Images contrasting "Point-Based" vs. "Set-Based" practices (from last reference below):
The first mentions of "set-based" practices was in:
A.C. Ward, J.K. Liker, J.J. Cristiano, and D.K. Sobek II, The second Toyota paradox: How delaying decisions can make better cars faster, Sloan Management Review 36 (1995), 43-61.
The follow-up article published in HBR brought even more attention to "set-based":
D.K. Sobek II, J.K. Liker, and A.C. Ward, Another look at Toyota's integrated product development, Harvard Business Review 76 (July-Aug 1998), 36-49.
The first paper to focus specifically on Toyota's set-based practices:
D.K. Sobek II, A.C. Ward, and J.K. Liker, Toyota's principles of set-based concurrent engineering, Sloan Management Review 40 (1999), 31-40.
Rather than focus on Toyota, this paper fleshes out the set-based concepts and how they can transform product development and systems engineering:
B.M. Kennedy, D.K. Sobek II, and M.N. Kennedy, Reducing rework by applying set-based practices early in the systems engineering process, Systems Engineering Journal 17, no. 3 (Autumn 2014), 278-296. (pdf available here)
The most extensive treatment of set-based practices, along with the associated enabling practices, is in this book:
P.W. Cloft, M.N. Kennedy, and B.M. Kennedy, Success is Assured: Satisfy your customers on time and on budget by optimizing decisions collaboratively using reusable visual models, Productivity Press, Taylor and Francis Group, 2019. (see SuccessIsAssured.com)
Origins of the Term: the term was first introduced along with the term "set-based concurrent engineering" to describe the practices observed in the Toyota Product Development System that made it 4X more productive than its competitors by Dr. Allen Ward, Dr. Durward Sobek, and their colleagues in papers published in the Harvard Business Review and Sloan Management Review (see the first three references below).
Definition: a design, analysis, and decision-making practice characterized by:
not selecting a particular values for design decisions but rather analyzing the full set of possibilities, not as a finite collection of a few points, but as an infinite set of points
rather than trying to “pick the best”, instead trying to “eliminate the weak”
rather than concrete visualizations of the actual geometry, instead focusing on visualizations of the full trade space and on the sensitivities between different targeted attributes and causal decisions
not selecting a particular design and then analyzing its performance, but rather analyzing the drivers of performance and limits on feasibility and using that knowledge to identify more desirable parts of the design space (of the full set of possibilities)
Origins of the Practice: we trace the first use of set-based practices to the Wright Brothers; after the typical lack of success that others were having in achieving manned powered heavier-than-air flight, the Wright Brothers stopped designing airplanes and instead turned their focus to closing three key knowledge gaps, capturing that knowledge in trade-off charts and tables, and then used that understanding to design the first successful aircraft. (see the last two references for more on that) The aircraft industry continued to heavily use trade-off curves until CAD and CAE software was adopted in the 1970's, leading to a transition back to point-based practices (and significant losses in productivity). Following World War II, many Japanese aircraft engineers went to work in the automotive industry, carrying with them set-based practices. More so than other companies, Toyota resisted adoption of new technologies that compromised those set-based practices.
Images contrasting too few points to have set-based knowledge vs. having enough points to gain set-based knowledge (from last reference below):