Attention is now turned to a set of assertions that express some of the lessons learned from the empirical studies. The presentation starts with two fundamental assertions about what characterizes enterprise modeling. The proceeding presentation is structured according to the two main phases of enterprise modeling development and dissemination. For each of the phases, assertions regarding objectives, process, artifacts, and actors are made.

7.2.1    Perceived challenges of enterprise modeling

The first assertion concerns the relationship between the modelers and the domain, and is fundamental in the sense that it influences many of the other assertions.
 
Assertion 1 

The perceived challenges of enterprise modeling are influenced by the relationship between the modelers and the type of work that is modeled: 

When the type of work that is modeled is well known to the modelers, enterprise modeling essentially becomes an act of representation. 

When the type of work that is modeled is unknown to the modelers, enterprise modeling essentially becomes an act of sense-making. 

By perceived challenges is meant how much effort the actors themselves perceive modeling to take, or how much effort it has been observed to require. By type of work is meant the four categories of the taxonomy outlined in section 3.6.

Concerning to which extent work is known to the modelers, observations from the four empirical studies suggest the following assertion:

 
Assertion 2 

Work dominated by replication risk and material artifacts is more likely to be well known than work dominated by design risk and information-based artifacts.

Consequently, assertion 1 and 2 combined suggests that the challenge of modeling RM type of work is more likely to be that of representation, while the challenge of modeling DI type of work is more likely to be sense-making.

To support the assertions, the following empirical evidence can be put forward:

Replication risk, material artifacts (RM): All four studies included modeling of the physical production chain. In VPT, representation of the production chain was explicitly stated to be the simplest to model. The same was contended in PA30, using the physical production chain as the backbone of the model. In Gazz, the type of work modeled was RM, and the modeling effort was more representation than sense-making. In TEK-S, M3, M4 and M7 represent RM type of work (see table 6.3), and none of the models were perceived as challenging to create or understand (there were no disagreements over the contents).

Replication risk, information-based artifacts (RI): Modeling of information-based work dominated by replication risk was observed in VPT, PA30 and TEK-S. However, both VPT and PA30 used the physical production chain as a backbone for their modeling. The VPT project group leader explicitly considered modeling of RM type of work to be simpler than RI type of work. In TEK-S, all seven models except M3 include work of RI type. As discussed in section 6.6.1, there were some disagreement over the M2 model, but not concerning the aspects that are dominated by replication risk. Modeling of work involving information-based artifacts was explicitly stated to be more difficult than modeling of physical processes (page 126).

Design risk, material artifacts (DM): Work within this category was not observed modeled in any of the projects.

Design risk, information-based artifacts (DI): The M2 model in the TEK-S project is the prime evidence for the perceived challenges of modeling DI type of work. The validity of the M2 model was heavily disputed, as discussed in section 6.6.4. In VPT, the attempt at modeling creative work at the R&D center was found difficult, and in PA30, the interviewed modelers stated that modeling of creative and unpredictable processes was difficult.

From a theoretical point of view, the difference in knowledge of work dominated by replication risk versus design risk can be considered intrinsic to the very definition of the concepts. Processes associated with replication risk are presumably more widely agreed upon than processes associated with design risk, as the products are known, and the process is repeated and can be observed over and over.

Concerning the perceived distinction between matter and information, one may argue that physical artifacts with their more direct manifestation than purely information-based artifacts are easier to make sense of and easier to agree upon. This is consistent with Dahlbom's (1992:119) differentiation of artifacts' "thingishness", where material, social and ideal artifacts (in that order) are perceived to become less and less objective and stable.

The rest of the assertions in this chapter are made in the context of modeling of work that is not well known, i.e., where modeling is more an act of sense-making than mere representation of a socially agreed upon reality. This is more in line with the perspective on enterprise modeling taken in this thesis.