Building Trust Through the Use of Standards
Letter from the Executive Director, March 2024
A core element of trusting something is knowing its source. You might recognize someone as being unreliable or having provided inaccurate information in the past. You might question their motivations and therefore may have reason to doubt that they are providing the full scope of the story, or wonder if they have cherry-picked facts that support their perspective. This is increasingly the problem with artificial intelligence (AI) systems that are generating content. It might not even be possible to trust an audio clip you’ve heard or an image you’ve seen.
Trust has become a growing concern in our community for several different reasons. While the history of political media is nothing new, there has been a marked increase in partisan political media, and many things, even if nonpolitical in nature, are now perceived through a partisan lens. This has led to an associated decrease in trust in media. Social media algorithms have been shown to increase this polarization within media consumption. The rise of machine-generated text, image manipulation tools, and deep fakes of audio and video has exacerbated these challenges to trusted communications. These broader political concerns weren’t generally a problem in scholarly communications.
However, in the past couple of years, the issue of research integrity has grown in prominence, and many of these broader issues are finding their way into scholarly research as well. There has been an increase in the use of (or perhaps just monitoring of) image manipulation tools. A number of likely predatory publications have assumed intentionally misleading brands to attempt to mislead authors. There have been issues regarding special issues and questionable editorial practice.
Ensuring research integrity is a key problem for our community. Standards support the networks of trust. While NISO isn’t involved in the discussions around editorial practice per se, such as the leadership provided here by COPE, we are engaged in other elements of the publication process related to integrity and its signaling in the community. This is the focus of the Communications of Expressions of Concern (CREC) project. This project, funded in part by a grant from the Alfred P. Sloan Foundation, released a draft for public comment last fall. The committee has been reviewing the comments and is expected to submit its recommendations for approval by the NISO Topic Committee next month, with publication following thereafter, if approved.
Recognition of authorship or source is also important in reputation building. Citations are one way in which we build knowledge and track it through history. References are the practical implementation of the notion that “we stand on the shoulders of giants.” Understanding the context of a new idea and placing it in the thread of knowledge helps to grasp the meaning and value of a new discovery. All of this is well established in scholarly communications. Provenance is another important issue today, from several different perspectives. Persistent identification can help alleviate concerns about provenance. For example, the use of the ORCID system to track authorship has become a useful method of tracking people’s contributions. Other issues remain.
Last month, I participated in the launch meeting of an ISO International Workshop Agreement (IWA) project to launch a new Unique Media Identifier (UMiD) organized by DIN and the Global Media Registry. The purpose of this new identifier is to unambiguously identify media distribution channels and brands for content. It aims to enhance integrity in digital search, indexing, streaming, and social media platforms by uniquely identifying sources of content in the distribution chain. Rather than identifying the individual content objects, the UMiD would identify the means of distribution so that content can be connected to its source. Confusion about content’s source can result in misattribution, incorrect indexing, mismatching, and promotion/demotion of content in algorithmically managed systems. As proposed, the identifier is meant to be a transparent means of tracking back to the channel owners, a means of promoting trust. While not serving to whitelist or blacklist content, the identifier will provide a unique and cross-industry method for identifying these sources. For the scholarly community, the UMiD would address the problem of distinguishing between a long-established title and a similarly named one that just started and could be predatory. An ISO International Workshop Agreement is the lowest-level consensus output at ISO and is not a standard, although it could develop into one later if it gains traction. While it will not solve many of the harder integrity problems, it will add to the network of trust that exists in the online environment. Work on the IWA is expected to progress quickly, with a goal of producing a public consultation draft in August and a final output by February 2025.
There are other areas where integrity and trust building could involve the development of standards. Trust of AI systems is critical as they become more widely used and adopted. This was a theme of some of the sessions at the NISO Plus Conference in Baltimore last month. For example, tracking provenance of AI outputs to their sources for verifiability and recognition was an idea that might be explored for standardization. The NISO staff is preparing a report of the outcomes of NISO Plus, which we hope to be sharing with the community in the coming month.
Of course, there are many other initiatives related to research integrity. I’ll be moderating a session at the upcoming Society for Scholarly Publishing conference in June entitled, “Only you can prevent research integrity fires!” It should be a wonderfully fun session where these topics will be discussed. I look forward to seeing you there!
Sincerely,
Todd Carpenter
Executive Director, NISO
