The Phung Hoang Management Information System: Vietnam, 1972.

Phung Hoang Management Information System Command Manual, March 1972

Extract from Phung Hoang Management Information System Command Manual, March 1972.

The Phoenix Program, or Phung Hoang, was a "pacification programme" developed by the CIA with the assistance of a joint civil-military organisation called MACV, developed to identify covert actors working with the Vietcong. Phung Hoang aggregated intelligence from civilian informants, Vietcong suspects detained in prisons, South Vietnamese Police investigations and other US-led pacification operations. Suspected Vietcong would be rounded up, interrogated, and as one CIA account puts it: "if they resisted, they were killed" (Finlayson, 2008). Most historical accounts position Phung Hoang as a chaotic and violent kill-or-capture regime that relied on compromised intelligence and driven by the need to continuously report positive results to the analysts back in Washington DC. The consequences of programme vary: Tim Weiner puts the number of Vietcong suspects killed at 20,000 "at a minimum" (2007: 394), while Seymour Hersh quotes official South Vietnamese government (GVN) statistics that put the figure at 41,000. In a macabre twist of logic, the real number killed may have been deliberately inflated by MACV in order to portray Phung Hoang as being more "successful" than it actually was—statistical misreporting was widespread in other Vietnam-era pacification programmes such as the Hamlet Evaluation System. Another thing these various programmes had in common was that, behind the sprawling violence, a supposedly rational procedure systematised operations to mitigate error, increase expediency and maximise impact on the Vietcong's political and paramilitary infrastructure.

The above image is taken from the Phung Hoang Management Information Subsystem (PHMIS) Command Manual (1972: 11), a document that gives some insight into the manners in which kill-or-capture programmes were bureaucratised. The manual is attributed to CORDS (Civil Operations and Revolutionary Development Support) as well as MACV (Military Assistance Command Vietnam), with assistance from the Computer Sciences Corporation—a company still offering "next generation IT solutions" today. The document abstract explains that "the manual describes the operation of the system, the input requirements, error correction procedures, and the various system outputs." (5) In brief, the system is fed with report cards for suspects, who are then assimilated into a database. Monthly reports are automatically produced, sent up the chain of command for analysis, and finally suspect lists are sent back down the chain to relevant divisions in Vietnam for what is euphemistically termed as "neutralisation". Timelines are clearly specified: Neutralisation reports "are to arrive in Saigon no later than the 5th day of the month following the month reported" (28).

System objectives are listed in an abridged form below (9):

  • Maintain files containing biographic data on identified members of the Vietcong Infrastructure (VCI).
  • Provide accurate statistical information on the neutralisation of the VCI.
  • Provide Phung Hoang with the capability to respond to inquiries by interrogating the database.

In these objectives we can see the formation of a regime of information which, through the practice of individuating Vietnamese populations and grouping them according to various personalised vectors, aspires to secure territories based on what are ostensibly blunt facts and objective data. Database headings such as birth place, birth year, sex, party membership, residence are filled out for each suspect. It is upon these factors that one person is distinguished from another, and on this tenuous basis that a suspect is confirmed VCI and thus sought out: "VCI suspect is confirmed. (three separate reports establishing his identity as a VCI)". (16)

Looking at the above flowchart, these objectives are demonstrated in a clear procedural logic. It is divided into the administrative territories of various organisations, their lengthy bureaucratic titles encoded as acronyms: District and Provincial Intelligence Operations and Coordination Centers (DIOCC/PIOCC); Phung Hoang Directorate (PHD); National Police Command Data Management Center (NPC/DMC); and finally, Military Assistance Command/Data Management Agency (MAC/DMA). Information travels through a sequence of input-output boxes, each box symbolising a task of auditing, reviewing, processing, updating, reporting, inquiring, managing, and so on. These task-boxes have different icons, indicating paper records, punchcards, actions, and reviews. Thick black arrow-lines indicate the intended movement of data through the system, while thinner lines demonstrate the various feedback loops and meta-processes. Dotted lines indicate error processing.

How the division of labour operates across the various institutions involved in the PHMIS is visible--although it is not clear whether each discrete task-box is work undertaken by a human or a computer. The manual specifies elsewhere that CPU time required on a monthly basis is under 60 minutes, with district-level advisors devoting eight hours per month on input reports and a four-person team in Saigon reviewing data full-time. (12) It is also specified that the PHMIS is written in COBOL, a programming language optimised for business applications developed by Grace Hopper, and runs on the IBM S/360, a pivotal machine in the history of computing for its modularity and standardised protocols (Cortada,2012: 66).

The accuracy of input data is vetted at each point of processing, audited at four separate points, and with multiple passes of "Error Correction": "The edit/correction cycle is repeated until all errors are corrected." (10) It's perhaps not difficult to imagine how a surveyor contracted to report on the PHMIS might be impressed with its gestures towards due diligence: If the data is being so thoroughly audited, if it has to pass through some many "gates" of analysis and vetting, perhaps one might be lead to the conclusion that mistakes are minimal? The pristine clarity of the flowchart does not necessarily play out in practice, however. Robert Komer, the CORDS chief who but two years previous lauded the use of computer systems to expedite and simplify intelligence collection and queries ("Impact of Pacification"), states the following in a draft memo titled "The Phung Hoang Fiasco":

"Judging from the incredibly poor dossiers at most PIOCCs and DIOCCs I visited, there is all too little prior evidence available in most cases as to whether a man killed, captured, or rallied is really a VCI. Fingerprints are rarely used to establish identification and photos almost as little." (1970: 3)

The issue Komer raises hints at the complicated reality behind the surgical logic of the PHMIS flowchart. The problem with such visualisations of procedure is that they are necessarily a simplification, and as such they exclude. The "information flowchart" occupies a grey zone, an intermediary between two forms of violence: the schematisation of individuals according to a machine-processable database logic; and the subsequent categorisation of the database's population into those who are and those who are not subject to physical violence and/or internment. That this flowchart is intended to train the PHMIS operators in the use of the system and its processes means it is all the more impactful: in isolating it from the moments of its biopolitical force, the discrete administrative tasks are divorced from their violent implications.

In their continued practices, such systems take on a role of legitimising not only themselves, but the wider conflict they are situated within. They produce statistical cosmetics that efface the violence from war, acting as evidence that there is due process—rhetorical counterweights to moral objection. Blunt facts can be adopted as sharp instruments of political strategy, as we are especially aware of in these times. The lesson then is precisely to query the assumed causality of the flowchart: to see the lines between each action as processes in themselves; to deconstruct each constituent task-box into a recursive sequence of sub-tasks; to look outside its boundaries; to fracture its procedural logic in order to demonstrate its micropolitics.


  • Computer Science Corporation. [Web]
  • Cortada, James. "The Digital Flood: Diffusion of Information Technology Across the U.S, Europe, and Asia." Oxford University Press, 2012. Print.
  • Finlayson, Col Andrew R "A Retrospective on Counterinsurgency Operations." October 2008. [Web]
  • Komer, Robert. "The Phung Hoang Fiasco." July 30 1970. [Web]
  • Komer, Robert. "Impact of Pacification on insurgency in South Vietnam." Rand Corporation. 1970. [Web]
  • Hersh, Seymour. "Moving Targets". New Yorker, Dec 15 2003. [Web]
  • PHMIS Command Manual. MACV/CORDS. March 1972. [Web]
  • Weiner, Tim. "Legacy of Ashes: The History of the CIA". London: Penguin, 2007. Print.
  • Wisner. "Phoenix Goes Underground". [Web]