Expert Networking and International Governance: Questions of Democracy

Networks of experts coordinated or orchestrated by international bodies have become so widespread and influential that they are said to shape a new world order. Standards for consumer safety, investor protection, and environmental sustainability are governed by appeals to the epistemic authority of experts. Typically, formal international organizations orchestrate cross-border constellations of public–private collaborations between groups that are deemed to have relevant knowledge. This trend is part of a depoliticization of decision-making; policy issues are framed as technical problems that should be kept at a distance from party politics. The question here is how to conceptualize and assess this development in democratic terms. In political theory, three kinds of approach have evolved in response to this trend. At one extreme, the argument is that governance beyond the state cannot be legitimate until it has implemented modes of representation and contestation familiar from the domestic context. At the other extreme, the argument is that legitimacy beyond the state should be decoupled from democratic concerns and be legitimated on technocratic grounds. Between these two poles is the argument that democracy does not have to resemble the domestic model in organizational terms and can fruitfully be reconceived or reinterpreted in the international context. Versions of the reinterpretive approach are currently popular under different theoretical labels. It is fruitful to use it as a model for considering questions of democratic legitimacy for the expert networks that constitute or interact with international organizations. In following the reinterpretive route, a natural starting point is to consider what the key evaluative dimensions of democracy are. At an abstract level, democracy is about three main considerations: 1. Authorization: The people are the rightful principals of public action. It is necessary to consider how people can be empowered to challenge and potentially veto opinions that flow from expert networks. 2. Attitude: Democratically justified institutions express the right kind of concern for people as equals. There are important questions about how the technical rationalities of expert networks can show consideration for a reasonable pluralism of perspectives and how “soft law” can address subjects with appropriate respect for citizens’ claim to justification and rule of law. 3. Area: The authority of democratically legitimate institutions must be matched by a defined sphere of answerability. For the area of expert networks, this issue concerns both the scope of expert mandates and whether there is a fit between mandate and actual practice. The task for an assessment of the democratic legitimacy of expert networks is to consider more fully what each of these evaluative dimensions imply in the relevant context.

LinkIt: A Tool for Assessing and Providing Formative Feedback for Structural Knowledge

It is challenging to assess the effectiveness of learning and training. Most evaluators rely on multiple choice, fill-in-the-blank, and essay questions. These are time-consuming, provide little formative value, and provide no visualization of the knowledge domain. LinkIt uses constrained concept mapping to elicit, score, represent visually, and provide immediate feedback on student knowledge networks by comparing them to expert networks.

Who are behavioural public policy experts and how are they organised globally?

Behavioural public policy has spread internationally over recent years. Worldwide, expert units are translating insights from behavioural sciences into policy interventions. Yet, behavioural expert networks are a puzzling case. They seem to oscillate between two modes of collective action: as an epistemic community, they are based on the consensual belief that biases in behaviour pose a problem for policymaking. As an instrument constituency, they bring together a diversity of actors, unified not by consensual beliefs about problems but by practices of promoting behavioural instruments as solutions. Drawing on a review of literature, this article provides a systematic analysis of the relation between epistemic communities and instrument constituencies. It argues that there has been an ‘agency shift’ from one mode to the other. The implications are that experts should be aware of the fact that the instruments they are proposing might develop a political life of their own.

An Approach to Changing Competence Assessment for Human Resources in Expert Networks

An expert network is a community of professionals in a specific field, united by an information system, in which different tasks are solved. One of the main tasks in expert networks is the selection of specialists with specified competencies for joint problem solving. The main characteristic of an expert network member is a set of competencies, which includes both functional aspects and personal qualities. For this reason, the procedure for selecting specialists and ranking them is critical. Such a procedure uses specialists’ competence assessments from the expert network. If these assessments are out of date, the project results can be unsuccessful. This article proposes an approach aimed at automating the assessment of the specialists’ competencies based on the projects results. This approach consists of a reference model and an algorithm of competence assessment change for human resources. The paper also includes an algorithm evaluation on generated data.

MS-NET: modular selective network

We propose a modular architecture of Deep Neural Network (DNN) for multi-class classification task. The architecture consists of two parts, a router network and a set of expert networks. In this architecture, for a C-class classification problem, we have exactly C experts. The backbone network for these experts and the router are built with simple and identical DNN architecture. For each class, the modular network has a certain number $$\rho$$ ρ of expert networks specializing in that particular class, where $$\rho$$ ρ is called the redundancy rate in this study. We demonstrate that $$\rho$$ ρ plays a vital role in the performance of the network. Although these experts are light weight and weak learners alone, together they match the performance of more complex DNNs. We train the network in two phase wherein, first the router is trained on the whole set of training data followed by training each expert network enforced by a new stochastic objective function that facilitates alternative training on a small subset of expert data and the whole set of data. This alternative training provides an additional form of regularization and avoids over-fitting the expert network on subset data. During the testing phase, the router dynamically selects a fixed number of experts for further evaluation of the input datum. The modular nature and low parameter requirement of the network makes it very suitable in distributed and low computational environments. Extensive empirical study and theoretical analysis on CIFAR-10, CIFAR-100 and F-MNIST substantiate the effectiveness and efficiency of our proposed modular network.