Examining Bias in Construction Kits Solving the Combinatorics The - PowerPoint PPT Presentation

Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Examining Bias in Construction Kits Solving the Combinatorics The 15 th International Conference on DESIGN Problem Evaluating Bias Findings Conclusions Dr. James Gopsill Future Work Questions University of Bath, UK 21 st –24 th May, 2018 1 / 24

Contents Examining Bias in Construction Kits Dr. James Construction Kits 1 Gopsill Construction Kits Solving the Combinatorics Problem 2 Solving the Combinatorics Problem 3 Evaluating Bias Evaluating Bias Findings Findings Conclusions 4 Future Work Questions Conclusions 5 Future Work 6 Questions 7 2 / 24

Construction Kits Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem Evaluating Bias Findings Conclusions Future Work Questions (a) Lincoln Logs (b) Lego (c) Minecraft Figure 1: Digital & Physical Construction Kits 3 / 24

Construction Kits Interesting Facts Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem Evaluating Bias Findings Conclusions Future Work Questions Figure 2: Lego There are approximately 62 bricks per person of the Earth’s population. 4 / 24

Construction Kits Interesting Facts Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem Evaluating Bias Findings Conclusions Future Work Questions Figure 3: Minecraft There are 75M monthly players of the digital Constuction Kit Minecraft. 5 / 24

Construction Kits Research Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem Evaluating Bias Findings Conclusions Future Work Questions Figure 4: Lego Therapy 1 1G. Owens et al. “LEGO Therapy and the Social Use of Language Programme”. In: Journal of Autism and Developmental Disorders 38.10 (June 2008), p. 1944. 6 / 24

Construction Kits Research Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem Evaluating Bias Findings Conclusions Future Work Questions Figure 5: Serious Play 2 2S. Klaus-Peter et al. “T oolkit-Based Modelling and Serious Play as Means to Foster Creativity in Innovation Processes”. In: Creativity and Innovation Management 24.2 (2015), pp. 323–340. 7 / 24

Construction Kits Research Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem Evaluating Bias Findings Conclusions Future Work Questions Figure 6: Embedding rules within bricks 3 3D. Mathias et al. “Design variation through richness of rules embedded in LEGO bricks”. In: Proceedings of the International Conference on Engineering Design . 2017. 8 / 24

Construction Kits Meeting of the minds Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem Evaluating Bias Findings Conclusions Future Work Questions Figure 7: Meeting of the minds 9 / 24

Construction Kits Bias Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem But how much influence does a Construction Kit Evaluating Bias Findings have on the freedom to design? Conclusions Future Work Questions 10 / 24

Construction Kits Bias Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem Evaluating Bias Findings Conclusions Future Work Questions Figure 8: Construction kit used to design the Sydney opera house 11 / 24

Construction Kits Bias Examining Bias in Construction Audience Participation Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem Evaluating Bias Findings Conclusions Future Work Questions Figure 9: 2 2 × 4 LEGO bricks ( B 2 , 4 ( 2 ) ) Using the two LEGO pieces you have been given, place one on top of the other. 12 / 24

Construction Kits Bias Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem Evaluating Bias Findings Conclusions Future Work Questions Figure 10: The 46 combinations of B 2 , 4 ( 2 ) 13 / 24

Construction Kits Bias Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem Evaluating Bias Findings Conclusions Future Work Questions Figure 11: 14 morphologically unique combinations of B 2 , 4 ( 2 ) 14 / 24

Construction Kits Bias Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the 4 2 4 2 4 2 2 1 4 8 4 4 4 1 Combinatorics Problem Evaluating Bias Figure 12: Number of pathways of B 2 , 4 ( 2 ) Findings Conclusions Future Work Questions Thus, there are multiple pathways to each combination. And, the number of pathways to each combination differs. So is the kit biasing particular solutions eventhough they are all valid? (Keep hold of your combination. We will come back to it at the end.) 15 / 24

Solving the Combinatorics Problem T o explore this further, we generated a breadth-first search algorithm. Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem Evaluating Bias Findings Conclusions Future Work Questions Figure 13: Breadth-first search algorithm 16 / 24

Solving the Combinatorics Problem Examining Bias in Construction Kits Dr. James Gopsill Construction Kits Solving the Combinatorics Problem Evaluating Bias Findings Conclusions Future Work Questions Figure 14: University of Bath’s Balena HPC (3,072 cores, 18TB main memory and 300TB storage providing a peak performance of 57Tflops) 17 / 24

Evaluating Bias Examining Bias in Construction Kits Dr. James Gopsill Construction Kits T wo Cases of B 2 , 4 ( n ) bricks: Solving the Combinatorics NAC No additional constraints Problem Evaluating Bias NRC No rotation constraint Findings Job size Conclusions Future Work 64 cores Questions 6 hours 18 / 24

Evaluating Bias Gini Coefficient Examining Bias in Construction Kits Dr. James 100% Gopsill � n � n j = 1 | x i − x j | i = 1 G = (1) Construction A � n � n Kits Line of 2 j = 1 x j i = 1 Solving the Cumulative Non- Combinatorics Problem share of Bias Evaluating Bias Lorenz path- Findings Curve ways Where x i is the number of pathways Conclusions to combination i , x j is the number of Future Work pathways to combination j and n is Questions B the number of combinations. Cumulative 100% In terms of the Lorenz curve in share of com- Figure 15, this is the ratio of the area binations (or- under the Lorenz curve ( B ) and the dered) area for a non-biased kit ( A+B ). Figure 15: Lorenz curve of the distribution of pathways to combinations 19 / 24

Findings Examining Bias in Construction Kits Dr. James T able 1: Number of pathways ( P ) and morphologically equivalent combinations ( M ) of B ( 2 , 4 ) ( n ) Gopsill P Construction Bw,h ( n ) P M P min P max A non-biased A biased B G M Kits NAC Solving the Combinatorics Problem B 2 , 4 ( 2 ) 92 14 6.6 2 16 620 137 472 0.31 12.3 × 103 2.6 × 106 98.0 × 103 1.6 × 106 B 2 , 4 ( 3 ) 429 28.7 4 104 0.39 Evaluating Bias 2.2 × 106 33.3 × 103 36.9 × 109 557.2 × 106 21.4 × 109 B 2 , 4 ( 4 ) 66.6 4 832 0.42 Findings 497.8 × 106 3.0 × 106 14.9 × 103 744.8 × 1012 4.5 × 1012 370.8 × 1012 B 2 , 4 ( 5 ) 166.2 4 0.51 Conclusions NRC Future Work B 2 , 4 ( 2 ) 42 8 5.3 2 8 162.0 49.0 131.0 0.27 Questions 2.5 × 103 172.5 × 103 10.8 × 103 126.0 × 103 B 2 , 4 ( 3 ) 139 17.9 4 32 0.29 198.5 × 103 4.6 × 103 449.7 × 106 10.5 × 106 300.6 × 106 B 2 , 4 ( 4 ) 43.5 4 448 0.34 19.7 × 106 193.4 × 103 2.0 × 103 1.9 × 1012 18.7 × 109 1.1 × 1012 B 2 , 4 ( 5 ) 101.6 4 0.42 90% reduction in the design space when adding a constraint 86% reduciton in P max 20% reduction in G 20 / 24

Findings Examining Bias in Construction Kits Dr. James Gopsill 1 NAC NAC Construction Kits 150 0 . 8 NRC NRC Solving the Combinatorics Problem 0 . 6 100 Evaluating Bias M G P Findings 0 . 4 Conclusions 50 0 . 2 Future Work Questions 0 0 2 3 4 5 2 3 4 5 n n (a) Mean pathway to combination ratio (b) Gini Coefficient Figure 16: Characteristics of bias within B ( 2 , 4 ) ( n ) construction kits 21 / 24

Conclusions Examining Bias in Construction Kits Dr. James Gopsill But how much influence does a Construction Kit have on the Construction Kits freedom to design? Solving the Combinatorics Problem Evaluating Bias A significant bias does exist within the brick-style (LEGO ™ ) construction Findings kits Conclusions However, it may not be perceived by the designer due to the Future Work Questions exponential rate of increase in the number of pathways and combinations Additional constraints reduces the design space but has little effect on the bias within the kit 22 / 24

Future Work Examining Bias in Construction Kits Dr. James Gopsill Construction Kits 4 2 4 2 4 2 2 1 4 8 4 4 4 1 Solving the Combinatorics Figure 17: Back to the pathways Problem Evaluating Bias Findings Conclusions Future Work Questions 23 / 24