1 00:00:08,599 --> 00:00:10,190 Welcome back! 2 00:00:10,190 --> 00:00:14,859 So far we have been talking mainly about legacy infrastructure systems. 3 00:00:14,859 --> 00:00:19,730 You may have started to wonder about new infrastructure systems. 4 00:00:19,730 --> 00:00:24,989 They do not yet suffer from wear and tear, and you may expect them to be well documented, 5 00:00:24,989 --> 00:00:27,149 in each and every detail. 6 00:00:27,149 --> 00:00:32,079 Surely you would expect such a system to behave more predictably? 7 00:00:32,079 --> 00:00:36,059 I am sorry to disappoint you. 8 00:00:36,059 --> 00:00:41,469 For a small system at a very local level it may be possible to model the system accurately 9 00:00:41,469 --> 00:00:44,129 and to predict its behavior. 10 00:00:44,129 --> 00:00:48,079 However, most infrastructure systems are much larger, 11 00:00:48,079 --> 00:00:54,649 and have a tendency to grow continuously, as a result of economies of scale and network 12 00:00:54,649 --> 00:00:57,760 externalities. 13 00:00:57,760 --> 00:01:02,969 Many technologies applied in infrastructure systems are characterized by decreasing cost 14 00:01:02,969 --> 00:01:09,900 per unit of output with increasing scale, until a certain optimum size of operation. 15 00:01:10,000 --> 00:01:19,800 That is why electricity infrastructures are still dominated by large scale thermal power plants. 16 00:01:20,909 --> 00:01:26,359 The growth of infrastructure networks is largely explained by the economic driver to exploit 17 00:01:26,359 --> 00:01:31,020 economies of scale, thus making the service more and more affordable 18 00:01:31,020 --> 00:01:35,390 for increasing numbers of users. 19 00:01:35,390 --> 00:01:42,350 The other driver for network growth, the phenomenon of network externalities, 20 00:01:42,350 --> 00:01:48,060 is the phenomenon of increasing user value with an increasing number of users being connected 21 00:01:48,060 --> 00:01:49,539 to the network. 22 00:01:49,539 --> 00:01:52,039 Let me give you an example. 23 00:01:52,039 --> 00:01:59,039 A telephone connection would be of little value if only a few other people have a phone. 24 00:01:59,700 --> 00:02:06,700 Anyone connecting to the telephone network or the internet increases the value of the 25 00:02:06,700 --> 00:02:09,619 network to other users. 26 00:02:09,619 --> 00:02:14,680 Anyone buying a smart phone increases the usefulness of such phones to other people 27 00:02:14,680 --> 00:02:18,160 already using a smart phone. 28 00:02:18,160 --> 00:02:24,050 It is due to network externalities that, once a new infrastructure service takes off, 29 00:02:24,050 --> 00:02:33,000 the system tends to grow until the usage of that service is universal, or near-universal. 30 00:02:33,200 --> 00:02:40,060 Economies of scale and network externalities explain why infrastructure systems for energy, 31 00:02:40,060 --> 00:02:43,890 transport, telecommunication and information services 32 00:02:43,890 --> 00:02:47,870 have a natural tendency to grow into huge systems, 33 00:02:47,870 --> 00:02:52,060 comprising a huge number of subsystems, links and nodes, 34 00:02:52,060 --> 00:02:56,840 all of which are interdependent in several ways. 35 00:02:56,840 --> 00:03:02,910 If one subsystem is not functioning well, this may have far-reaching repercussions on 36 00:03:02,910 --> 00:03:06,710 the functioning of the overall system. 37 00:03:06,710 --> 00:03:10,300 The interdependence of the subsystems can take various forms, 38 00:03:10,300 --> 00:03:17,300 from simple linear dependencies to multiple, non-synchronous relationships. 39 00:03:17,660 --> 00:03:22,240 The non-linearities caused by feedbacks between subsystems, 40 00:03:22,240 --> 00:03:28,670 across system levels and time scales, are the main cause of emergent behavior of 41 00:03:28,670 --> 00:03:34,950 the aggregated system, that is the system as a whole. 42 00:03:34,950 --> 00:03:39,700 As the number of subsystems and interrelationships increases, 43 00:03:39,700 --> 00:03:43,130 and as those interrelationships become more diverse, 44 00:03:43,130 --> 00:03:50,200 it becomes more difficult to gain an overall view of the system and to know all the feedback loops. 45 00:03:50,410 --> 00:03:54,680 Eventually, the system will become so complex that the 46 00:03:54,680 --> 00:04:00,540 analyst can no longer recognise or model it at all. 47 00:04:00,540 --> 00:04:04,540 Fortunately, the emergent behavior of infrastructure systems 48 00:04:04,540 --> 00:04:08,480 shows remarkably consistent patterns. 49 00:04:08,480 --> 00:04:14,710 Even if I do not come home at exactly the same time every day, 50 00:04:14,710 --> 00:04:21,000 and though my individual use of electric devices in my home differs from day to day, 51 00:04:21,000 --> 00:04:27,590 the aggregated pattern of road congestion and electricity demand during the day is very 52 00:04:27,590 --> 00:04:30,880 similar, with slight variations between working days 53 00:04:30,880 --> 00:04:36,450 and weekends, and with seasonal variations. 54 00:04:36,450 --> 00:04:43,450 These recurrent patterns play a crucial role in the operation of infrastructure systems. 55 00:04:44,100 --> 00:04:48,480 Infrastructure operators recognize these patterns and know how to deal with them. 56 00:04:48,480 --> 00:04:54,650 They use models that predict how their control actions influence the aggregate behavior of 57 00:04:54,650 --> 00:04:58,570 the system, even if they cannot model how this behavior 58 00:04:58,570 --> 00:05:05,570 emerges from the interacting elements at the micro-level of the system. 59 00:05:06,380 --> 00:05:08,560 As a user, I do not care, 60 00:05:08,560 --> 00:05:12,910 since the value of the infrastructure for me is determined by the system's performance 61 00:05:12,910 --> 00:05:15,400 at the aggregate level. 62 00:05:15,400 --> 00:05:19,510 What difference does it make for me what cables or switches are used, 63 00:05:19,510 --> 00:05:26,510 as long as I can make a phone call and watch television in a comfortably heated home? 64 00:05:27,960 --> 00:05:35,800 Another factor contributing to the predictability of infrastructure system behavior is path dependency. 65 00:05:36,400 --> 00:05:41,430 Since most infrastructures use technologies that are characterized by strong economies 66 00:05:41,430 --> 00:05:45,700 of scale, they include many large scale, 67 00:05:45,700 --> 00:05:51,540 capital intensive installations with an economic lifetime of decades. 68 00:05:51,540 --> 00:05:57,490 This feature implies that the physical system is fairly stable. 69 00:05:57,490 --> 00:06:02,240 Also the transportation and distribution networks are capital intensive. 70 00:06:02,240 --> 00:06:06,590 They represent huge sunk costs. 71 00:06:06,590 --> 00:06:13,590 Sunk costs are costs that have already been incurred in the past and that cannot be recovered. 72 00:06:15,220 --> 00:06:22,800 Even in a tiny country like the Netherlands (only 37,350 square km), 73 00:06:22,800 --> 00:06:27,630 the distribution networks for electricity, natural gas, 74 00:06:27,630 --> 00:06:36,200 drinking water and sewage each represent more than 100,000 km of underground cables or pipelines 75 00:06:36,230 --> 00:06:41,700 to serve its 17 million inhabitants. 76 00:06:41,750 --> 00:06:51,200 The high voltage and high pressure transmission networks each reach a length of almost 10,000 km. 77 00:06:51,200 --> 00:06:55,990 The cost of these networks cannot be recovered if we were to decide today that it would be 78 00:06:55,990 --> 00:07:01,260 smarter to use an entirely different technology. 79 00:07:01,260 --> 00:07:08,260 We have made these costs in the past, they were already occurred and cannot be recovered, 80 00:07:08,260 --> 00:07:14,800 so we are dealing with the phenomenon of sunk costs. 81 00:07:14,800 --> 00:07:19,150 It is highly unlikely that we would adopt that smart new technology, 82 00:07:19,150 --> 00:07:26,150 if it would entail the need to build a new network requiring billions of Euros investment. 83 00:07:26,930 --> 00:07:32,410 The sunk costs represented by the existing system make it more likely that we will stick 84 00:07:32,410 --> 00:07:34,630 to the established system. 85 00:07:34,630 --> 00:07:39,010 In other words, technological choices that we made a long 86 00:07:39,010 --> 00:07:43,810 time in the past, have created a certain path dependency: 87 00:07:43,810 --> 00:07:53,800 they dictate many of the choices we make today about expanding and innovating our infrastructure systems. 88 00:07:54,120 --> 00:08:00,680 The path dependency created by past technology choices and capital investments does not mean 89 00:08:00,680 --> 00:08:05,050 that established infrastructure systems will never become obsolete. 90 00:08:05,050 --> 00:08:15,000 If a new technology comes around which promises far better performances, far better value for the user, 91 00:08:15,000 --> 00:08:19,500 possibly an entire new service, 92 00:08:19,560 --> 00:08:26,560 it is likely to be adopted if it can compete with the established infrastructure. 93 00:08:27,030 --> 00:08:32,000 A prime example is mobile telephony, that requires comparatively small "network" 94 00:08:32,000 --> 00:08:38,279 investments, so it is cost-effective, while bringing unprecedented flexibility in 95 00:08:38,279 --> 00:08:43,809 telecommunication, and revolutionary new services. 96 00:08:43,809 --> 00:08:48,240 Many developing economies around the world have embraced mobile telephony, 97 00:08:48,240 --> 00:08:55,240 while leapfrogging the copper wired fixed telephone infrastructure. 98 00:08:55,740 --> 00:09:01,850 Developed economies have been slower to adopt the mobile phone and the new services it enables, 99 00:09:01,850 --> 00:09:11,000 such as mobile money transactions, than for example, African countries like Kenya. 100 00:09:11,490 --> 00:09:18,900 Studies on complex systems often use the concept of agents for interacting elements in the system. 101 00:09:19,490 --> 00:09:24,269 In general, an agent is a model for any entity in the system 102 00:09:24,269 --> 00:09:31,089 that acts according to a set of rules, depending on input from the outside world. 103 00:09:31,089 --> 00:09:36,420 An agent can be an automatic on-off switch in a local control system, 104 00:09:36,420 --> 00:09:41,939 it can be a sophisticated software entity that is capable of intelligent control actions, 105 00:09:41,939 --> 00:09:46,410 it can be a human controller or any other decision maker, 106 00:09:46,410 --> 00:09:50,970 somewhere in the infrastructure system. 107 00:09:50,970 --> 00:09:54,220 By now, it should be clear that our definition of 108 00:09:54,220 --> 00:09:59,579 infrastructure does not only refer to the physical network. 109 00:09:59,579 --> 00:10:06,500 In our view, an infrastructure system includes 110 00:10:06,500 --> 00:10:13,600 - besides the transport and distribution networks - the carriers, conversion and storage facilities 111 00:10:13,600 --> 00:10:15,689 as well as the governance, 112 00:10:15,689 --> 00:10:22,689 management and control systems that are needed to make the system meet its functional specifications 113 00:10:23,290 --> 00:10:28,100 and its social objectives. 114 00:10:28,100 --> 00:10:33,949 In all parts of the system, social agents or actors as we call them, 115 00:10:33,949 --> 00:10:40,949 are making big and small decisions that influence the behavior of the system. 116 00:10:41,860 --> 00:10:48,100 The complexity of infrastructure systems in the social domain is the subject of the next video lecture. 117 00:10:48,139 --> 00:10:50,500 Thank you for your attention.