Analysis Of QPW Data


We most frequently use a natural force-based network analysis to analyze organization data collected in the Pasteur data base. This analysis produced an adjacency diagram. In these diagrams, a default repelling force is established between each pair of roles. There is also an attracting force between pairs of roles that are coupled to each other by collaboration or mutual interest; a stable placement occurs when these forces balance.

Here is the picture that results by applying this analysis to QPW:

BorlandNet.gif

Each rectangle represents a role. Each role is colored proportional to how much it is coupled to the rest of the organization as a whole. Roles are connected with lines that indicate the strength of interaction between the respective roles. The thick, yellow lines indicate strong interaction; medium red lines are moderate interaction; and thin, green lines are the weakest interaction. Roles are grouped so that the ones that interact most closely with each other are closest to each other on the diagram, while those with the least mutual coupling are the furthest from each other in the diagram.

There are several things worth noting in these pictures that set them apart from most other organizational process models we've made. Here is a summary of those properties:

  • The QPW process has a higher communication saturation than 89% of the processes we've looked at. The adjacency diagram shows that all roles have at least two strong connections to the organization as a whole. The project's interaction grid is dense. The coupling per role is in the highest 7% of all processes we have looked at. This is a small, intensely interactive organization. We find patterns like ThreeToSevenHelpersPerRole, CouplingDecreasesLatency, FewRoles and ProducerRoles in this structure.
  • There is a more even distribution of effort across roles than in most other processes we've looked at. The roles in the adjacency diagram are shaded according to their intensity of interaction with the rest of the organization. In the QPW process, Project Manger and QA glow brightly; Coders a little less so; Architect, Product Manager, and Beta Sites are "third magnitude stars"; and Tech Support, Documentation, and VP still show some illumination. Most "traditional" processes we've studied show a much higher concentration of interaction near the center of the process. That is, most other processes comprise more roles that are loosely coupled to the process than we find in QPW. That may be because QPW is self-contained, or because it is small. It may also be because the process was "intense": a high-energy development racing to hit an acceptable point on the market share curve. We see the patterns DistributeWorkEvenly and FewRoles.
  • Project Manager and Product Manager are tightly coupled, central roles in the process. These managerial roles were filled by individuals who were also key technical contributors to the project (they wrote real code), which contributed to their acceptance and success as process hubs.
  • Product Manager was a role that was employed only after a year of development.
  • Quality Assurance is a tightly coupled and central role. Many organizations consider QA to be an external function, outside their organization and process. At Borland, QA becomes a way of life after development has converged on a good design and a stable user interface. For QPW, this was about 12 months into development. Again, this is EngageQualityAssurance.
  • The CEO (Philippe) figures strongly in the organization. In a company of thousands of employees, it is unusual to find the CEO as tightly coupled to development as we find in QPW. It is instructive to examine the responsibilities associated with Philippe Kahn's role: Ensure product is commensurate with current market environment; ensure product market coordination is done in a timely and cost-effective manner; determine pricing, product positioning; shape public perceptions and handle PR for the product prior to and after ship; determine cosmetic changes to keep consistency among all Borland products and to call out certain features (in other words, usability testing); playing jazz to avoid press questioning on ship dates. This is a combination of PatronRole, LegendRole (Philippe Kahn was an icon of 1980s software culture), and FireWalls.
  • The overall interaction grid pattern is uncharacteristic of what is found in other processes:


BorlandGrid.gif



Interaction grids show patterns of interactions in an organizations, and are particularly useful when the organization is large or when its interactions are dense. We most often use an interaction grid where roles are ordered on both axes by their degree of coupling to the organization as a whole. The most integral roles are placed near the origin. Most other processes exhibit a characteristic pattern of points along the axes, with lower point density and lower intensity for increasing distances from either axis. In QPW, there is a general lessening of density and intensity as one moves toward the northeast quadrant of the interaction grid. The northwest and southeast quadrants of the Borland grid remain more dense than we've seen in other processes. This is a combination of DistributeWorkEvenly and ResponsibilitiesEngage.

Between 30% and half of the processes we've studied exhibit a pattern called schismogenesis ([BibRef-Bateson1958]; see also TheOpenClosedPrincipleOfTeams). To summarize, schismogenesis is a term from classic anthropological literature that describes a tendency for societies to stratify into sociological "comfort zones." This phenomenon appears in interaction grids as a clustering of points around the diagonal. For organizations where this phenomenon is present, the effect is particularly pronounced in the northeast quadrant of the interaction grid. It indicates that organizations contain splinter groups.

The QPW process is characteristically "anti-schismogenetic." That is, there is blank space around the diagonal of the interaction grid, particularly in the northeast quadrant. While we have seen graphs with random scatterings of points, the QPW graph is the first where the points seem to abhor the diagonal, yet fill out the rest of the graph.