Values First Software Engineering

Lead Research Organisation: Lancaster University
Department Name: Computing & Communications

Abstract

BACKGROUND
Values are deeply held principles guiding decision-making processes of individuals, groups and organizations. Software is inevitably affected by values: the organizational values of the project sponsor, research partners, and developers. Some values (e.g. financial value) are easier to quantify than others (e.g. trust, responsibility) with the latter often dismissed in software production processes as lacking of measurable evidence. This is problematic because all values, including less-easy to measure ones, influence how people use, access and engage with software systems with far-reaching impact not only on the commercial success of software products, but more widely on society.

VISION
Values-First SE is a systematic and disciplined approach to the elicitation, articulation, and deliberation of human values in software production. Given the pervasiveness of software and its impact on society, we - developers, researchers, clients, and end-user - must strengthen our capacity and confidence to externalise the values-sets built into software and use them to track how software behaves. Recent examples such as the Google boycott stem from the (often unintentional) breach of implicitly held values-systems: simply put, companies do not want to be associated with extremist values perceived as opposite to those held by society. However, the interplay between values held by software industry (e.g. prestige, social responsibility), clients (e.g. financial, care for their customers and employees), and end-users (e.g. trust, social justice) is complex, difficult to articulate and rarely fully captured by current SE decision-making processes.

CONTEXT
Awareness of the impact of software on politics, society, and the environment is not new: from cyber-security to environmental informatics, to digital-health, a large body of ethical computing exists looking at mechanisms that can guide developers and managers' responsibilities (e.g. code of Ethics). What is new is the unprecedented scale, reach and complexity of such impact and the urgency for developers to "be prepared to be responsible".

One of the biggest challenges for developers is that the full impact of values choices in the code developed is often unseen and unintentional: when writing software, often the platform obfuscates the process even to the software developer. How can software developers be prepared to be responsible when it is not clear what they should be responsible for? For example, in the Android Software Development Kit (SDK), the geocoding of location is done by sharing of precise location with a third party organization (including Google). The implications of sharing this data, how it is stored, treated and reused is not fully explained to the developer in tutorials, in the SDK documentation or in the IDE at the time of writing the code.

Those simple lines of code for the developer could have an unseen impact on the encoded values of the software produced. However, current SE methods do not seem to provide the means to follow values through a software development process and, in particular, there are currently no methods that allow values to be used as a reference framework for decision making at key stages of software development. Articulating, negotiating, and capturing human values across SE decision-making processes is precisely Values-first SE main aim.

Planned Impact

OVERARCHING GOAL
To lay the foundations for an international and diverse community of practice by attracting industry and international research partners, together with national and local cultural institutions around a common intent: the production of values-conscious software. Such common intent is actioned through the systematic investigation of concepts, methods and techniques that industry case-studies find to be effective in eliciting, documenting and negotiating values throughout SE decision-making processes.

This project will support such community of practice and strengthen its capacity through flagship use-cases from which to draw lessons learned and guidelines for university teaching, professional training, outreach and school education.

Values-First SE has the potential to change the way decisions are taken during software production and the way software development is taught. This three-strand impact strategy articulates these potentials:

STRAND 1: IMPACT ON INDUSTRY
Values-First SE potential for innovation is significant in supporting software industry in finding new and effective ways for managing the social, political, and environmental responsibilities of the software that they develop. This strand focuses on the industrial exploitation of research results by working in close partnership with our business partners.
(Key participants: Zühlke Engineering, IYWTO/Cleanweb).

STRAND 2: IMPACT ON RESEARCH & FUNDERS
The overarching goal of this strand is to trigger a cultural shift towards how values are considered in software production; this has the potential to influence future research agendas in the EPSRC Digital Economy (DE) and SE research programmes.
(Key participants: DesiRE/IRISA, Monash, TUW, UofT)

STRAND 3 IMPACT ON OUTREACH/EDUCATION
Given the crucial role of new generations in responding to technology shaping societal changes it is particularly important to translate research into outreach and educational activities therefore promoting science to cultural institutions and the general public.
(Key participants: LUSU, SMG)
 
Description The role of human values in SE decision-making processes is widely acknowledged, but so too is the challenge of researching values in a systematic way.

Our main contribution to the field has been the identification and development of a theoretical framework respond to this challenge, a range of tools designed in accordance with this framework, and a collection of case studies using such tools reaching out to more than two-hundred software practitioners from industry, research, and education. These tools use a combination of quantitative, qualitative and design thinking approaches and have been designed to be both robust and replicable, and creative and provocative - our findings have been published in top SE venues (e.g. CSCW2018, ESEM2018, CHASE2019, and ICSE-NIER 2019)

For example, one of our tools, the Values-Q-sort (V-CS), uses an established but little known (in SE) mixed-method to structure interviews around specific values statement and capture how different values shape SE technical decisions. In practice, for each Q-Sort statement, a number of 'how-to' practical examples can be elicited and documented. For example, in algorithms deciding which job adverts to show, a good practice associated with the V-QS statement "It is important to me that I do not discriminate against others when developing software" would be to ensure there is no gender bias influencing which adverts are shown, resulting in a potential trade-off with cost effectiveness.

We developed the V-QS by mapping the principles of the Code of Ethics for Software Engineers (drafted by the Association of Computing Machinery - ACM) to Schwartz Universal Values, an established values model drawn from Social Psychology. We found that the ethics code provides a body of language relatable for SE practitioners. Emerging findings not only show the significance of values as distinguished from, though connected to, ethics; but also how different meanings that the same value has for each software practitioner and different ways in which they may translate into code.
Exploitation Route We have received (and accepted) several invitations to run workshop and train practitioners in the use of our tools (e.g. the values q-sort). We have made the designs available on request; the next step is to further validate the tools and complement them with more detailed documentation and full on-line access.
Sectors Communities and Social Services/Policy,Digital/Communication/Information Technologies (including Software),Education,Environment,Healthcare,Government, Democracy and Justice,Culture, Heritage, Museums and Collections,Security and Diplomacy

URL http://www.valuesincomputing.org
 
Description Values-First SE (VFSE) is about understanding how human values work in software production. We investigate both easier to measure values such as the commercial success of a software product and more difficult to capture values such as fairness, trust and 'public good'. Over the past twelve months we have engaged with more than two hundred software practitioners from industry, research and education. Our work has helped to identify the tensions within and between software development teams, stakeholders and end-users. By doing so, our research may help to anticipate technical decisions that could lead to breaches of moral standards and interfere with the very public good that it wishes to support. IMPACT IN EDUCATION: We work at a public University, we have a duty of care towards our future generations, and we take this duty seriously. For us education means: "to prepare students to think critically, knowledgeably, reflectively, and empathically " (The Denver Manifesto, 2017 - h http://tinyurl.com/denvermanifesto ) To this end, we have adapted and used our research tools in educational contexts. These include undergraduate courses in software engineering, workshops for doctoral candidates in cybersecurity, and international summer school sessions on Artificial Intelligence & Ethics. We have recently being invited to run a session on AI and Values with 6th form students early this July. EXAMPLE OF IMPACT IN EDUCATION: The Values Q-Sort has been used in the Software Design Studio at Lancaster University as part of a reflective, practice-based software design module for undergraduate students in their 2nd year of their BSc(Hons) in Software Engineering. The resulting values statements were included in the technical reports and submitted for assessment. The exercise was used to spark reflection on team values and was used by each project team to include their own values statement in their project plan. In addition, We modified the Values Q-Sort to apply to Artificial Intelligence (using the Asilomar AI Principles as a basis for the statements). This was used at the European Forum in Alpbach, Austria (EFA2018) with 80 undergraduate and postgraduate students from a variety of disciplines as well as in other research contexts, such as the CIRN2018 workshop, Prato, Italy and with academic staff at Monash University, Melbourne, Australia. IMPACT IN INDUSTRY: We are a small agile team and, not only we have worked with our original industry partners, but have also established a strong collaboration with the digital team of a large UK-based co-operative enterprise. Together, we have designed a 4-week digital sprint that is currently on-going. Activities have included: - Taster & ThinkUp sessions - Q-Sort sessions (teams and/or individual) - On-line surveys - In situ observations - In depth Interviews - Team workshops - Facilitated focus groups A final report and a formal presentation of results will be arranged for the end of April / start of May 2019
First Year Of Impact 2018
Sector Digital/Communication/Information Technologies (including Software),Education,Other
Impact Types Cultural,Societal

 
Description M. McLuhan Centenary Fellowship in Digital Sustainability
Amount $10,000 (CAD)
Organisation University of Toronto 
Sector Academic/University
Country Canada
Start 07/2018 
End 06/2019
 
Description Seed Funding: from EPSRC LWEC Senior Fellowship (Prof Blair)
Amount £3,500 (GBP)
Organisation Lancaster University 
Sector Academic/University
Country United Kingdom
Start 01/2019 
End 03/2019
 
Description SpriteHub - Security Privacy Identity Trust in the Digital Economy 
Organisation University of Manchester
Country United Kingdom 
Sector Academic/University 
PI Contribution Access to national and international network of security / privacy scholar
Collaborator Contribution Networking activities
Impact No Outputs yet
Start Year 2019
 
Description UKRI CDT in Artificial Intelligence, Machine Learning and Advanced Computing 
Organisation Swansea University
Country United Kingdom 
Sector Academic/University 
PI Contribution Invited to join the UKRI CDT Steering Committee
Collaborator Contribution Feedback on Strategy
Impact No outputs yet (Joined in Nov 2019), contribution to draft EDI strategy
Start Year 2019
 
Title Values First SE Tools and Techniques 
Description A family of new tools and techniques developed as part of the project aimed at the articulation, mapping and tracking of human values in software production decision making process. These tools have been designed for and used with software practitioners in research, education and industry. Description of such tools and emerging results have been published in peer-reviewed world leading SE venues (e.g. CSCW2018, ESEM2018, CHASE2019, ICSE-NIER2019). 
Type Of Technology New/Improved Technique/Technology 
Year Produced 2018 
Impact We are at the early stage of the use of such tools but so far more than 200 software practitioners have been using a selection of such tools. For further details please refer to the paper: Winter, ER, Forshaw, SW, Hunt, L & Ferrario, MAFC 2019, Towards a Systematic Study of Values in SE: Tools for Industry and Education. In Proceedings of the 41st International Conference on Software Engineering - New Ideas and Emerging Results track (ICSE-NIER2019). IEEE/ACM, New York. 
URL http://eprints.lancs.ac.uk/131669/1/ICSE_NIER_accepted_preprint.pdf
 
Description 2 x Values in Computing Masterclasses 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Undergraduate students
Results and Impact 2 x 2hr masterclasses on values in computing delivered to first and third year students on BSc in Computer Science, Huddersfield University
Year(s) Of Engagement Activity 2019
 
Description 4 x 'Research Jams' on Values in Computing with University of Toronto PhDs students and staff (Feb, March, May 2018 and Feb 2019) 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Postgraduate students
Results and Impact Suggested, co-organised and facilitated 4 research jams with PhD students and faculty staff from iSchool and Comp Science Dept at the University of Toronto - project partner and sponsor of my Fellowship in Digital Sustainability (DCI / iSchool). This event series reached around ~30 staff and students.
Year(s) Of Engagement Activity 2018,2019
 
Description A week-long seminar on AI & Ethics (European Forum Alpbach 2018) 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Undergraduate students
Results and Impact Designed and delivered a week-long seminar on Artificial Intelligence and Ethics - done in collaboration with Prof Peter Kirchslagher, Lucerne (Theology and Ethics). Each session was three hour long and attended by 80 students.

The overall aim of the seminar week was to understand, reflect, articulate and deliberate on the Ethical implications of AI for society. A jointly written draft for an AI & Ethics Manifesto was co-produced during the seminar as key tangible outcome and learning process reference point.
Year(s) Of Engagement Activity 2018
URL https://www.alpbach.org/en/session/sem18-seminar-08-artificial-intelligence-and-ethics-08-16-09-30/
 
Description AI - A Tricky Game 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact A panel debate organised by Austria AI society and Forbes - I was one of the pannelist.
Year(s) Of Engagement Activity 2018
URL https://www.werkstaette-wattens.at/de/programm/2018-08-17-state-of-ai
 
Description AI and Values 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact This was an invited workshop held at the Faculty of IT, Monash University, Australia. The workshop attracted 25 participants - mainly faculty staff and PhD researchers. The aim was to take a deep dive into human values, examine how they work, and what structures they may exhibit. Specifically, our twofold objective was to capture the diversity of meanings for each value and their interrelationships in the context of AI. I did do so by using some of the tools and techniques developed as part of the Values in Computing (ViC) research. Further collaborations and meet up are planned in 2019, particularly in relation to the Dagstuhl seminar on Values in Computing on 14-19th July which I co-organise.
Year(s) Of Engagement Activity 2019
 
Description Industry Seminar: Human Values in Software Production, IBM, Rio De Janeiro 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Values in Computing is about understanding how human values work in software production. We investigate both easier to measure values such as the commercial success of a software product and more difficult to capture values such as fairness, trust and 'public good'. Our work helps to identify the tensions within and between software development teams, stakeholders and end-users. By doing so, our research may help to anticipate technical decisions that could lead to breaches of moral standards and interfere with the very public good that it wishes to support.
Year(s) Of Engagement Activity 2019
 
Description Invited Speaker at Not-Equal Summer School, Swansea University 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Postgraduate students
Results and Impact A range of Phd Students, practitioners, Scholars and third sectors representative attends the talk as part of the Not-Equal Summer School (UKRI funded Network-plus). this sparked conversation and plan for future collaborations
Year(s) Of Engagement Activity 2019
URL https://not-equal.tech/not-equal-summer-school/
 
Description Invited Speaker: Coop Digital All Hands, Federation House, Manchester 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact ViC was invited to present research work carried out with Coop digital, this was present to all Coop-digital staff present at the event and invited school pupils, feedbakc from the presentation was very positive
Year(s) Of Engagement Activity 2019
 
Description Invited Talk - Values in Computing (PERCCOM Summer School, Lappeenranta, University of Technology, Finlad) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Postgraduate students
Results and Impact This talk introduces some of the tools and emerging findings from Values in Computing work at Lancaster University, UK. A broader perspective on the issues connecting tech industry, academic research, and governance framed the conversation as well as provocations from the current state of affairs and the metaphysical roots of the binary system.
Year(s) Of Engagement Activity 2018
URL http://codecamp.fi/doku.php/perccom_summer_school_2018
 
Description Invited Talk - Values in Computing - Connecting the Bits 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Postgraduate students
Results and Impact The talk introduced some of the tools and emerging findings from latest " Values in Computing" work at Lancaster University, UK and from activities carried out at UofT as part of the my Fellowship.
Year(s) Of Engagement Activity 2018
URL http://dci.ischool.utoronto.ca/2018/04/24/values-in-computing-connecting-the-bits/
 
Description Invited Workshop - Alien AI: The Criterion Of Intelligence 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Postgraduate students
Results and Impact More than 20 participants (PhD "TIPS' students - , Trust Identity Privacy and Security), practitioners, and people from industry took part to this workshop reflecting, sharing knowledge and discussing actions in regards to future developments of AI, privacy and security.
Year(s) Of Engagement Activity 2018
URL http://pactman.uk/building-a-community-of-uk-tips-researchers-2/
 
Description Invited panellist and speaker: Ethics and Algorithmic Mediation of Social Processes (EMAPS), PUC Rio, Brazil 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Postgraduate students
Results and Impact Invited to speak about their work at PUC-Rio's research group in Ethics and Algorithmic Mediation of Social Processes (EMAPS) and to join a panel of experts in data science, semiotic, psychology, and law to reason about the ethical and human values aspects of the design, development, use and management of data-intensive and intelligent systems.
Year(s) Of Engagement Activity 2019
URL http://www.hcc.inf.puc-rio.br/EMAPS/?II_Workshop_2019___English
 
Description Invited talk: Coop Digital Presents..., Federation House, Manchester 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact A lively debate about values in software producation and ethics in the workplace with software professional
Year(s) Of Engagement Activity 2019
URL https://www.eventbrite.co.uk/e/co-op-digital-presents-human-values-in-software-production-tickets-73...
 
Description Seminar: Understanding & Measuring Human Values in SE, Leicester University 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Postgraduate students
Results and Impact Knowledged and practice exchange with colleagues and students at the Computing Science department at the University of Leicester with focus on human aspects in SE.
Year(s) Of Engagement Activity 2019