Online disclosure neither now nor never: When would a “later” button work?

Authors


Abstract

The combination of personal demographic data, taste preferences, public disclosure of friend networks and now increasing usage of tools for instantly updating status (e.g., Twitter, Facebook) has fundamentally altered users' understanding of the temporality of information and its (semi-)permanence. A recent exploratory study found users were more willing to share items as time passed, and a major finding included a latent willingness (40% of sharing scenarios) to disclose information at a later date, neither “now” nor “never”. This study further explicates these findings by examining in greater detail the nature of the role of temporality in disclosure willingness.

Introduction

Research has repeatedly shown that computer-mediated communications (CMC) lead to higher levels of disclosure of personal information (Tidwell and Walther 2002). Recent studies have examined the role of increasingly common social media and social network services (SNS) on disclosure in a variety of contexts (Mazer et al. 2007; Tufekci 2008). The combination of personal demographic data, taste preferences, public disclosure of friend networks and now increasing usage of tools for instantly updating status (e.g., Twitter, Facebook) has fundamentally altered users' understanding of the temporality of information and its (semi-)permanence. A recent exploratory study investigated users' willingness to disclose information with respect to how long ago that information may have been created or captured. Users were more willing to share items as time passed, and a major finding included a latent willingness (40% of sharing scenarios) to disclose information at a later date, neither “now” nor “never” (Russell and Kramer-Duffield 2009). A recently released piece of software augmenting Twitter – HootSuite – advertises among its chief selling points the ability to “pre-schedule tweets” (Twitter messages).

Research Design

This study further explicates these findings by examining in greater detail the nature of the role of temporality in disclosure willingness. Initial investigations suggested that while there is a “now, later, never” tripartite division in terms of disclosure willingness, users make less fine-grained and more binary distinctions in terms of audience and intimacy level of information. This is perhaps due to the flattening effect of social platforms like Facebook and Twitter, which mostly allow you to say “yes” or “no” in determining whether someone is your “friend” or whether a piece of information should be shared at all.

Further research will therefore collapse intimacy and audience distinctions and further explore the nuance – or lack thereof – regarding latent delayed-disclosure desire. We will test a “specific later” versus “generic later” proposition along several sets of comparisons, investigating:

RQ1: What is the effect on willingness to disclose of specifying a future date (one day, one week, one month) at which users might (automatically) share a given piece of information, as compared with a generic phrasing of “later”?

RQ2: What is the effect on willingness to disclose of allowing users to themselves specify a future date at which users might (automatically) share a given piece of information, as compared with a generic phrasing of “later”?

If, as is suggested by the ongoing research, users do possess a latent or emergent desire to augment their disclosure behaviors by adding “later” to their temporal toolkit, it is important both for researchers and designers to investigate the more specific nature and contours of this desire. This research aims to quantify these desires as a step towards establishing a more complete picture of evolving user practice online.

Methods

This study as the initial investigation preceding it will utilize Amazon's Mechanical Turk system for the gathering of a large amount of data and ensuring a consistent set of results. As noted in Kittur et al. (2008), the Mechanical Turk is a quite useful methodology but that “special care is needed in formulating tasks in order to the capabilities of the approach.” Among their suggestions which we followed was use of a simple mathematical CAPTCHA, which helped to eliminate just 175 results from our original result set of 2205 (some were presumably bots attempting to game the Mechanical Turk system).

Bearing in mind that there is currently no mechanism to construct a traditional study design with mutually exclusive groups, the complete data available through the system makes it possible to check for consistency among data provided by discrete workers and to remove bots and respondents with perceived malign intent. These basic measures and quality checks on the data allowed for a very clean and extensive set of results, gathered quickly and at relatively little expense.

Discussion

In the past we have had a binary choice to make when sharing information: to share, or not to share. We reserved the right to wait until later and share with a new current audience, a story or information from the past (though it is and always has been difficult or impossible to move in the other direction – to “unshare”), but outside of very particular circumstances have not had a ‘later’ button. The few exceptions to this general rule are for the most part legal documents like wills, trusts, or legally-sealed government documents, which specify that certain materials, writings, etc. be revealed or opened access to “upon death”; upon a child achieving the age of majority; or after a given time period has past. In day-to-day life, the few examples involve customs like Christmas presents or time capsules, where admonitions of “don't open until Christmas” or “wait 30 years until opening” are generally obeyed. The difference between all of these actions and the preferences suggested by this research is that the delays mentioned above apply to a very particular action with a predetermined, limited and specified audience, while our research indicates that the desire for a ‘later’ button is prevalent regardless of audience (even while audience has some effect on the precise parameters of the professed sharing predilections).

In technology circles, or spaces of innovation more generally, it is received wisdom that legal frameworks lag behind changes in practice substantially and sometimes to the detriment of innovators. In this case, however, legal frameworks may both be ahead of the game and offer us a set of approaches that address these latent or emergent desires for a ‘later’ button. Apart from the limited practice of post-dating blog posts, and the recently-introduced capabilities of HootSuite mentioned above, there is thus far little technical innovation along these lines.

Conversely, in addition to the above-mentioned legal precedents of wills, trusts and government documents, there has been substantial thought, discussion and innovation in terms of legal frameworks around copyright issues and temporality. A proposal has also been made in the open source software community by Zooko Wilcox-O'Hearn, whose Transitive Grace Period Public Licence (http://zooko.com/tgppl.html) proposes an opportunity for authors of code to retain a limited license provided they distribute the full code “no later than 12 months” after the creation of the work. The latter proposal is the matter of some controversy and discussion (see http://tinyurl.com/zooko1 and http://tinyurl.com/zooko2) in the Open Source community, with some fearing a proliferation of open licensing protocols may lead to confusion or un-enforceabillity. Other Open Source advocates have proposed the formation of an encrypted “escrow service” for open source code whereby access to the code would be restricted until a specified “later” date, after which time it would be licensed under a fully open and unrestricted protocol.

In at least one case, these proposals are being put into action already. Creative Commons has extensive experience with these issues as pertains to its licenses, and its Founders' Copyright allows users to register works in the manner not of current copyright but rather as it was initially set forth in US copyright law. Under the Founders' Copyright, Creative Commons buys a creator's copyright for $1 and licenses it exclusively to the creator fora 14-year term with an optional 14-year renewal, after which all works become part of the public domain. These approaches all offer potential frameworks by which users can think differently about issues of disclosure and temporality.

Conclusion

It seems likely that as users spend more and more of their time online, in situations where they disclose information about themselves in various contexts, they will become increasingly aware of the aspects of temporality regarding information they share, do not share, or could share. Our initial investigations that users may already be more aware of these issues, and desirous of additional temporal affordances in addressing them, than is currently assumed. As users, designers and researchers deal with the ongoing disclosure and maintenance of personal information online, it is imperative that further developments be carried out with substantial forethought to the range of issues explicit and implicit therein. This requires, in our view, both a thorough and ongoing examination of users' views and desires, and also a thoughtful examination of relevant precedents and current models suggestive of possible approaches. This research is another step in piecing together the ever-evolving picture of individual disclosure behaviors and perceptions online.

Ancillary