No abstract

Though designers must understand systems, designers work differently than scientists in studying systems.  Design engagements do not discover whole systems, but take calculated risks between discovery and intervention. For this reason, design practices must cope with open systems, and unpacking the tacit guidelines behind these practices is instructive to systems methodology. This paper shows that design practice yields a methodology which applies across forms of design.  Design practice teaches us to generate ideas and gather data longer, but stop when the return on design has diminished past its cost.  Fortunately, we can reason about the unknown by understanding the character of the unbounded.  We suppose that there might as well be an infinite number of factors, but we can reason about their concentration without knowing all of them.  We demonstrate this concept on stakeholder systems, showing how design discovery informs systems methodology. Using this result, we can apply the methods of parametric design when the parameters are not yet known by establishing the concentration of every kind of factor, entailing a discovery rate of diminishing returns over discovery activities, allowing the analysis of discovery-based trade-offs.  Here, we extend a framework for providing metrics to parametric design, allowing it to express the importance of discovery.

No abstract

Probabilistic programming is a programming language paradigm receiving both government support [1] and the attention of the popular technology press [2]. Probabilistic programming concerns writing programs with segments that can be interpreted as parameter and conditional distributions, yielding statistical findings through nonstandard execution. Mathematica not only has great support for statistics, but has another language feature particular to probabilistic language elements, namely memoization, which is the ability for functions to retain their value for particular function calls across parameters, creating random trials that retain their value. Recent research has found that reasoning about processes instead of given parameters has allowed Bayesian inference to undertake more flexible models that require computational support. This article explains this nonparametric Bayesian inference, shows how Mathematicaʼs capacity for memoization supports probabilistic programming features, and demonstrates this capability through two examples, learning systems of relations and learning arithmetic functions based on output. ‡ Nonparametric Bayesian Inference Bayesian statistics are an orderly way of finding the likelihood of a model from data, using the likelihood of the data given the model. From spam detection to medical diagnosis, spelling correction to forecasting economic and demographic trends, Bayesian statistics have found many applications, and even praise as mental heuristics to avoid overconfidence. However, at first glance Bayesian statistics suffer from an apparent limit: they can only make inferences about known factors, bounded to conditions seen within the data, and have nothing to say about the likelihood of new phenomena [3]. In short, Bayesian statistics are apparently withheld to inferences about the parameters of the model they are provided.

This chapter provides a stakeholder discovery model for distributed risk governance suitable to machine learning and decision-theoretic planning. Distributed risk governance concerns when the underlying risk is not localized or has unknown locality so that any initial interaction with stakeholders is limited and educational and participatory initiatives are costly. Therefore, expecting the initial reaction to communications is critical. To capture this initial reaction, the authors sample the population of potential stakeholders to discover both their concerns and knowledge while handling inaccuracies and contradictions. This chapter provides a stakeholder discovery model that can accommodate these inconsistencies. Stakeholder discovery provides a timely strategic assessment of the risk situation. This assessment forecasts projected stakeholder actions to find if those actions are in line with their strategic interests or if there are better choices using reinforcement learning. Unlike other reinforcement learning formulations, it does not take the state space, criteria, potential observations, other agents, actions, or rewards for granted, but discovers these factors non-parametrically. Overall, this chapter introduces machine learning researchers and risk governance professionals to the compatibility between non-parametric models and early-stage stakeholder discovery problems and addresses widely known biases and deficits within risk governance and intelligence practices.