1. Threshold Models of Technological Transitions Utrecht University Summerschool Complex Systems August 2015

2. Introduction A large part of complexity research deals with conditions under which autonomous particles or agents suddenly show coordinated behaviour leading to the emergence of macroscopic patterns. In the social sciences this is an old question as human agents are in principle driven by personal contexts, yet sometimes show remarkable coordinated behaviour. Think of social unrest, social norms, fashions, media hypes, etc. The threshold is generally expressed as the number of other agents already adopting. Hence, many technologies are slow to diffuse due to this coordination, a.k.a. lock-in problem. There is a wide interest because of the need of sustainability transitions. The take away message holds that there are many different but related ways to explain sudden transitions, which means that empirical research really has to go to the micro level to understand mechanism or mechanisms. Similarly, policy will only work well if the exact process underlying technology adoption is well understood.

3. Structure I will discuss: – The classic lock-in model of competing technologies – The modified lock-in model of transitions – Informational cascades – The NK-model – Percolation model Background literature:

4. The example of cars, bikes and planes PAGE 430/09/09

5. And the dominant designs that followed PAGE 530/09/09

6. A more recent example … PAGE 630/09/09

7. Dominant design PAGE 730/09/09

8. Lock-in Path dependence Irreversibility Multiple equilibria Unpredictability Population consists of 50-50 distribution of R-agents and S-agents Sequential decision- making

9. Modified lock-in model of technological transitions

10. Modified lock-in model

11. Lock-in model

13. Informational cascades

14. NK-model

17. 17 Percolation in a social network

18. 18 Percolation in a social network

19. 19 Different network structures

20. 20 Upper bound to diffusion: 45º line (perfect information) Phase changes: from a non- diffusion to a diffusion regime Regular and Small world networks very inefficient Thresholds depend on network structrures

21. 21 Thresholds depend on network structrures