Jens Riegelsberger 1 M. Angela Sasse John D. McCarthy 2 November 2006 University College London Human-Centred Systems Group Department of Computer Science A Framework for Trust Applied to Social Computing Current affiliations: 1 Google UK, 2 LBI UK
Feb 12, 2016
Jens Riegelsberger1
M. Angela SasseJohn D. McCarthy2
November 2006
University College LondonHuman-Centred Systems GroupDepartment of Computer Science
A Framework for TrustApplied to Social Computing
Current affiliations: 1Google UK, 2 LBI UK
Trust Research in HCI> Trust has gained prominence in HCI research
> Trust in web sites (e-commerce: Egger, Sapient, Corritore et al.)
> Trust in humans (virtual teams, online advice: Olson et al., Swerts et al.)
> … Trust in ambient technologies(CHI 04 Workshop)
Dis-embeddingInteraction is stretched over time and space and involves complex socio-technical systems [Giddens, 1990]
… pervasive in modern societies
(e.g. catalogue shopping)
‘Lack of Trust’ online
More risk> Privacy (more data required)> Security (open system)> Own errors …
More uncertainty> Inexperienced with decoding cues> Less surface cues are available> Cues might have no significance (“anyone could set up a good-looking site”)
Symbols vs. Symptoms
Relevance of Trust Research> Trust is of high importance for individuals and society
> Mediating interactions can result in lower trust (widely publicised ‘lack of trust’ online)
> Effect may be partially due to lack of familiarity, …
> but trust is an ongoing concern in mediated interactions:• Lack of central control• More explicit information required• Decreased number of trust signals
Trust Background> Only required in the presence of risk and uncertainty> “… willingness to be vulnerable based on positive expectations about the actions of others” [e.g. Rousseau et al., 1998; Corritore et al., 2001]
> Based on assessment of ability and motivation [Deutsch, 1956]
> Assessment can result in cognitive trust [Lahno, 2002, Lewis & Weigert 1985]
> But human trust-decisions are also based on immediate pre-rational affective reactions [Corritore et al., 2000, Lewis & Weigert 1985]
TRUSTEETRUSTOR
1 Signals
TRUSTEETRUSTOR
1 Signals
TRUSTEETRUSTOR
Separation in Space
+ UNCERTAINTY
Outside Option 1 Signals
TRUSTEETRUSTOR
2a Trusting Action2b Withdrawal RISK
Separation in Space
+ UNCERTAINTY
Outside Option 1 Signals
TRUSTEETRUSTOR
2a Trusting Action2b Withdrawal
3a Fulfilment 3b Defection
RISK
Separation in Space
+ UNCERTAINTY
Outside Option 1 Signals
TRUSTEETRUSTOR
2a Trusting Action2b Withdrawal
3a Fulfilment 3b Defection
RISK
Separation in Time
+ UNCERTAINTY
Separation in Space
+ UNCERTAINTY
Current HCI Trust Research
> Focused on increasing trust (overcoming ‘lack of trust’)> … but well-placed trust is of high importance
for long-term acceptance
> Largely concerned with cognitive trust> … but trust is also based on affective reactions to
interpersonal cues
> No coherent theoretical foundation> Measurement often based on self-reports and Prisoner’s Dilemma studies
Framework for Trust > 1. Step: Focus on incentives for trustworthy behavior> 2. Step: Identify signals
>> Design guidelines for systems that allow well-placed trust
> Draws on work by >> Bacharach & Gambetta (Dept. of Sociology, Oxford)
>> Raub et al. (ISCORE, Utrecht)
> Why should a trustee ever fulfill?> Intrinsic and Contextual Properties
>> …. support trustworthy action>> …. signal trustworthiness
Outside Option 1 Signals
TRUSTEETRUSTOR
2a Trusting Action2b Withdrawal
3a Fulfilment 3b Defection
RISK
Separation in Time
+ UNCERTAINTY
Separation in Space
+ UNCERTAINTY
TRUSTEETRUSTOR
TRUSTOR TRUSTEE
Temporal
Institutional
Social
ContextualIncentives
Trust
Context Signal Incentive
ContextualProperties
TRUSTOR TRUSTEE
Temporal
Institutional
Social
ContextualIncentives
Trust
IntrinsicProperties
Context Signal Incentive
IntrinsicProperties
TRUSTOR TRUSTEE
Temporal
Institutional
Social
ContextualIncentives
Trust
Ability
Context Signal Incentive
IntrinsicProperties
InternalisedNorms
Benevolence
Types of Trust
TRUSTOR TRUSTEE
> Expectation of futureencounters is an incentivefor trustworthy behaviour> A defrauded trustor might withdraw from future interactions or retaliate
> Past experience gives information about trustor’s personal properties
RequirementsStable identities, traceability of outcomes to actors and actions
ContextualIncentives
Trust
Costs
Benefits
IntrinsicProperties
Benevolence, Internalised
Norms
Temporal
Temporal
Social
Institutional
TRUSTOR TRUSTEE
> Trustors share informationabout trustees’ past behaviour in the form of reputation> Reputation is a ‘hostage’ in the trustor’s hand
Requirements> Reliable & unbiased reputation aggregation > Incentives for contributingreputation information> Shared understandingof cooperation & defection
ContextualIncentives
Trust
Costs
Benefits
IntrinsicProperties
Benevolence, Internalised
Norms, Ability
Social
Temporal
Social
Institutional
TRUSTOR TRUSTEE
> Law, ContractsRequirements: > A priori definitions … little flexibility> Low cost of investigation and punishment comparedto risk
> Organisations, brands> Trust is vested in roles rather than individualsRequirements: > Clear and reliable signs of institutional affiliation
ContextualIncentives
Trust
Costs
Benefits
IntriniscProperties
Benevolence, Internalised
Norms, Ability
Institutional
Temporal
Social
Institutional
Questions of trust in social networking sites
> The site as trustee> Respect for privacy (no selling on of personal information, no spam)> Ability to keep personal data secure> Stay in business (to make initial effort of registering worthwhile)
> Other users / people as trustees> Respect for boundaries of context> Respect for norms of conduct
USER SITE
Institutional• Legal systems • Consumer protection agencies
Temporal• Interest in repeat business• Depending on presenceof alternative trustees
Social• Depending on presenceof alternative trustees
ContextualIncentives
Trust
Costs
Benefits
IntrinsicProperties
Technical Competence,
Corporate ethics
Contextual Properties
USER OTHER USER
InstitutionalLegal systems Site’s code of conduct
TemporalInterest in repeat interaction
SocialReputation system
ContextualIncentives
Trust
Costs
Benefits
IntrinsicProperties
Induction into norms of forum
ContextualProperties
Jens Riegelsberger [[email protected]]
M. Angela Sasse [[email protected]]
John D. McCarthy [[email protected]]
http://www.cs.ucl.ac.uk/staff/jriegels
University College LondonDepartment of Computer Science
Trust in Mediated Interactions