Invitation to tender: Development of an agent-based model of the data economy

Mon Sep 10, 2018

Experienced organisation or consortium of organisations to help the ODI scope and develop a prototype (toy) agent-based model of the data economy

10 September 2018
Tender reference: RDP11-001
Call for tenders by the Open Data Institute
Contact: [email protected]

Please note: due to the volume of applications, the earliest we will be able to respond to you following the internal review is 15 October.

Summary and timeline

Aim

The objective of this work is to help us understand what is possible with agent-based models, test the feasibility of applying them in our work and understand how they can effectively be communicated to key stakeholders. We intend to explore these questions by engaging a delivery partner(s) to produce a specification and working prototype of a ‘toy’ agent-based model of the data economy (or parts of)  that can be used to simulate the response of the data economy to the implementation of different data policies.

By data economy, we mean that part of the UK’s economy in which government, companies and organisations ’leverage data and analytics to improve operational efficiencies, to make better tactical and strategic decisions, and to create innovative products, services and business models… It is also a connected economy in which partners and competitors alike share data and integrate business processes where the resulting benefits to overall markets, the enterprises themselves and customers outweigh the risks of such collaboration.’

By a ‘toy’ model we mean a model that are abstract in nature which focuses more on broad structure, rules and interactions of key agents rather than a model that is driven by large databases orientated to make the model ‘real’ as far as possible.

The successful department, organisation or consortium of organisations will work in collaboration with the Open Data Institute (ODI) who will provide input, guidance and review throughout.

Background

As part of our innovation programme, we are exploring the feasibility of new methods to help policy makers make better decisions around data. This includes the recently launched second UK Tech Innovation Index, a project carried out by the DataCity and supported by the ODI, that looked at novel techniques to identify clusters of innovative UK businesses.

We have recently launched a new project, focusing on whether it is possible to simulate the effects of different data policy interventions on the data economy using agent-based models. To date we have started by conducting some initial research, held conversations with industry experts about how we might proceed and internal and external debate about what policy questions we would like to test.

Through working with a partner organisation or organisations, we hope to improve our understanding of what is possible with agent-based models and use external expertise to test the feasibility of applying them in our work. We are also keen to understand whether they can be useful for policy-makers in making decisions, particularly in how the results can be analysed and communicated effectively.

Audience

The primary audience for this piece of work is (UK) government policy makers with responsibilities for data policy. The secondary audience is the ODI and similar organisations who provide advice to policy makers about the data economy.

Deliverables

We want to better understand agent-based models and test their feasibility in application to data policy-making, through practical experience. As such, the deliverables in this tender revolve around two distinct practical phases: developing a model specification/behaviour rules, and developing a prototype toy model based on this specification.

We recognise that our project timeframes and objectives are ambitious, and as such, are flexible on the level of emphasis delivery partner(s) wish to place on each phase. We encourage proposals that focus on helping us to achieve internal ODI and stakeholder learning, alongside the delivery of a useable specification and prototype.    

Phase 1: developing a model specification

The first stage of the project will revolve around creating a detailed specification of the agent-based model, including identifying specific questions, a description of the agents, the rules for behaviour and identifying relevant data sources. This process would be carried out in close collaboration with the ODI, building on our expertise and the scoping work we have done so far on the project – in terms of questions and model scoping.

We would also be keen to receive proposals that involve ways of collaboratively creating rules for models based on wider input from our stakeholders – through workshops and surveys, for instance. The output of this work can be in any suitable, suggested format that can be published under an open licence on the web. The audience for this output is the ODI itself, similar organisations, policy-makers and others looking to use agent-based approaches to inform policy making.

Phase 2: developing a prototype toy model

The second stage of the project will focus on building a prototype or MVP toy agent-based model based on the specification from phase 1. We expect the delivery partner(s) to build, calibrate and validate the model based on the specification, and deliver a prototype which can be independently run by the ODI after completion of the project. Given that we anticipate a low level of available quantitative data, we anticipate validation of the model will be through a group of experts, rather than against a data source.

The delivery partner(s) will work closely with the ODI to explore how the outputs and insights generated by the model can be disseminated and communicated to key stakeholders in an accessible and useful manner. All code, outputs and data generated in phase 2 are expected to be published under open licences to allow others to reuse and build upon the work done.

Given the timeframes and objectives, we are keen to receive suggestions on the best approach to deliver a viable prototype within these constraints – for example using an agile methodology, using existing software frameworks or models, or prioritising certain aspects of development.

Activities

Through both phases of the project delivery partner(s) will be required to work closely with the ODI team. We have internal capacity to identify and reach out to research participants, host and facilitate events, and disseminate outputs, and would expect delivery partner(s) to use this capacity to help deliver this ambitious project. The ODI will take responsibility for identifying and convening relevant stakeholders, to enable the delivery partner(s) to focus on other areas.

Within any proposal, we would be keen to understand the details about how the chosen delivery partner(s) would work closely with the ODI, collaboratively with relevant stakeholders and openly. This may include co-location at the ODI’s offices in London, regular face-to-face meetings, stakeholder workshops and being available remotely (eg skype, email, slack).

Duration

8 October 2018 – 8 February 2019

Value of award (excl. VAT)

Up to £40,000

Questions to ODI by

5pm BST on Wednesday 19 September 2018

ODI responses by

5pm BST on Friday 21 September 2018

Costed proposals due by

5pm BST on Friday 28 September 2018

Tender decision by

5pm BST on Monday 15 October 2018

Contract awarded

5pm BST on Friday 19 October 2018

ODI brief winning applicant(s)

Week of 21 October 2018

ODI progress reviews

Biweekly or as needed

Final work delivered by

8 February 2019

Tender reference:

RDP11-001

Contact

[email protected]

Terms of payment

50% of the agreed value of the award will be paid on completion of the first phase of work (delivery of the specification), and the remaining 50% will be paid upon completion of the work, including satisfactory responses to all feedback from the ODI.

Form of tender response

Interested parties should submit a costed proposal (in English) to [email protected] which includes:

  • the tender reference in the email subject line
  • a short explanation of your proposed approach (eg methodology, scope), and why you are well-placed to do the work (no more than 4 pages)
  • a description of the team who will do the work, including bios (no more than 1 page)
  • the costing should be at activity level, but feel free to provide more detail (no more than 2 pages)
  • a writing sample of a similar specification and/or output from similar model

If you have any questions about the tender, please contact [email protected] quoting the tender reference. The ODI reserves the right to make both anonymised questions and answers public or shared with other organisations having stated their interest.

Decision criteria

All proposals will be assessed as described in our public procurement policy. In addition, for this procurement we will be looking for:

  • Commitment to enabling and guiding the ODI to better understand agent-based modelling.
  • Evidence of a well thought out approach to delivering this ambitious project under the constraints identified.
  • Ability to communicate outputs from complex models to non-technical audiences.