants for as long ago as anyone cares to remember. Until the 1980s it also shared, along with Lahore, the status of the political pulse of the country, or at least of the urban parts of Pakistan. To succeed nationally, movements had to make a mark on the city, and currents that emerged in Karachi frequently influenced the national mainstream. The very features of Pakistani society that are represented so prominently in Karachi are the ones that are often thought to challenge the coherence and stability of the nation state. Foremost, of course, are ethnic and religious sectarian heterogeneity. But there is also political fragmentation, economic disparity, demographic pressures, steady erosion of the state’s institutional capacity and the heavy footprint of international conflict. This paper is based on the premise that even as Karachi’s politics were cut off from the national mainstream, the city became a vantage point on the crisis of the nation state in Pakistan. In fact, the city’s exceptionalism since the mid-1980s could be seen as part of a wider process of political disarticulation in Pakistan: a disarticulation that at times threatens the basic make- up of the state. Yet the state has managed to pull back from the brink on several occasions, and so has the city. A closer look at the city and its own ‘experiments’ with the limits of institutional elasticity holds lessons for an understanding of the precarious resilience that has characterised the nation state itself in Pakistan. The aim of this paper is to provide a perspective on institutional breakdown in Karachi, using the prevalence of conflict and violence – particularly of the civil variety – as an index for this. The paper begins with a description of the conflict and violence that has become associated with Karachi. Trends and patterns in conflict and non-state violence are identified here, as outcomes that require explanation. The second section provides relevant background information on the city’s placement within the national polity and economy. Some of the main explanations for Karachi’s trends and patterns of conflict and violence − relating to ethnic identity, organisations, economic frustration, and political conspiracies − are reviewed in the third section. The role of external factors such as migration and the influences of the war in Afghanistan are also highlighted. An alternative approach focusing on two incontrovertible aspects of Karachi’s urbanity – migration and informality − is outlined in the fourth section. It is proposed here that the analysis of conflict and violence in Karachi can benefit from an understanding of the very processes that made Karachi an open city in the first place. The high rate of migration was correlated with an erosion of the formal sector’s capacity for supplying basic infrastructure or regulating the use of existing infrastructure. This resulted in the rapid growth of an informal sector for the provision of residential land, basic public utilities, transport and contract enforcement. The informalisation of public 1 provisioning, which was often aided and abetted by state organisations, was premised on two institutional deviations: first, the large-scale legitimisation of private non-state arrangements for contract enforcement; and second, the strengthening of existing or nascent social networks based on bonds of family, ethnicity and religious and sectarian identity. Qualitative accounts of the histories of land use in six very different localities of the city are interpreted using this alternative approach. Finally, it is argued that this alternative approach offers a way of understanding not just the breakdown but also recovery or the prospects of a return to political negotiation. Conflict and violence Current patterns of non-state violence Karachi is a busy city that provides sustenance to its 15 million residents through industry, trade, commerce, services and charity. It is also a violent city – probably the most violent city of comparable size in South Asia. In 2001, the murder rate in Karachi was 4.04 per 100,000 people compared with 1.67 in Mumbai and 3.79 in Delhi (CPLC 2009). Between 1994 and 2004, there were 8,816 casualties reported through incidents of violence including murders, torture by non-state actors, kidnappings for ransom, vigilante reprisals, bombings and suicide attacks.1 There are estimated to be a significant number of lethal weapons, many of them unlicensed and illegally held, in private hands in Karachi.2 These statistics, however, do not fully convey many of the paradoxes of Karachi’s non-state violence. The capacity for violent disruption that lurks just beneath the surface manifests itself at rare instances when the city is brought to a grinding halt by one of it several claimants. Otherwise it is business as usual, with traffic jams, bustling markets, late night shopping and dining, drives by the sea front, and an infectious carefree mood. There are few ‘no-go’ areas in the city, and hardly any neighbourhoods that are normally ‘closed’ to the entry of ethnic or religious outsiders. Yet it takes only a rumour for the shutters to go down, and for any locality to turn inwards on those occasions when the situation becomes abnormal. There have been days when pitched battles have raged between armed groups – divided along party-political, ethnic or gang lines – throwing up temporary fronts across otherwise invisible and well-traversed boundaries. Those days are not common, but everyone seems to know that they could come at short notice. Specifically, there are currently several identifiable proximate sources of non-state violence in Karachi. First, a number of political parties operating in the city maintain armed cadres who are organised and trained in the use of lethal weapons. Inter-party violence has tended to flare up around particular events such as public meetings, rallies and elections. There are also underlying tensions between political groups that are blamed for targeted killings of rival members and armed cadres. These groups are discussed in greater detail below, but it needs to be noted that their activities are not directly linked to jihadi forms of religious extremism. Second, the use of firearms is common in a range of crimes including robberies, carjackings, vehicle thefts, burglaries, drugs trafficking, protection rackets and kidnappings for ransom. It is presumed that all of these types of crimes involve some level of organisation and networks. There are allegedly close connections between criminal and party-political networks. In some 1 Authors’ estimates based on annual reports of the Human Rights Commission of Pakistan (HRCP 1994-2004). 2 Between 1993 and 1996, firearms were the agent of injury/death for 3,392 victims (Chotani et al. 2002) 2 parts of the metropolis, such as the old city quarter of Lyari, rival gang leaders have had relations with various party factions. In other cases it is alleged that the local armed cadres of some parties are directly involved in organising crimes and protecting petty criminals. Some conspicuous cases of kidnapping for ransom have been linked with powerful rural patrons. Third, there have been acts of violence that are thought to be linked to jihad and religious extremism.3 Most of the known groups connected with jihadi violence belong to the Sunni Deobandi sect – even though there are many Sunni Deobandi clerics and organisations that do not support jihadi violence within Pakistan. There have been several suicide bomb attacks on local and foreign targets that have claimed the lives of many civilians since 2001. Many of the now-proscribed jihadi groups that have been declared terrorist organisations have had a presence in Karachi (Newsline, August 2003). It is alleged that the US journalist Daniel Pearl was abducted and murdered by an Al-Qaeda cell operating in the city, and it is possible that there are other such cells that remain hidden. Before 2001 the main public activity of the religious extremists groups in the city was to raise funds and to recruit members through the many mosques and seminaries. These groups were also alleged to have carried out bombing and assassination campaigns against the Shia Muslim community that is present in large numbers in Karachi (Newsline, July 2002). A brief chronology There has been a steady escalation in violence in Karachi – led by political violence – since the early 1970s. Up until then political, student and trade-union activity was largely peaceful and violence was limited to isolated brawls that rarely led to a casualty. The main source of violence during such movements was the police, and on rare occasions the military that were called in to quell agitation. Sectarian religious violence between Sunni and Shia mobs did take place occasionally, but this posed a security challenge only on particular days in the year when religious marches of rival sects happened to cross each others’ paths. Much earlier in the late 1940s the city had witnessed communal violence against the minority Hindu community, most of whose members were forced to flee to India (Khuhro 1998). The first political movement during which non-state actors – mostly mobs of youths – used any significant amount of violence was in the early 1970s when the newly-elected provincial assembly of Sindh passed a law requiring the teaching of the Sindhi language in all educational institutes (Pakistan Forum 1972). This was seen by many members of the Urdu- speaking majority of Karachi as an assault on their cultural and political position in the province and the city (Rashid and Shaheed 1993). There were cases of ethnic violence as well as other acts of mob violence, but the situation was quickly brought under control through enforcement and political negotiation (Rahman 1995). Sustained mobilisation followed elections in 1977, which the opposition claimed were rigged by the government of the populist Prime Minister Zulfikar Ali Bhutto. As the government responded with repression the movement in Karachi turned violent with mob attacks, arson and targeting of the government’s political sympathisers and supporters (Weinbaum 1977). The PNA movement was, arguably, the last major movement in which Karachi was part of, or even led, the national mainstream. In Karachi the PNA movement (named after the opposition Pakistan National Alliance) was headed by right-wing Islamic religious parties that 3 The word jihad literally means struggle in Arabic. In Pakistan is has come to connote combat in the cause of Islam. It is in this sense that the term jihad is used here. 3 had become implacable foes of Zulfikar Ali Bhutto’s left-of-centre Pakistan People’s Party (PPP). The PNA movement became the pretext for a military coup led by General Zia-ul-Haq, and he imposed a highly repressive martial law regime that banned political activity. General Zia’s regime faced considerable resistance and he made alliances with religious groups to implement Islamic Sharia law. In Karachi, where opposition to Zia had considerable support on college campuses, pro-Zia Islamist groups began to receive weapons and training from state agencies. This was the beginning of the war between Soviet forces in Afghanistan and their mujahideen opponents. The Pakistan military’s support for the Afghan jihad had two immediate implications for Karachi. The Islamist groups, particularly the student-based Islami Jamiat-e-Tulaba (IJT), became militarised due to their involvement with the Afghan mujahideen, and there was a proliferation of small arms in private hands as weapons destined for the Afghan jihad found their way to unregulated arms markets (Tully 1995). Campus opposition to General Zia and the IJT in Karachi took several forms. Besides leftists groups – such as the various factions of the National Students Federation (NSF) – there were a number of ethnically based groups, some of them also with leftist rhetoric, among the Baloch, Pashtun Sindhi and other ethnic minorities. These latter groups saw the Zia regime as a manifestation of Punjabi ethnic domination in the country. Their connections with the rural hinterland, particularly in the Pashtun tribal areas, allowed the ethnic groups to also acquire weapons in order to stand up to the IJT on the campuses. The formation of an ethnically based group claiming to represent the interests of Karachi’s Urdu-speaking Muhajir community completed the picture.4 The All Pakistan Muhajir Students’ Organisation (APMSO), which was to give birth to the Muhajir Quami Movement (MQM) or the Muhajir national movement, was founded as a rival to the IJT. Although IJT was ethnically diverse, its parent party the Jamaat-e-Islami had a strong reserve of support in the Urdu-speaking population of Karachi. Many of the founders of the APMSO were disaffected IJT members, who wanted to replace Islamic solidarity with a call to the distinct ethnic identity of the Muhajirs. The APMSO quickly challenged the IJT and seriously undermined the latter by weaning away many of its potential supporters among the Muhajirs. It also followed in the pattern set by the IJT and ethnic student federations in acquiring arms. Campus politics had been steadily militarised, and by the mid-1980s it was virtually impossible for unarmed student groups to physically survive in the city’s colleges and universities. The emergence of the MQM from the APMSO meant that the politics of armed cadres left the confines of the campuses and entered the communities. The party had disproportionate support among the Muhajir youth, many of whom were attracted to its militant and macho reputation. The idea that the Muhajirs formed a distinct ethnic group – like the Sindhis, Punjabis, Balochis and the Pashtuns – became a rallying point. The organisation took full advantage of simmering ethnic resentments among the Muhajirs – many of whom felt disenfranchised in ‘their own’ city. In a series of violent encounters with other ethnic groups, particularly the Pashtuns, the MQM established a reputation for militancy (Gayer 2003). 4 The word ‘muhajir’ literally means migrant or refugee, and in the context of Karachi refers to Muslim migrants from India who arrived in the city following the bloody partition of the country at the time of independence in 1947. Latterly, descendents of the original migrants also began to class themselves as muhajir and argued that they formed a distinct ethnic group. The term evokes the Islamic tradition of hijrat, or migration in the face of religious persecution. . 4 The first major ethnic confrontations took place over seemingly innocuous issues. A hit and run incident involving a Pashtun-owned bus and a Muhajir schoolgirl led to widespread riots in which large numbers of buses were set alight. Another incident was sparked off by alleged sniper fire on an MQM procession from an Afghan refugee neighbourhood. This also led to large-scale rioting in which many people lost their lives and homes were destroyed. The next to flare up was the working class neighbourhood of Orangi where Muhajirs and Pashtuns lived in close proximity. Automatic weapons were used by both sides, and there were a number of horrific incidents of people being burnt alive in their homes (Karim 1995). These events marked a qualitative shift in the nature and scale of violence in Karachi: during 1985, the Karachi police recorded 608 cases of rioting which claimed 56 lives (Richards 2007). By 1987 the MQM was ready to contest local elections, and in the national elections that followed in 1988 it emerged as an overwhelming victor in Karachi. Political representation did not lead to an abatement of political violence. On the eve of the 1988 elections there was a massacre of Urdu-speaking Muhajirs in a series of random drive-by shootings in the city of Hyderabad. Extreme Sindhi ethnic nationalist groups were widely suspected of carrying out this crime. There was virtually immediate retaliation against Sindhis in Karachi for which the MQM was blamed. Ethnic tensions between Sindhis and Muhajirs remained high even after the elections, despite the fact that the MQM and the Sindhi-dominated PPP were coalition partners. The MQM left the coalition in 1989 and this was followed by a period of unrelenting political and ethnic violence in Karachi. A new factor in this phase of violence was the abduction of political and ethnic rivals. Parts of the city started to become ‘no-go’ areas for the police and security services, and it was reported that parties operated detention centres where victims would be tortured and even executed (HRCP 1992). From 1990 onwards there were sporadic attempts on the part of the security services to confront armed ethnic groups in Karachi, particularly the MQM. Police action was launched in 1990 but was aborted after the military backed away (Ziring 1991). Another operation, this time led by the military itself, was started in 1992 under the cover of a bloody split in the ranks of the MQM, but this too got bogged down, apparently due to the lack of good intelligence. From 1993 till 1996 there was an intelligence-led campaign with cooperation between police, paramilitary forces and the army to seriously disable the military capacity of the MQM. The state’s action led to mass arrests of MQM members, supporters and their family members. Large number of youths (estimates range between several hundreds and several thousands) were killed in actual or faked encounters with the security services (HRCP 1992-2007). Many young men escaped and took refuge in rural areas – paradoxically among ethnic communities that had themselves been targets of ethnic violence in Karachi. Others fled abroad to the US, UK, South Africa and Malaysia. MQM resistance was also fierce. During the early period of the security operations there were retaliatory attacks on state-security personnel and on other ethnic groups. It was common to find trussed up dead bodies in gunny sacks with torture marks. Many of these were alleged police informers (HRCP 1996). The party also retained the ability during that time of calling and enforcing general strikes in the city causing serious economic disruption and losses. Many parts of the city witnessed internal displacement as people from the ‘wrong’ ethnic group were forced to shift to safer neighbourhoods (Khan 2003). It is estimated that hundreds of families of MQM members and supporters – especially those families who had lost a member in a police encounter – were forced to go into hiding. They were unable to return to their homes because of fears of police harassment. There were also 5 demographic shifts within the city as people moved to neighbourhoods in the southern districts of the city that were relatively less disrupted by strikes and riots. It can be argued that the 1993 to 1996 period marked a peak in ethnic conflict and political violence of certain types in Karachi.5 Chottani, Razak and Luby (2002) estimate the rate of violent injuries to be 23 per 100,000, and the rate of violent deaths to be 13 per 100,000 between 1993 and 1996.6 Since that period the MQM and other ethnic parties have regrouped, rearmed and revived themselves. They have also displayed, very occasionally, the ability to repeat the excesses of the 1990s. The most notable such event was the violence on May 12, 2007, which is discussed below. But there are also important changes. The MQM, for example, formally changed its name from Muhajir Quami Movement to the Muttahida Quami Movement (United National Movement). Dropping the ‘Muhajir’ label was seen as an important act of symbolism because it allowed the party to claim to be a national rather than ethnic party (Richards 2007). The lowering of ethnic political violence since the late 1990s does not mean that such violence has disappeared altogether. There have been periods when inter-party rivalries have led to targeted assassinations of party cadres and supporters. The link between lower level party cadres and criminal elements has also thrived. Fighting between youth cadres of parties is sometimes indistinguishable from gang warfare. Even though ethnic violence might have abated for now, other forms of violence have emerged in recent years. Between 1994 and 2006, extremist Sunni militants waged a campaign of bomb attacks and assassinations against the city’s Shia minority (Abbas 2001). A number of mosques and congregations were bombed. There was a systematic campaign to target educated professionals within the Shia community – presumably in the expectation that this would lower the community’s morale and encourage its members to seek emigration (Korejo 2002). It is estimated that 26 doctors were assassinated.7 The recent pattern of sectarian violence is very different from the Shia-Sunni violence that occurred up until the 1980s, when rival religious processions might break out into mob attacks and riots. There is known to be much overlap between groups that target Shias and those who have undertaken or supported terrorist attacks on foreign and national targets since 2001. Targeted assassinations thought to be carried out by jihadi groups actually predate 9/11 and the US-led war against the Taliban and Al-Qaeda in Afghanistan. Militant Sunni organisations such as the Sipah-e-Sahaba Pakistan (SSP) and Lashkar-e-Jhangvi are known to have had close ideological and operational links with jihadi groups such as Jaish-e-Mohammad (Rana 2004). The war in Afghanistan and the Pakistan government’s support for the US-led coalition produced an angry response in some of the religious circles in Karachi. There were calls for recruitment for jihad and many young men – mostly but not exclusively from ethnic Pashtun localities – volunteered to fight alongside the Taliban. A new dimension was added to the pattern of violence in the city, with attacks on foreign and national high value targets. US 5 According to the authors’ calculations from HRCP annual reports, 4,642 people were killed in shootings, encounters and political violence between 1994 and 1996 (see Annex 1). 6 The authors also determine the nature of violence as being ethnic and politically motivated through gathering data from ambulance service logbooks. They found that the number of casualties and injuries increased during periods of strikes. Moreover, they discovered that casualties were disproportionately high (in contrast with population) in Korangi (22% of all homicides), Malir (8%), Nazimabad (8%) and Orangi (8%). Urdu speakers are a majority in Orangi and Nazimabad, while Malir and Korangi are multi-ethnic. 7 Authors calculations from HRCP annual reports 6 consular personnel and premises were targeted on several occasions by suicide bombers. There was a suicide bomb attack on French naval engineers that killed over a dozen people (Rana 2004). There were unsuccessful assassination attempts on important Pakistani officials, such as President Musharraf and the army commander of Karachi. There were also major bomb attacks on public rallies that claimed the lives of hundreds of people. The first was an explosion at a Barelvi Sunni religious gathering in April 2006 that eliminated the entire leadership of the Sunni Tehreek (Sunni Movement). This organisation claims to represent the Barelvi Sunnis who distinguish themselves from the Deobandi Sunnis. The former focus on ritual and devotion in contrast with the latter who are highly doctrinaire. While the Sunni Tehreek is a militant organisation – with armed cadres – it opposes and is opposed by the jihadi groups who are mostly Deobandi Sunni (Rana 2004). On the streets the Sunni Tehreek’s main rivalry, however, is not with the jihadi groups but with the nominally secular MQM. This is because many of the Sunni Tehreek cadres are former MQM members who split off or were expelled due to infighting (Shah 2003). The vendetta has been carried over into the new organisation. There were no claims of responsibility on the Sunni Tehreek rally, but there was a war of words between the MQM and the supporters of jihadi groups, with each shifting the blame on the other. In October 2007 the public reception for Benazir Bhutto’s return to Pakistan from her ten- year exile was the target of a major bomb attack. The attack on Bhutto’s rally, which she survived, killed over 130 people and maimed several hundred others. The attack was blamed variously on jihadi groups, Al-Qaeda, Taliban and elements within Pakistan’s security apparatus opposed to her policies (Dawn News, October 19, 2007). In this case too there was no admission of responsibility. After a lull of some two years, a Shia religious processions were bombed in December 2009 and February 2010, claiming over sixty lives. These attacks on Shia Muslims were claimed by spokespersons for the Pakistani Taliban. Summary of trends and patterns It is useful to summarise the main trends and patterns of conflict and violence in Karachi. The above review has shown that the mid-1980s marked a turning point both in terms of the extent and the nature of political violence. Whereas before, political violence had been exceptional and connected to specific policies and actions, now it became widespread and systemic. Political organisations and entire communities became militarised. Ethnicity was a salient line of division, but so was party-political and religious affiliation. There was a blurring of boundaries between political cadres and protection rackets, with the consequence that criminal and political violence became indistinguishable. Organisations acquired the capacity to enforce city-wide shut-downs and strikes, and state-security agencies lost their overwhelming superiority in the exercise of coercive power. There are inter-connections between the three broad sources of violence identified above: ethnic/party-political conflict, criminal violence and the jihadi threat. But there are also distinctive features. The jihadi violence has recently consisted of sectarian or high profile targeted assassinations, and bomb attacks including suicide bombs. Ethnic/party-political conflicts sometimes take the form of targeted killings but do not involve bomb attacks. These conflicts can and do involve ‘mass’ actions such as shut-downs and strikes, mob violence and even territorial battles. The jihadi violence does not include ‘mass’ or open action, but it can include sporadic attempts at enforcing Taliban-style restrictions on music and dress codes in selected pockets of the city. 7 This paper takes that view that although jihadi violence is a more urgent threat to state security and stability, ethnic/party political and criminal violence, which are inter-connected, are more important sources of insecurity and the danger of breakdown in Karachi. The arming of the city, the emergence of identity-based politics, and the weakening of state institutions are key shifts since the mid-1980s. Some of these conditions are the same ones that also make Karachi vulnerable to jihadi violence. Recent city-wide violence This chronological review will be incomplete without reference to two recent city-wide acts of political violence that revealed the geographical contours of the emerging contest in Karachi. The first was what has come to be known simply as ‘May 12’. There was widespread fighting on that day in 2007 when rallies from different parts of the city heading to the airport to receive the deposed chief justice of the country came under attack. The ensuing violence claimed over fifty lives in different parts of the city and left many people injured. There was a virtual shutdown for three days, and a serious danger of the situation escalating into an open ethnic conflict. The second instance was the reaction in Karachi to the news of the assassination of Benazir Bhutto in Rawalpindi in northern Pakistan: widespread rioting engulfed the city and there were several deaths; while about thirteen banks were burnt or robbed, and fifteen factories, ten petrol pumps and more than seven hundred vehicles torched (The News, December 30, 2007). The then-ruling MQM was widely held responsible for the May 12 events. The party had clearly stated its opposition to the deposed chief justice’s visit to Karachi, and to the lawyer’s movement against the Musharraf regime in general. Opposition parties which then included the PPP and the ethnic Pashtun Awami National Party (ANP) were supporting the lawyer’s movement, and planned to turn out in large numbers to greet the defiant judge. The night before the judge’s visit MQM cadres were seen setting up road blocks and barricades shutting off access to the proposed route of the chief justice’s reception processions. The city had been divided geographically between a central corridor along a north-south axis, and the periphery. On the day itself all state-security forces – police, paramilitaries and army – were conspicuous by their absence. At around midday, the time when the chief justice’s plane landed, there were sniper attacks on opposition processions that been stopped at barricades at different points in the city. Many opposition supporters were also armed and there were pitched gun battles across the city. The MQM managed to stop the chief justice’s welcome, thus forcing him to return to Islamabad after waiting within the airport for several hours. Ambushes and attacks on processions were well-planned and eyewitness accounts suggest that many of the attackers organised into military-style units acting in concert. In addition to sniping and gun battles there were a number of instances of abduction, many of which led to execution-style killings (HRCP 2007b). Although the MQM had an upper hand in terms of scale and organisation of combat, the opposition also managed to put up sustained resistance on the streets. The violence quickly spread from the processions to neighbourhoods along a notional front-line that roughly corresponded with the boundary between central and outer districts around the city. MQM offices were set alight in the outer districts as well as in cities across the country. Benazir Bhutto’s assassination on December 27, 2007 provoked an angry response in Karachi and other cities of Sindh. Gangs of youth, many of them armed, spread out from working 8 class neighbourhoods within the city and in the outlying settlements and there was widespread rioting and arson (The News, December 30, 2007). On this occasion, too, the state-security forces were withdrawn from the streets. Petrol pumps were early targets and they remained shut for several days, thus paralysing traffic in the city. Although many of the central precincts of the city were also affected by rioting – and thus remained closed for several days – there was a sense of a centre-periphery geographic division. All of the main entry and exit routes serving the city were effectively blockaded, including routes leading to the harbour area. The differences between the two recent city-wide acts of violence are obvious. The first was planned, involved direct confrontation between armed groups, and resulted in high levels of casualties but relatively little damage to property. The second, by contrast, was spontaneous, and there was extensive looting and damage to property. The similarities, however, are instructive. In both cases the main protagonists, at least on the surface, were political parties. State-security forces were absent or withdrew, and order was quickly restored when they reappeared. There was a geographic division of the city between the centre and the periphery. On both occasions there was a danger that the violence would escalate along ethnic lines, but this danger was quickly ameliorated by political parties and leaders through public gestures of reconciliation. Finally, it is striking that the political backdrop of both cases of recent city-wide violence were events relating to the national mainstream. Karachi’s violent response in turn had an influence on national politics. The May 12 violence resulted from MQM’s decision not to allow the opposition movement – which had received public support in other parts of the country, notably north-central Punjab – to encroach into what the party regarded as its territory. The violence had the opposite effect. The resistance in Karachi and the casualties raised the pitch of the opposition movement and made the government’s position harder to defend. The post-Benazir violence was also a response to an event in national politics. The PPP had been a minor player in the city with its support concentrated in the rural areas and smaller towns of Sindh and Punjab. The scale and spread of violence suggested that something might have changed. Karachi was back in the national mainstream after decades, and now in its own violent manner. Pulling away from the brink? The remainder of this paper is not about conflict and violence, but about understanding the conditions that may have led to a qualitative change in Karachi’s security environment and its isolation from the national mainstream in the mid-1980s. Karachi’s experience suggests another paradox, and thus a potential source of insight. There are a number of instances – some drawn-out and others rapid – when a clear danger of escalation of violence and breakdown has been reversed. The ten-year period from the late 1980s to the late 1990s saw dramatic changes in the fortunes and military capacity of the MQM; and there have been other periods when criminal violence has peaked and then declined. The two recent cases of city-wide violence offer dramatic windows on instances when escalation was seen as a palpable threat and the situation was rapidly brought under control. Any understanding of Karachi’s drift into conflict and non-state violence, therefore, also needs to be able to explain the tendency thus far of pulling back from the brink. 9