Every airport has it: a three-letter code making sure that a bag labelled ‘COR’ is travelling with its owner to Córdoba in Argentina (and not to Cork in Ireland, which is ‘ORK’). Other examples are phone or bank account numbers, licence plates for cars, ISBNs for books and tax identification numbers. Consumers and professional communities are surrounded by all sorts of unique identifiers that provide clarity and distinction in cases where ambiguity is the enemy.
In the media sector, many brands and channel names sound similar, sometimes by coincidence but often on purpose when malicious actors try to create confusion on which they can capitalise. This can mislead audiences and advertisers as well as the algorithmic-driven recommender systems of search or social media platforms that direct our attention and advertisers’ dollars.
Clear definitions and terminology are a condition for the functioning of any complex system. Our shared information space is clearly – and increasingly – one of these systems where standards are developed in parallel to allow for more interoperability and safety. Unfortunately, the opposite seems to be the result so far. At the personal level of consumer safety and in terms of national security, the internet is increasingly perceived as a threat against anything ranging from the wellbeing of an individual to the rule-of-law based world order.
The dilemma of content moderation
In an age of disinformation and elevated geopolitical rivalry, reliable information is needed more than ever. It deserves prominence and visibility, while harmful and illegal content must be demoted or deleted. Both categories are, however, hard to define and identify. This is already a challenge in the case of illegal activities online and almost impossible in the ‘lawful but awful’ category. There are convincing reasons to not even try – who has the legitimacy to judge and on what grounds? Censorship and ‘content moderation’ are essentially the same thing, though with different purposes and alleged intentions.
On the other hand, freedom of speech does not mean that anything goes in the physical world or online. Balancing civil liberties and personal safety was never easy and more recently has become an increasingly polarising and toxic matter of ideology and ‘culture wars’. But politics aside, the sorting, filtering and indexing of content is also a necessity of usability. The whole knowledge of humankind may now be available literally at our fingertips, but it is worthless without the structure and tools to access it. This basic requirement sparked the success story of one search engine a generation ago and it is driving the breakthrough of generative AI today.
Like it or not, the functioning of our shared information space depends on digital infrastructure, platforms and technology. Underlying protocols and algorithmic-driven recommender systems determine how we see the world online. In order to work properly, they require real-time data to determine which piece of content is promoted, demoted or deleted based upon inclusion lists of trustworthy sources of content and exclusion lists of bad actors. Currently, these datasets are provided by an increasing number of external actors, such as commercial ad-tech providers and NGOs, but are also compiled internally by platform owners.
Even if a web domain or social media account is always distinct, it might not be immediately clear to which media outlet or company it belongs.
A problem occurs when these different lists result in ambiguities or mismatches. This might be the case with brands of the same name (there are several dozen media outlets called ‘Phoenix’), or with affiliates, syndicated channels or sister brands of the same origin, but with different editorial lines (e.g. Al Jazeera or Fox). Even if a web domain or social media account is always distinct, it might not be immediately clear to which media outlet or company it belongs. This could lead to conflicting or wrong signals, misleading algorithmic indexation and negative impacts on site integrity and user experience. Bad actors may even exploit this deficiency and try to game the recommender systems with similar sounding names of channels, accounts or brands.
Introducing a unique identifier
This is the reason why a unique identifier could be helpful in mitigating ambiguities along the distribution chain of online content. Of course, all actors involved are using their proprietary sets of identifiers already today. They are a necessity for the management of large datasets, be it as a tech platform, regulator or advertiser. However, harmonising them is the prerequisite for interoperability and as such, the added value of a unique identifier. It would not replace any existing model of structuring media-related data but provide an additional, unified code for a more seamless exchange between them and thus help to safeguard the integrity of the information ecosystem overall. Driven by political priorities, regulatory logic and economic interests, the demand for the convention of a unified taxonomy has become even more visible over the past months. Whether it is the enforcement of legislation in the field of digital policy, consumer protection or competition, the attempt to protect brand safety, subsidising public interest journalism or fighting foreign influence operations online, the unambiguous identification of the source of a piece of content is always key.
Accuracy in attribution begins with clarity in terminology. Different data models with diverging categories derive mainly from the range of distinct purposes behind them. For example, a national regulator uses a different logic and vocabulary to a global social media provider or marketer. Accordingly, the meaning of terms like ‘channel’ or ‘brand’ or ‘asset’ or ‘outlet’ can contain and conceal important nuances – let alone in different languages – that matter when datasets are matched and merged.
The Global Media Registry is a German non-profit organisation that operates the global Media Ownership Monitor initiative.1 It has experienced these challenges first-hand when building and maintaining its database. A conversation with different stakeholders produced more questions and, eventually, the plan to initiate the development of a unique media identifier (UMId) for channels and brands. Partners such as the European Platform of Regulatory Authorities and its British member Ofcom supported the idea from the start. As an authoritative and fully self regulatory instrument, industry standard setting through the International Organization for Standardization (ISO) was selected and pursued. This includes a thorough application process, whereby the proposal is vetted by experts throughout the global ISO community of national standardisation bodies, mainly to avoid duplication or overlaps with existing norms. This exercise produced a number of useful references. For example the international standard name identifier was not considered to be redundant and could feed into UMId development. In addition, the German national standardisation body, DIN, volunteered to facilitate the initiative on behalf of the ISO.
The ISO method
An ISO standard is built through collaboration and transparency. The demand needs to be clearly articulated at the start, but the exact outcome is not determined. In this case it is up to the experts participating in the process to agree by consensus on the final shape and syntax of the UMId. This can range from a simple number or abbreviation (such as those used for airport codes) to a string of numbers and letters that include additional information (for example, the two-letter country prefix in the IBAN system) to a multi-tiered system. The operating model will also be developed and decided during the standard setting workshop. A variety of different examples to consider already exist, ranging from a fully automated distribution of the identifier to institutionalised models of curation and governance (ISBN and ICANN, for example).
During the preparatory phase the proposal could also be refined and checked for the potential of misuse or resistance. With a risk based approach, the main question would be to what extent the implementation of a UMId standard might result in unintended consequences or harms, such as being used as a means of censorship and thus encroach on freedom of speech. In this case, the answer is clearly that it couldn’t and doesn’t. First of all, identifiers that help to direct the distribution of content online exist already in large numbers and at several levels. The purpose of the UMId is to harmonise them, based on a mutually agreed taxonomy, not to invent something entirely new. Secondly, it is limited to the source level of channels or outlets. It would not be used for individual items of content. Thirdly, and perhaps most importantly, it is non-judgemental. It makes no statement about the quality of a source any more than the airport tag ‘COR’ has a role beyond ensuring that a bag follows a traveller to Argentina and does not end up in Ireland. It does not include any information about the length of the runway, for example, or the quality of coffee sold in the terminal building.
Safety by design
Many industries already rely on the concept of safety by design, where risk is minimised through a pro-active approach of embedding protocols and universal standards into processes rather than retrofitting safeguards after an issue has occurred. In the fight against disinformation however, the latter strategy still prevails, for example by trying to build resilience among audiences through so-called media literacy, or by means of fact checking and debunking false, potentially harmful content ex post, after publication. In a complementary way, the UMId’s purpose of enhancing the coherence and integrity of indexation and recommendation ex ante follows a source rather than content-driven logic.
No single measure will eradicate the curse of disinformation nor will it solve the broader problems of the internet. Many building blocks are required and they need to fit together in order to create a larger whole. Real-time interoperability along the algorithmically determined distribution chain of online content can be considered one of those – this is where unique media identifiers fit in. Establishing them through the transparent, consensual and self-regulatory process of ISO standard setting should be in the best interests of many stakeholders, both in the private and public sector. All of them are invited to join and contribute to an endeavour that begins in early 2024.
The kick-off meeting for the international workshop agreement on unique media identifiers for distribution channels and brands will take place on Wednesday 21 February 2024 at DIN, Burggrafenstr. 6, 10787 Berlin, Germany. The workshop will be open to registered participants only. An option to join virtually will be provided.
For more information or to register, contact: [email protected]