After an idea or new combination has been developed, the next step is eval- uation. Most ideas prove wanting and are discarded almost immediately.
Handbook of New Product Development Management
The first winnowing occurs with the individual’s own thought trials. The inventor and scientist Faraday described the process (Simonton, 1999: 27):
‘The world little knows how many thoughts and theories which have passed through the mind of a scientific investigator have been crushed in silence and secrecy by his own severe criticism and adverse examinations; that in the most successful instances not a tenth of the suggestions, the hopes, the wishes, the preliminary conclusions have been realized.’
In contrast to variation, however, selection also takes place outside of the individual. Although the roles are rarely defined so starkly, the poet Paul Valéry (Simonton, 1999: 27) proposes that ‘it takes two to invent anything.
One makes up combinations; the other chooses and recognizes what he wishes and what is important to him in the mass of things which the former has imparted to him.’ Campbell concurs: ‘Such considerations suggest comple- mentary combinations of talent in creative teams, although the uninhibited idea-man and the compulsive edit-and-record type are notoriously incom- patible office mates’ (Campbell, 1960: 105). One huge advantage of social interaction is that combinations can be judged by a greater number of selec- tion criteria. ‘Much of creative thought is opportunistic in the sense of having a wide number of selective criteria available at all times, against which the thought trials are judged’ (Campbell, 1960: 104).
From a managerial perspective, the challenge changes from generating new ideas to choosing the most promising ones for further development. Fortu- nately, it becomes easier (or at least more of a repeatable process) to manage the product development process as it progresses from idea creation to ultimate commercialization. A process, which relies initially on individual creativity – relatively unpredictable, possibly unbalanced, and seemingly ‘blind’ – begins to rely more heavily on social interaction and more ‘manageable’ processes.
The research literature on the selection phase reflects this shift as well. Rather than coming from psychology and sociology research (as the literature on vari- ation did), it comes from operations management, economics, and marketing and focuses on traditional product development issues such as prototyping, testing, and marketing. These literatures have begun to reach back into the cre- ative stages and consider promising applications for technologically ‘pushed’
ideas (Dahan and Hauser, 2001).
Recent research has documented a tremendous improvement in product development tools for both the variation and the selection phases. Drug devel- opment provides a powerful archetype (Drews, 2000). Firms and universities now maintain libraries with hundreds of thousands of compounds for recombi- nant search. These library components are recombined and tested for efficacy against thousands of possible disease models. Ideally, such high-throughput screening constitutes an automatic variation generator and selector, which can run largely by itself. The process is not always effective, since model fidelity varies greatly, and even if a ‘hit’ occurs, the mechanism remains to
• • • • • 122
be explained. Still, high-throughput screening remains an attractive model for the automation of creativity in the product development process.
Simulation provides another promising tool for selection and, to a lesser extent, variation (Thomke, 1998; see also Thomke’s chapter in this vol- ume). The combination of the inexorable advance of computing power and the improvement in digital representation of products and product systems has greatly increased the efficacy of modeling products before manufac- turing them – even before prototyping them. The approach is particularly effective for complex systems with thousands of interacting parts. If the interactions between individual parts can be modeled accurately, the overall system dynamics will emerge accurately and without the need for hierarchical design and control. Engineers can test overall product hypotheses without delving into the details of component–by-component interactions. The tools also uncover unpredicted interactions. For example, BMW engineers dis- covered that overall vehicle safety was improved by weakening a particular frame support (Thomke et al., 1999). Ideally, these tools are integrated into a design methodology that fully explores the recombinant space through system- atic generation of variation and experimentation against high-fidelity models (Thomke, 2003).
Firms are often faced with too many product development possibilities.
While this may seem to be an enviable position, it is in fact a bad situ- ation, especially from a production process perspective (Wheelwright and Clark, 1992). Given the inherent organizational and managerial difficulty of killing projects, firms tend to allow too many to continue in their develop- ment pipelines. In any production system, throughput time increases as the system approaches capacity, with the result that all projects become late and over-budget. When there is an overabundance of product development pos- sibilities, rigorous selection processes become imperative for the firm’s sur- vival. Implementation of aggregate project planning processes (Wheelwright and Clark, 1992) that incorporate capacity, risk, and market analyses can help managers to avoid behavioral biases in project selection and to jus- tify the painful process of project selection. However, recent simulation work has demonstrated how simple and reasonable green-light rules can also cause extreme variance in the arrival of finished products (Gino and Pisano, 2005).
We also draw attention to a new trend in technology development, the rise of open-innovation communities (von Hippel, 2005), because they pro- vide a new model for an evolutionary view of product development. Open- innovation communities generate variance by the voluntary contributions of thousands of people. Motivated by a variety of reasons – including personal need, distrust of corporate goals, or community spirit – volunteers submit innovations with little guidance. The innovations are selected by commu- nity leaders (who are chosen for their technical and leadership abilities) and
Handbook of New Product Development Management
tested by the entire community. Open-innovation communities tackle problems ranging from software for operating systems to genetic research on agriculture (Broothaerts et al., 2005) to sporting equipment (Shah, 2005).
It is not yet clear whether open-innovation communities are mostly capable of refining modular systems or whether they are also capable of seminal breakthroughs. Community infrastructure, such as SourceForge, that enables any user to start a project may prove that open-innovation communities are indeed capable of very original and creative front-end ideation.
Probably the most valuable lesson for the classical literature on product development is the value of early product release to a dedicated, motivated, and diverse community of users. This subjects products to a huge variety of operating conditions and quickly identifies bugs. The process requires strong community leaders to outline modular architectures and to keep the voluntary associations from forking. A firm can benefit from such an open approach (MacCormack et al., 2001) but needs to manage its community’s perception of the firm’s proprietary motives.