Diese Präsentation wurde erfolgreich gemeldet.
Die SlideShare-Präsentation wird heruntergeladen. ×

The influence of Group Decision Making on Architecture Design Decisions

Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige

Hier ansehen

1 von 55 Anzeige

The influence of Group Decision Making on Architecture Design Decisions

Herunterladen, um offline zu lesen

Group Decision Making influcencs Architecture Design Decisions. This presentation, given as a keynote at the MARCH 2019 workshop (https://is.ieis.tue.nl/research/bpm/MARCH/index.php/keynote/), tries to identifies GDM factors that influence architecture design decisions.

Group Decision Making influcencs Architecture Design Decisions. This presentation, given as a keynote at the MARCH 2019 workshop (https://is.ieis.tue.nl/research/bpm/MARCH/index.php/keynote/), tries to identifies GDM factors that influence architecture design decisions.

Anzeige
Anzeige

Weitere Verwandte Inhalte

Ähnlich wie The influence of Group Decision Making on Architecture Design Decisions (20)

Weitere von Henry Muccini (20)

Anzeige

Aktuellste (20)

The influence of Group Decision Making on Architecture Design Decisions

  1. 1. The influence of Group Decision Making on Architecture Design Decisions @MARCH 2019 Henry Muccini DISIM Department University of L’ Aquila, L’Aquila, Italy henry.muccini@univaq.it - @muccinihenry Slides available at: http://www.slideshare.net/henry.muccini/
  2. 2. Special thanks to Smrithi Rekha V. whose contribution is substantial for this research line. Some of this research has been also carried out with Ivano Malavolta, and Damien Andrew Tamburri.
  3. 3. “Group decision-making is a research area that aims to understand and develop methods to enhance the collective decision process” [IST2018]
  4. 4. Henry Muccini – SAGRA 2017 6 Architecting in a picture ADD
  5. 5. Henry Muccini – SAGRA 2017 7 Collaborative Decision Making ADD ADD ADD ADD GDM
  6. 6. Henry Muccini @ MARCH 2019@ICSA2019 8 Group Decision Making Group decision making = 1. multiple people taking decisions (social attachment) 2. multiple roles/stakeholders with different concerns (42010)
  7. 7. Henry Muccini @ MARCH 2019@ICSA2019 10 Group Decision Making • Group characteristics: • size, composition and cohesion • the stages in the formation of groups, • information exchange within the group, • GDM methods, • issues faced
  8. 8. Henry Muccini @ MARCH 2019@ICSA2019 11
  9. 9. Henry Muccini @ MARCH 2019@ICSA2019 12 https://bit.ly/2U4O1Fm
  10. 10. Henry Muccini @ MARCH 2019@ICSA2019 13 https://bit.ly/2JwJwj5
  11. 11. Henry Muccini @ MARCH 2019@ICSA2019 14 Main Factors (by the audience)
  12. 12. Henry Muccini @ MARCH 2019@ICSA2019 15 Main Factors (by me) Social Links among team members Social Links among team members ContextContext Type/Importance of Decisions Type/Importance of Decisions Conflicts management Conflicts management Norms and Regulations Norms and Regulations Group characteristics Group characteristics Selection Strategy (e.g., voting) Selection Strategy (e.g., voting)
  13. 13. Henry Muccini @ MARCH 2019@ICSA2019 16 Main factors by the GGPS 16 General Group Problem-Solving (GGPS), 1993 [6] (generic model of GDM) impacts impacts impacts
  14. 14. Henry Muccini @ MARCH 2019@ICSA2019 17 Main factors by Saaty and Vargas 17 Thomas L. Saaty and Luis G. Vargas, 2006 [5]
  15. 15. Henry Muccini @ MARCH 2019@ICSA2019 18Henry Muccini, Damian Andrew Tamburri, V. Smrithi Rekha: On the Social Dimensions of Architectural Decisions. ECSA 2015: 137-145
  16. 16. Why shall we, software architects, care?
  17. 17. Henry Muccini @ MARCH 2019@ICSA2019 20 RQ1: What are the existing group decision-making practices in real world SA groups? RQ2: Are the group decision- making techniques currently being practiced in line with techniques in GDM literature? RQ3: What are the challenges that SA groups face while making architecture-related group decisions? RQ4: How satisfied are SA group members with various aspects of GDM?26 different organizations 35 practitioners
  18. 18. Henry Muccini @ MARCH 2019@ICSA2019 21 RQ1 (GDM practices in industry) 1. 3-5 people involved in architecture decision making 1. Laughlin empirically proved that 3-5 is the ideal group size for an efficient performance 56% 26% 18% people involved in ADD up to 5 up to 10 bigger than 10
  19. 19. Henry Muccini @ MARCH 2019@ICSA2019 22 RQ1 (GDM practices in industry) 2. Teams are mostly homogeneous (middle management employees) 3. Mostly «discussion- based» and «democratic» GDM. 2. it could lead to groupthink 3. Democratic and Laissez- faire approaches. These may cause information asymmetry, that reduces team performance (no systematic may impair decision quality)
  20. 20. Henry Muccini @ MARCH 2019@ICSA2019 23 GroupThink The tendency of groups to try to minimize conflict and reach consensus without sufficiently testing, analyzing, and evaluating their ideas. The pressures for conformity restrict the thinking of the group, bias its analysis, promote simplistic and stereotyped thinking, and stifle individual creative and independent thoughts
  21. 21. Henry Muccini @ MARCH 2019@ICSA2019 24 Information Asymmetry In contract theory and economics, information asymmetry deals with the study of decisions in transactions where one party has more or better information than the other. This creates an imbalance of power in transactions, which can sometimes cause the transactions to go awry, a kind of market failure in the worst case
  22. 22. Henry Muccini @ MARCH 2019@ICSA2019 25 RQ1 (GDM practices in industry) 4. Distributed vs co-located teams 5. Consensus 83% 17% Distributed vs Co-located teams distributed teams co-located teams 83% 17% How to get to a consensus Discussion-based rule-based
  23. 23. Henry Muccini @ MARCH 2019@ICSA2019 26 RQ2 (GDM practice vs GDM theory) 1. GDM practices combine different GDM (unstructured) GDM methods 2. Different stakeholders are given different weights 2. seniority (65%), expertise (9%), specific factors (business, political, technological) (26%) 71 65 17 11 9 0 10 20 30 40 50 60 70 80 BRAINSTORMING CONSENSUS-BASED VOTING DELPHI AHP Type of used GDM
  24. 24. Henry Muccini @ MARCH 2019@ICSA2019 27 GDM methods • The participants freely propose a list of alternatives. Then, brainstorm over them to arrive at a final decision. A leader moderates (supporting the generation of ideas). Brainstorming • Alternatives are provided by the participants, and then voted.Voting • Experts answer questionnaires in a distributed and anonymous way. A facilitator provides an anonymous summary of the experts’ forecasts (after each round). (avoids influence between experts) Delphi •Consensus: several alternatives are listed, an effort is made to achieve maximum level of consensus. Selection: once there is high level of agreement among participants selected and the decision is made the best alternative are selected. Consensus- Selection • The problem is modelled as goals, alternatives and criteria. Participants are normally experts who do a pairwise comparison of alternatives based on certain criteria. The results are then synthesized to make the final decision AHP
  25. 25. Henry Muccini @ MARCH 2019@ICSA2019 28 RQ2 (GDM practice vs GDM theory) 3. How often major decisions are omitted? Why? A) Time Constraints 13% B) Lack of Communication and Consultation - 13% C) Lack of Knowledge and Expertise - 13% E) Poor Exploration of Requirements and Solutions - 13% F) dominance of certain members leading to others keeping quiet 25% 13% 47% 40% How often have major decision been omitted in the GDM Process never rarely often
  26. 26. Henry Muccini @ MARCH 2019@ICSA2019 29 RQ3 (Challenges in group decisions) 1. GMD issues 63 66 54 28 17 A. GOALS ARE NOT UNDERSTOOD BY THE ENTIRE GROUP B. LONG TIME TO ARRIVE AT FINAL CONSENSUS C. PRESENCE OF POWER DIFFERENCES D. ACCEPTING LOW-RISKY SIMPLE SOLUTIONS E. MORE EXTREME OR RADICAL DECISIONS IN A GROUP Challenges
  27. 27. Henry Muccini @ MARCH 2019@ICSA2019 30
  28. 28. Henry Muccini @ MARCH 2019@ICSA2019 31 RQ4 (SA members satisfaction of their GDM)
  29. 29. Henry Muccini @ MARCH 2019@ICSA2019 32 In summary GDM impacts the quality of the software architecture design decisions
  30. 30. Evaluation Framework
  31. 31. Henry Muccini @ MARCH 2019@ICSA2019 34 Evaluation framework definition RQ1) how to evaluate the architecture design decision methods’ suitability for group decision making? RQ2) how adequate existing architecture design decision methods are for group decision making? 34
  32. 32. Henry Muccini @ MARCH 2019@ICSA2019 35
  33. 33. Henry Muccini @ MARCH 2019@ICSA2019 36 General Group Problem- Solving (GGPS), 1993 (generic model of GDM) Thomas L. Saaty and Luis G. Vargas, 2006
  34. 34. Henry Muccini @ MARCH 2019@ICSA2019 37 Evaluation framework definition (3/3) 37 Problem Identification Development of alternatives Preference Indication Prioritizing Group Members Provision for conflict resolution Group Decision Rules Information Exchange and Recall Revisiting Information
  35. 35. Henry Muccini @ MARCH 2019@ICSA2019 38 2. ADD methods selection [10] Falessi, et al. Decision-making techniques for software architecture design: A comparative survey. ACM Computing Surveys (CSUR) 43(4) (2011) [16] Tofan, et al. Past and future of software architectural decisions a systematic mapping study. IST 56(8) (2014) 38 Only decision-making (DM) processes/methods Decision methods covering broad aspects of DM Coverage of different SA DM Dealing with conflicting multiple objectives We included Output: 22 DM processes/method [17-38]
  36. 36. Henry Muccini @ MARCH 2019@ICSA2019 39
  37. 37. Henry Muccini @ MARCH 2019@ICSA2019 40 3. Evaluation framework applied to ADD methods 40 Problem Identification Development of alternatives Preference Indication Prioritizing Group Members Provision for conflict resolution Group Decision Rules Information Exchange and Recall Revisiting Information Few methods present an explicit problem identification step. At best, the process starts with identification of alternatives (A good problem identification step -> better problem space analysis -> high quality GDM practice [5])
  38. 38. Henry Muccini @ MARCH 2019@ICSA2019 41 3. Evaluation framework applied to ADD methods 41 Problem Identification Development of alternatives Preference Indication Prioritizing Group Members Provision for conflict resolution Group Decision Rules Information Exchange and Recall Revisiting Information Very few methods allow for a group to discuss and evolve alternatives. Multi-criteria decision-making methods must allow for the generation and filtering of alternatives through a process of discussion and deliberation which ensures more participation of group members [5].
  39. 39. Henry Muccini @ MARCH 2019@ICSA2019 42 3. Evaluation framework applied to ADD methods 42 Problem Identification Development of alternatives Preference Indication Prioritizing Group Members Provision for conflict resolution Group Decision Rules Information Exchange and Recall Revisiting Information The selected methods allow for preference indication but it is mostly individuals who rank the alternatives. They do not seem to allow multiple stakeholders to indicate preferences.
  40. 40. Henry Muccini @ MARCH 2019@ICSA2019 43 3. Evaluation framework applied to ADD methods 43 Problem Identification Development of alternatives Preference Indication Prioritizing Group Members Provision for conflict resolution Group Decision Rules Information Exchange and Recall Revisiting Information (almost) none of the methods account for hierarchy or expertise differences among stakeholders.
  41. 41. Henry Muccini @ MARCH 2019@ICSA2019 44 3. Evaluation framework applied to ADD methods 44 Problem Identification Development of alternatives Preference Indication Prioritizing Group Members Provision for conflict resolution Group Decision Rules Information Exchange and Recall Revisiting Information no method accounts for conflict management strategies. The sources of conflict, levels of conflict and appropriate conflict resolution styles could be applied to the SA decision-making methods. Collaborative style of conflict resolution is the most popular [1], so, it shall be supported
  42. 42. Henry Muccini @ MARCH 2019@ICSA2019 45 3. Evaluation framework applied to ADD methods 45 Problem Identification Development of alternatives Preference Indication Prioritizing Group Members Provision for conflict resolution Group Decision Rules Information Exchange and Recall Revisiting Information Very few allow for multiple stakeholder preference and hence they alone discuss decision-rules. (The more rigorous and scientific the decision-rule is, the better the quality of decisions made [5], [14])
  43. 43. Henry Muccini @ MARCH 2019@ICSA2019 46 3. Evaluation framework applied to ADD methods 46 Problem Identification Development of alternatives Preference Indication Prioritizing Group Members Provision for conflict resolution Group Decision Rules Information Exchange and Recall Revisiting Information Two of the chosen methods seem to indicate the presence of shared visual representation of information. (Information recall has been found to be key in making the knowledge pool more rich)
  44. 44. Henry Muccini @ MARCH 2019@ICSA2019 47 3. Evaluation framework applied to ADD methods 47 Problem Identification Development of alternatives Preference Indication Prioritizing Group Members Provision for conflict resolution Group Decision Rules Information Exchange and Recall Revisiting Information Only two methods are iterative in nature. The more number of times the group is able to exchange information, uncover unshared information and revisit the alternatives, the higher the quality of decisions.
  45. 45. Henry Muccini @ MARCH 2019@ICSA2019 48 Reflections  Most of these methods were not explicitly designed to involve groups;  Very few methods provisioned an explicit systematic method for discussing and evolving alternatives;  None of the surveyed methods had specific aspects of conflict resolution;  Few methods allowed for multiple stakeholder preference  Why? 48
  46. 46. Ongoing work
  47. 47. Henry Muccini @ MARCH 2019@ICSA2019 50 Experimenting the GDM Objective : To understand the effectiveness and efficiency of SA groups through experiments Outcomes measured: • N° of valid alternatives- Quantity , • Quality of Alternatives as evaluated by judges with experts - Scores, • satisfaction of group members a. Group1 : No GDM, Homogenous b. Group 2: GDM, Homogenous c. Group 3: No GDM, Heterogeneous d. Group 4: GDM, Heterogeneous
  48. 48. Henry Muccini @ MARCH 2019@ICSA2019 51 • Results indicate that the Group Decision Making method plays a significant role in the generation of alternatives and those with GDM generate higher number of valid alternatives • GDM plays a significant role in the quality of alternatives. Those that use GDM generate better alternatives • When the feedback scores were summed up, they did not show any significant difference among the groups, though this has to be further analysed
  49. 49. Concluding
  50. 50. Henry Muccini @ MARCH 2019@ICSA2019 53 Not anymore on a Ivory tower The architect The team with the architect
  51. 51. Henry Muccini @ MARCH 2019@ICSA2019 54 “the graph of interactions, patterned relations and arrangements emerging between individuals in the same endeavor”... Organizational Social Structure Dev. 1 Dev. 4 Dev. 2 Dev. 3 Dev. N-1 Dev. N … Art. 2 Art. 3Art. 1 Art. 4 Art. N Task Allocations
  52. 52. Henry Muccini @ MARCH 2019@ICSA2019 55 The 13+ Organisational Types [TLVV2012] Damian A. Tamburri, Patricia Lago, Hans Van Vliet. Organizational social structures for software engineering - ACM Computing Surveys (2012) 1–35 (at least) 81 properties observed so far that define types
  53. 53. Henry Muccini @ MARCH 2019@ICSA2019 56 A first attempt at an overlap
  54. 54. Henry Muccini @ MARCH 2019@ICSA2019 57 In summary How to make GDM effective and efficient?
  55. 55. Henry Muccini @ MARCH 2019@ICSA2019 58 Systematic GDM Fitting into the Organizational Social Structure Development of alternatives Conflict resolution & convergence Prioritizing Group Members Minimizing Groupthink, Asymmetry, etc. Tool support

×