A Systemic look at the Stockwell Shooting

By Dan Jenkins

An article from the ‘The Ergonomist’ August 2009

The fateful events in Stockwell, on the 22nd July 2005, need little introduction. A manhunt was on for the perpetrators of the previous day’s attempted bombings. A gym membership card, found with one of the failed devices, connected Hussain Osman and the address ‘21 Scotia Road’ to the attacks. An operation was mounted at the address to apprehend Osman as he left the flat’s communal entrance. At 09.33 hrs a man, allegedly bearing a resemblance to Osman, left the flat. Officers followed him on his 33-minute journey to Stockwell Tube Station. Two minutes after he entered the station, members of the Metropolitan Police Service’s (MPS) specialist firearms department (CO19), entered the underground station with orders to ‘stop’ a suspected suicide-bomber. Surveillance officers directed them towards the suspect. Moments later, two of the CO19 officers approached the man and between them fired seven shots into his head and one into his shoulder from close range. This man was later found to be Jean Charles de Menezes (JCdM), a completely innocent Brazilian national.

The organisation involved, the MPS, failed to balance its responsibilities of preserving the life of the public, its officers, and suspects. Whilst the situation can, unquestionably, be defined as complex, dynamic, and safety critical, procedures to tackle suicide bombers, defined by the codename Kratos, have been in existence since 2003. Thus, the pertinent question is how was this outcome allowed to happen.

The aim of this article, and the research behind it, is not to attribute blame to any individual or organisation, but rather, to explore the interdependencies between the numerous actions, omissions, and decisions that led to this tragic outcome. Those familiar with James Reason’s Swiss Cheese Model will be aware that latent conditions and unsafe acts can combine to produce the conditions for accidents. Whilst superficially the death of JCdM can be directly linked to the officer’s decisions to fire the fatal shots, further examination reveals that their actions were predated by a number of latent, error-causing conditions. The AcciMap model, shown on the below (click to enlarge), captures many of the findings of the Independent Police Complaints Commission’s (IPCC) report in a single representation, modeling their interdependencies and the causal flow.

Had JCdM been identified or challenged earlier, in a more controlled manner, his death could have been avoided [1]. The system crossed the boundary of safe operation as soon as a suspected suicide-bomber was allowed to board public transport. It is evident that the direction of the Gold Commander to stop all residents of the Scotia Road flat, briefed five-hours earlier, was not carried out [2]. Firearms officers were not in a position to challenge JCdM until he had boarded the train at Stockwell. Furthermore, the firearms officers arriving on the scene had received limited information on the unfolding events, had they been better integrated into the identification, they may have been better informed.

Failings have been identified at all levels of the model. For example, at the lowest level, the observer ‘Frank’, positioned in a van right outside the flat’s communal door, failed to capture an image of JCdM. His ‘call of nature’ preventing him from switching the video camera on [3]. Had Frank connected the camera to the van’s power source with cables available [4], or had there been a second observer in the van, vital information could have been captured that may have led to the identification of JCdM.

A number of factors clearly predated the operation. Various communication breakdowns are identified as key factors. A failure to equip officers with radios capable of working underground [5] (identified by the Kings Cross fire over 17 years before [6]) meant that CO19 officers had no contact with their superiors after entering the station. The organisation lacked a clear, unambiguous lexicon relating to orders and rules of engagement [7]. According to the IPCC report, the ‘stop’ command was not clear [8]. Also, the comment from the surveillance officer to the firearms officers, ‘There he is’, could also be misunderstood as, ‘There the bomber is’, rather than, ‘There the suspect is’ [9]. A mismatch in understanding between the different levels of command is also evident. Information from officers on the ground was consolidated without considering the negative identifications [10], undoubtedly affected the perception of those in the distributed command centre.

Besides collating the findings of the IPCC into a single graphical representation, the approach also questions the suitability of the MPS’s organisational structure to support rapid-paced operations, particularly where reliable identification of a suspect is not possible. The hierarchal division of functions between teams (surveillance officers collecting information, the distributed senior officers making decisions, and the firearms officers challenging suspects) relies upon the complex flow of information. A number of predefined organisational structures are proposed, the choice of structure informed dynamically by the operation type and unfolding events. Thus in situations where a hierarchical structure is unable to match the pace of the operation, a predefined structure can be implemented, allowing the same individuals to collect information, make a decision and challenge the suspect.