02
Markus (1983) - Power, Politics, and MIS Implementation
SourceMarkus, M. L. (1983). Power, Politics, and MIS Implementation. Communications of the ACM, 26(6), 430–444.
About the article
The article analyzes two implementation cases, a financial information system (FIS) and a logistics/inventory system, and draws its insights and struggles from comparing them.
Brainwaves
Where this took my thinking.
- 43 years later, what did we learn?
- Are almost all information systems created to be in control? True for CRM, ERP, HR software like Workday. But is it true for all technology implementations? Where does AI fit?
- What about designing for full transparency for employees? Not data for managers only: anyone in the organization sees the same numbers. Maybe impossible for an HR system (or maybe not), but certainly conceivable for CRM or AI tool usage data.
- Company culture defines the fear: what have employees seen leaders do with additional data? That history, more than the system, decides how much fear and resistance there will be. The fears underneath: being controlled, being held responsible for something you cannot control, repercussions. Links back to the full-transparency choice.
- All implementations and transformations focus on the functional layer (system, usability, functionalities), while one level up we are all human beings, and it is all about the emotions surrounding the change.
- The AI context: AI is in essence a personal decision-support system. So you probably need the power perspective for rolling out AI across an organization, and a learning strategy for individual adoption. (Builds directly on the Ginzberg/Keen distinction.)
- Add a power-analysis variant to AI adoption work.
- Any adoption analysis needs: the power between groups before/without the system, and the power in the system's design. And distinguish actual loss of power from perceived loss of power.
- Technical and user-oriented explanations of resistance are hygiene factors, not the story.
Follow-ups
Queued from this note. Nothing here is in the library yet, deliberately.
📚 Deep-dive candidates
Possible future library additions, not added yet.
- openGinzberg, M. J. (1974). A Detailed Look at Implementation Research. Report CISR-4, Center for Information Systems Research, Sloan School of Management, MIT. Free on MIT DSpace.
- openThe references in Markus worth mining later.
🤖 AI deep-dive sessions to do with Finn
- openWhich companies actually implemented full internal data transparency? Verified starting point: Vineet Nayar, Employees First, Customers Second (Harvard Business Press, 2010), on his tenure as CEO of HCL Technologies: financial data of the company and its business units opened to all employees, and managers' 360-degree feedback (Nayar's own first, in 2005) published on the intranet for everyone to see. Find more examples.
- openHow are autonomy, power, agency, and influence related? All play a role here; what are the nuances between them?
What I marked in the text
Close paraphrases of the article, in reading order. Reference material.
- A political perspective on the implementation of management information systems in complex organizations.
- Resistance, defined as behavior intended to circumvent or redirect the system, connects to a loss of power.
- The alternative hypotheses are technical and user-oriented aspects. My note: those are hygiene factors.
- The balance of power view involves: the original balance of power among groups; the intention and motivation to gain power through the system; and political tactics around the process of implementing it, with the degree of resistance that follows.
- Most systems literature looks at outcomes of the system (degree of use, user satisfaction); the political perspective looks at outcomes for the organization: did the system achieve what we aimed for?
- Major determinants of power: dependence of others on the power holder; ability to provide resources; ability to cope with uncertainty; irreplaceability; ability to affect a decision-making process.
- The reason of being of management systems is to provide managers with useful information, presumably so managers can cope better with variances arising from their production technologies.
- Three power allocations to distinguish: the power distribution before the system; the power actually designed into the system; the organizational power after introduction.
- Forms of resistance: sabotaging the system; providing inaccurate data; not using it at all; keeping other sets of records; circumventing the intent while obeying the letter. This resistance can come from people very low in the hierarchy.
- People and organizations differ in how actively they seek power, but it is unlikely they will voluntarily give it up.
- Access to information is probably less important as a basis of power than the ability to control access to information, or to define what information is kept and manipulated in what ways.
- Distribution of data that makes a subunit's performance more visible is likely resisted by the exposed group and welcomed by those who want to influence them. If the "winner" sits vertically above in the hierarchy, resistance is much lower than when the winner is a peer. (Links back to the centralization vs decentralization debate.)
- Participation is a necessary but not sufficient condition for system success; technical quality and top management support are other necessary conditions, at least above threshold levels.
- The design team matters: departments outside the design team are far more likely to resist. Not every department has to be in the room; knowing that peers were part of the design already helps.
- Ginzberg and Keen: in an individual context (a personal decision-support tool) implementation is mostly about learning; in complex rollouts across very different groups it is mostly about power.
“Central to the political perspective are the intentions, motivations, and desires of key actors, users and designers alike.”
Markus (1983), verbatim