As pointed out by Hummelbrunner, the Logical Framework Approach (LFA) was initially designed as a military planning approach, for a context of strong central control and sharply defined goals. USAID adapted the LFA to the development context in the late 1960s. Since then, it has been adopted in development planning and monitoring throughout the world, to attain a near-monopoly in development planning. The LFA has its merits, as it helps to conceptualise interventions in a uniform, formalised way. Logical frameworks ("logframes"), if well-designed, can provide a convenient, simple overview of the main features of an intervention.
The problem is, "[the LFA] reflects a management style which demands precisely structured and quantifiable objectives ("management by objectives"), assuming that the actors dispose of all relevant information and operate in rather stable environments. The focus is on the delivery of activities and outputs, and on the achievement of intended effects through intended routes." In development, you are likely to deal with multiple stakeholders who hold different opinions and interests, in fluctuating situations which may require more flexibility than the LFA can accommodate. Hummelbrunner introduces the terms "logic-less frame", "lack-frame" and "lock-frame" to illustrate the dilemmas of applying a simple tool to a complex reality. "Logic-less frames" are invented to justify an existing project design, e.g. when institutional donor funding is sought for an on-going activity. The "lack-frame" over-simplifies, omitting vital aspects of a project as it ties activities and objectives into a linear cause-to-effect chain. Finally, the "lock-frame" blocks learning and adaptation to new opportunities and risks. Power imbalances between unequal "partners" in development programmes exacerbate these difficulties. Hummelbrunner presents his observations from a real-life case, an EU programme in Eastern Europe: "tunnel vision" dissociating projects from their contexts, "mechanistic" styles of implementation which focus on conformity to pre-established plans, and lack of involvement of local actors, local assets and expertise are listed as common problems associated with excessive reliance on the LFA. In monitoring and evaluation, the LFA is of limited value, as "tunnel vision" and the over-emphasis on compliance with pre-defined plans fail to capture the complexity of development processes. "People operate with a much higher level of complexity than can possibly be included in a logframe, so the neat logic does not work in reality."
Hummelbrunner shows how variations of the LFA can palliate some of these deficiencies: Project Cycle Management (PCA) offers more opportunities for participation and adjustment - provided its LFA elements are not over-emphasised; Social Network Analysis (SNA) captures complex relationships that linear stage models such as the LFA cannot do justice to; and Outcome Mapping (OM) focuses attention on changes in behaviours, relationships, actions and activities of individuals and groups, as seen from different perspectives and taking into account the contributions of other actors.
Systemic Project Management offers a total departure from the LFA, picturing development work as intervening in a constellation of mutually interlocking systems and sub-systems with their own processes, which are more or less directly related to producing (or inhibiting) the desired outputs and outcomes. In contrast to the centralised control system underlying the LFA, Hummelbrunner proposes three level of systemic management: normative management, i.e. policy that defines the vision guiding the project; strategic management by supervisory entities, e.g. steering committees, which decide on the effective use of resources ("to do the right things"); and operative level management ("to do things right") carried out through decentralised coordination teams or service providers. No pyramidal hierarchy of command and control is needed; systemic project management relies on the self-organising capacity of the sub-systems.
All these different options deserve to be explored more systematically - as the author concludes: "A frame exists for a more 'logical' use of logframe - provided this frame is noticed and used."