In conversation with the very excellent RunCoCo project at Oxford University last Friday, I revisited a question which will, I think, prove central to my current research – establishing trust in an online archival environment. This is an important issue both for community archives, such as Oxford’s Great War Archive, as well as for conventional Archive Services which are taking steps to open up their data to user input in some way – whether this be (for example) by enabling user comments on the catalogue, or establishing a wiki, or perhaps making digitised images available on flickr.
A simple, practical scenario to surface some of the issues:
An image posted to flickr with minimal description. Two flickr users, one clearly a member of staff at the Archives concerned, have posted suggested identifications. Since they both in fact offer the same name (“Britannia Mill”), it is not immediately clear whether they both refer to the same location, or whether the second comment contradicts the first.
Which comment (if either) correctly identifies the image? Would you be inclined to trust an identification from a member of staff more readily than you’d accept “Arkwright”‘s comment? If so, why? Clicking on “Arkwright”‘s profile, we learn that he is a pensioner who lives locally. Does this alter your view of the relative trustworthiness of the two comments (for all we know, the member of staff might have moved into the area just last week)? How could you test the veracity of the comments? Whose responsibility is this? If you feel it’s the responsibility of the Archive Service in question, what resources might be available for this work? If you worked for the Archive Service, would you feel happy to incorporate information derived from these comments into the organisation’s finding aids? Bear in mind that any would-be user searching for images of “Britannia Mills” – wherever the location – would not find this image using the organisation’s standard online catalogue: is potentially unreliable information better than no information at all? What would you consider an ‘acceptable’ quality and/or quantity level for catalogue metadata for public presentation? You might think this photograph should never have been uploaded to flickr in its current state – but even this meagre level of description has been sufficient to start an interesting – potentially useful? – discussion. Just as a relatively poor quality scan has been ‘good enough’ to enable public access outside of the repository, although it would certainly not suffice for print publication, for example.
Such ambivalence and uncertainty about accepting user contributions is one reason that The National Archives wiki Your Archives was initially designed “to be ‘complementary’ to the organisation’s existing material” rather than fully integrated into TNA’s website.
In our discussion on Friday, we identified four ways in which online archives might try to establish trust in user contributions:
- User Profiles: enabling users to provide background information on their expertise. The Polar Bear Expedition Archives at the University of Michigan have experimented with this approach for registered users of the site, with apparently ambiguous results. Similar features are available on the Your Archives wiki, although similarly, few users appear to use them, except for staff of TNA. Surfacing the organisational allegiance of staff is of course important, but would not inherently make their comments more trustworthy (as discussed above), unless more in-depth information about their qualifications and areas of expert knowledge is also provided. A related debate about whether or not to allow anonymous comments, and the reliability of online anonymous contributions, extends well beyond the archival domain.
- Shifting the burden of proof to the user: offering to make corrections to organisational finding aids upon receipt of appropriate documentation. This is another technique pioneered on the Polar Bear Expedition Archives site, but might become burdonsome given a particularly active user community.
- Providing user statistics and/or manipulating the presentation of user contributions on the basis of user statistics: i.e. giving more weight to contributions from users whose previous comments have proved to be reliable. Such techniques are used on Wikipedia (users can earn enhanced editing rights by gaining the trust of other editors), and user information is available from Your Archives, although somewhat cumbersome to extract – in its current form, I think it is unlikely anybody would use this information to form reliability judgements. This technique is sometimes also combined with…
- Rating systems: these can be either organisation-defined ratings (as, for instance, the Brooklyn Museum Collection Online – I do not know of an archives example) or user-defined (the familiar Amazon or e-Bay ranking system -but, again, I can’t think of an instance where such a system has been implemented in an archives context, although often talked about – can you?). Flickr implements a similar principle, whereby registered users can ‘favourite’ images.
A quick scan of Google Scholar reveals much research into establishing trust in the online marketplace, and of trust-building in the digital environment as a customer relationship management activity. But are these commercial models necessarily directly applicable to information exchange in the archives environment, where the issue at stake is not so much the customer’s trust in the organisation or project concerned (although this clearly has an impact on other forms of trust) so much as the veracity and reliability of the historical information presented?
Do you have any other suggestions for techniques which could be (or are) used to establish trust in online archives, or further good examples of the four techniques outlined in archival practice? It strikes me that all four options above rely heavily upon human interpretation and judgement calls, therefore scalability will become an issue with very large datasets (particularly those held outside of an organisational website) which the Archives may want to manipulate machine-to-machine (see this recent blog post and comments from the Brooklyn Museum).