Presentation is loading. Please wait.

Presentation is loading. Please wait.

Projection and the Reality of Routines – reflections of a computational modeller Bruce Edmonds Centre for Policy Modelling Manchester Metropolitan University.

Similar presentations


Presentation on theme: "Projection and the Reality of Routines – reflections of a computational modeller Bruce Edmonds Centre for Policy Modelling Manchester Metropolitan University."— Presentation transcript:

1 Projection and the Reality of Routines – reflections of a computational modeller Bruce Edmonds Centre for Policy Modelling Manchester Metropolitan University

2 Background I am a computational modeller who does agent-based social simulation, I have been playing with models of organisational routines (which I am NOT going to exhibit), where the organisation has complex combinatorial problems to solve efficiently and the composite individuals have a variety of skills, but where I did not have to distinguish automatic and conscious cognition. This brought me to face some problems: –Agents need to understand what is supposed to happen so it can participate –However reasoning about the full complexity of what might happen is infeasibly complex –It seems almost inevitable that division of tasks and their negotiation happens at a rough, almost fictional level, with the detail invented/adapted on the fly Projection and the Reality of Routines, Bruce Edmonds, Workshop on Modelling Routines, Manchester, 25 th Nov. 2014, 2

3 The Ostensive and Performative Routines can be understood in two ways: 1.The Ostensive – how people think about and talk about routines (O) 2.The Performative – the regular patterns in what people do (P) In this routines are similar to other kinds of social phenomena (e.g. norms/conventions) It may be thought that one can study routines from either O or P Or that one of O or P is the more ‘real’ However, I will argue that the interaction between O and P levels is essential to their operation and characteristics Projection and the Reality of Routines, Bruce Edmonds, Workshop on Modelling Routines, Manchester, 25 th Nov. 2014, 3

4 Ostensive-Performative Interaction There are, at least, two kinds of interaction between the O and P levels: –P  O Induction: recognising regularities in the patterns of what people do this is fallible – only some of the more obvious regularities will be independently recognised these might be checked/corrected/confirmed by talking to others about these (at O level) –O  P Interpretation: using O knowledge of routines to decide what to do O versions of routines will inevitably have to be adapted to cope with the unexpected sometimes O and P versions will basically differ if the rhetoric of what happens is different from reality Projection and the Reality of Routines, Bruce Edmonds, Workshop on Modelling Routines, Manchester, 25 th Nov. 2014, 4

5 Abduction/Projection of Routines Induction of routines (P  O) is rarely pure ‘bottom-up’ pattern recognition (i.e. starting from a blank slate of knowledge) But rather, is biased by the existing O knowledge an agent has In other words, an agent will try to ‘fit’ or ‘project’ perceptions of P behaviour onto their existing O knowledge Projection and the Reality of Routines, Bruce Edmonds, Workshop on Modelling Routines, Manchester, 25 th Nov. 2014, 5

6 Shared Knowledge Where experience of a context is shared, this knowledge can be used by others to predict/understand deviations of P from O (distinguish noise from signal) Also an intentional understanding of others may allow for the recognition of what was intended to occur (O), separately from what did occur (P) Discourse (at the O level) and the needs of coordination will force some functional compatibility between O conceptions of P (but maybe rough) Furthermore, O discourse may result in O and P becoming more mutually consistent (through adaption in O or in P) Projection and the Reality of Routines, Bruce Edmonds, Workshop on Modelling Routines, Manchester, 25 th Nov. 2014, 6

7 Different Conceptions of Routines Different groups of people might have different representations (O) of what people are (or should be) doing due their different goals, expectations, concepts, etc. As long as either (a) the differences are understood by each group or (b) the differences happen to make no difference to coordination between these groups these differences can persist and be stable For example management and workers can have very different conceptions of routines Projection and the Reality of Routines, Bruce Edmonds, Workshop on Modelling Routines, Manchester, 25 th Nov. 2014, 7

8 New Routines New patterns of behaviour (P) may arise for various reasons (error, individual creativity, force of circumstance etc.) These patterns may be copied by others (P) They may be recognised and enter the discourse concerning coordination (O) Thus priming others to recognise them Over time the routines may become socially entrenched (at O and P levels) Projection and the Reality of Routines, Bruce Edmonds, Workshop on Modelling Routines, Manchester, 25 th Nov. 2014, 8

9 Priming In the process of developing new routines, they can be shaped by ‘priming’ or suggestion of what they ‘are’ or by knowledge of what routines have been in similar situations in the past This may not directly affect the P level but by influencing the O processes of coordination (for example setting up mutual expectations) it can constrain P to conform (more) to this O account Projection and the Reality of Routines, Bruce Edmonds, Workshop on Modelling Routines, Manchester, 25 th Nov. 2014, 9

10 O Institutional Memory When routines fall into disuse due to circumstances (at least at the P level) they may remain as part of the O lexicon (in documents and talk) When circumstances change, making the previous patterns possible again, then due to priming and the complexity of coordination, they may be re-seeded at the P level again due to their O existence Projection and the Reality of Routines, Bruce Edmonds, Workshop on Modelling Routines, Manchester, 25 th Nov. 2014, 10

11 The Reality of Routines Comes about due to the (maybe weak) connection between O and P levels, so that routines may be talked about and coordinated But routines are only possible due to the flexibility of P from O, that it can be greatly adapted to circumstances whilst still being recognisable and referable The two levels and the interaction is essential for their effectiveness Rountines are a phenomena that arises from the combination of O and P levels Projection and the Reality of Routines, Bruce Edmonds, Workshop on Modelling Routines, Manchester, 25 th Nov. 2014, 11

12 Some Hypotheses Arising The more complicated the coordination and the more predictable the situation, the more P will correspond to O The less constrained the coordination and the more uncertain the situation, the more P will be adapted and not necessarily correspond to any O discourse about it Discrepancies between P and O might not be enough to cause either: –an update of O if the social consensus underpinning O is against this –an update of P if circumstances make this difficult to do Projection and the Reality of Routines, Bruce Edmonds, Workshop on Modelling Routines, Manchester, 25 th Nov. 2014, 12

13 Reflections The task of modelling routines in all their glory is quite complex. It combines the difficulty of: –The AI symbol grounding problem –How social consensus and negotiation occurs –The fluid use of analogy and ‘fictions’ –The actual adaptability in the face of specific situations But, I do think it IS possible! Projection and the Reality of Routines, Bruce Edmonds, Workshop on Modelling Routines, Manchester, 25 th Nov. 2014, 13

14 The End Bruce Edmonds http://bruce.edmonds.name Centre for Policy Modelling http://cfpm.org Modelling Routines Website (where the slides will be): http://modellingroutines.wordpress.com


Download ppt "Projection and the Reality of Routines – reflections of a computational modeller Bruce Edmonds Centre for Policy Modelling Manchester Metropolitan University."

Similar presentations


Ads by Google