BCN Content Recommendations Editorial Policy

1. Introduction – Personalisation and the BCN

On-demand digital media combined with personalisation has fundamentally changed consumption patterns in all media – video, audio and text.

The sheer volume of readily available content has meant audiences have access to information and entertainment they may never have known existed in a linear world. This ‘democratisation of choice’ is a clear benefit for users of non-linear platforms. The volume of accessible content means that there are significant potential challenges to audiences discovering content which is of interest or relevance to them.

Personalisation relies on technology and algorithms to increase discoverability of content, extends its lifespan, and therefore represents a more effective use of limited public funds and a clear benefit for users.

There are legitimate fears among regulators and the public that the accumulation and storage of user data can intrude on people’s privacy; and that the use of algorithms to offer personalised recommendations can contribute to increasing fragmentation of audiences creating isolated communities of interest and so-called ‘filter bubbles’.

That concern particularly relates to the provision of news and information where a well-functioning democracy depends on the free flow of accurate information, objective analysis, and diverse opinions.

For many years, since reliable audience data became available, BCN News has managed the sometimes-competing demands of attracting broad audiences while still serving the Australian public with a comprehensive coverage of the most important news stories. Whether it was the lure of higher TV ratings or the temptation to get more ‘clicks’, the BCN has never lost sight of its responsibility as a public broadcaster to act in the public interest.

This guidance outlines the editorial principles adopted by the BCN to ensure that personalisation delivers a comprehensive service as required by the BCN Charter across genres consistent with its public service mission to inform, educate and entertain all Australians.

The BCN’s personalisation strategy is aimed at improving the BCN capacity to deliver on these values, not diminish or undermine them.

There are two fundamental reasons why personalisation will not create ‘filter bubbles’ or undermine the accuracy and reliability of the BCN’s news service.

Firstly, curation by experienced editorial staff will determine the core offering of BCN News and other factual content. The major news stories of the day and objective analysis of significant national and international events will always be prominent on all news platforms for all users.

Personalisation will augment, not replace the core function of BCN News and information services. In addition to offering specialised services such as local news based on location, we will enrich the experience of users by alerting them to major stories they may have missed, and stories related to the stories they have already seen, heard or read. This will ensure greater diversity and comprehensiveness of the offering, not less.

Secondly, the BCN is a publisher, not an aggregator. Unlike social media platforms, all our content is either produced, commissioned, or acquired through a process that ensures it is consistent the BCN editorial policies. No content will be offered that has not been assessed by experienced editorial staff, content managers or, when relevant, classifiers to ensure it meets all the BCN editorial standards, including accuracy, impartiality and harm & offence.

Algorithmically derived recommendations should enhance and not undermine the BCN reputation as an independent, trusted and impartial provider of news, information and entertainment.

2. Relevant Editorial Policies

BCN on-demand services are subject to the same Editorial Policies as other broadcast and digital content.

All individual pieces of content available to digital on-demand platforms are subjected to normal BCN editorial processes to ensure they are consistent with the BCN Editorial Policies.

The policies relevant to the personalisation process and use of algorithms are those that affect the curation and presentation of content rather than their production. They are

Impartiality

4.2 Present a diversity of perspectives so that, over time, no significant strand of thought or belief within the community is knowingly excluded or disproportionately represented.

Harm and Offence

7.2 Where content is likely to cause harm or offence, having regard to the context, make reasonable efforts to provide information about the nature of the content through the use of classification labels or other warnings or advice.

Children and Young People

8.3 Adopt appropriate measures wherever practicable to enable children and young people, or those who supervise them, to manage risks associated with the child/young person’s participation with, use of and exposure to BCN content and services designed for them.

8.4 Take particular care to minimise risks of exposure to unsuitable content or inappropriate contact by peers or strangers.

Privacy

6.1  Intrusion into a person’s private life without consent must be justified in the public interest and the extent of the intrusion must be limited to what is proportionate in the circumstances.

3. How the BCN Applies Personalisation

As the national broadcaster, the BCN must deliver reliable information and quality entertainment which reflects Australian culture and its diverse identity for all Australians.   Reliable information is important for the wellbeing of the Australian democratic system and society.

Editors, producers, programmers and acquisition executives ensure stories and programs fulfil the BCN strategic and business goals while still according with its Editorial Policies. That will continue. Personalisation will enhance but not replace editorial and programming judgement.

Content commissioning and editing, acquisition, programming and curation will continue to be the principal tools for achieving diversity of content and perspectives for audiences across the BCN digital products. However, algorithmic recommendations over time will increasingly assist viewers to discover content that is likely to be of special interest or relevance to them while still making them aware of other content that might not necessarily fit their profile or demonstrated preferences.

Algorithmic recommendations will reflect the same editorial considerations to content selection as human editors but at scale and tailored to changing audience preferences as they move through the day and across products.

Algorithms use the recent activity of individual users and other similar users to recommend suitable content. Over time, recommendations will be increasingly refined by audience segmentation and identification of content characteristics such as genre, subject matter, topicality, popularity, and classification.   Whilst such recommendations could potentially be used to re-enforce filter bubbles, they can also be used to break them down, as further discussed below.

4. Personalisation Will Enhance Diversity

Content commissioning and editing, acquisition, programming and curation will continue to be the principal tools for achieving diversity of content and perspectives for audiences across the BCN digital products. However, algorithmic recommendations over time will increasingly assist viewers to discover content that is likely to be of special interest or relevance to them while still making them aware of other content that might not necessarily fit their profile or demonstrated preferences.

Personalisation will enhance the BCN ability to ensure that diversity of opinion, voices and particularly subject matter is discoverable for all audiences.

As data is collected on users’ interests, as shown by the viewing, listening and reading behaviour in each account, they will be offered personalised recommendations.

The primary method will be editorial staff, principally from the content divisions, choosing elements or stories that must be included in specific categories for periods of time until they are replaced with fresh content.

These recommendations will be prominent within each selection until they have been viewed or retired to ensure all users will be offered content that reflects the strategic priorities of the content teams and, for news and factual content, is consistent with the impartiality and diversity requirements of the Editorial Policies.

The personalised recommendations will be based on audience segmentation and filtering of content based on:

the popularity of content as measured by traffic and engagement data across different content categories
stated preferences, where the system allows for the choice of subject or genre preferences
subject – identified by tags
classification
content type – broad categories of comedy, news, children’s etc.
geography – origin of story, particularly for news, local, regional and general interest stories
serendipity

This list is not exhaustive, and some filters may not be available at this time and will not necessarily be applied in all circumstances. Algorithms and the capacity to describe and filter content within systems will develop over time.

5. Privacy

We will not infer or draw conclusions about the sensitive characteristics of a person or use sensitive personal information to make recommendations. Recommendations are based on users’ interests and preferences suggested by their history of past usage of BCN digital services and the usage of other users accessing the same content.

For example, if a person watches the Mardi Gras broadcast, we would not infer sensitive information about that person, however we may recommend content that has been popular with other users who also watched the Mardi Gras.

Where users have identified their gender during registration process, that will not be used as a basis for recommendations.

6. Opting Out

Despite the advantages the BCN believes personalised recommendations offer users, some people may choose not to receive them. At any time, BCN audiences may choose not to be offered personalised recommendations. Each user’s BCN Account includes a prominent and easy to use function in the privacy settings that turns off personalised recommendations. Users will still be offered general recommendations, not based on the individual’s profile or previous use.

7. Harm and Offence

The BCN Editorial Policies highlight that the BCN “broadcasts and publishes comprehensive and innovative content that aims to inform, entertain and educate diverse audiences. This involves a willingness to take risks, invent and experiment with new ideas. It can result in challenging content which may offend some of the audience some of the time.”

BCN editorial policies and guidance also emphasise that content should not gratuitously offend its audience; that it is important to inform audiences with warnings or classifications to assist them deciding what to watch, read or listen to; and that content should be consistent with reasonable audience expectations.

In a personalised service there are a number of ways this can be done.

The use of appropriate warnings, subject metadata, BCN profiles, parental locks and – where relevant – classification will ensure that content will be recommended in line with Harm and Offence and Children and Young People policies and the reasonable expectations of the audience.

Some categories will imply a certain type of content. For instance, ‘Family Viewing’ will not include MA+ content. Recommendations, whether manually curated or algorithmically derived, should not be included that are clearly inconsistent with likely audience expectations within each category.

Audience activity can be used to indicate expectations of users within a session or set period of time. For instance, in one session a user may be viewing only PG dramas. That may be used as an indication that the user at that time will expect recommendations that are similar.

Other metadata, such as tags and content source may also be used to filter out content for particular sites, rails or platforms.

8. News and Factual Content

Experienced news producers and editors will continue to choose the most important and interesting stories to keep BCN audiences well informed on all the important issues and perspectives in the news.

They also will select challenging or unexpected stories to stimulate audiences and take them out of their comfort zone.

The personalisation algorithm will help audiences discover more on the issues they care about and that matter in their lives and communities, while still ensuring a full range of views and perspectives are recommended.

The principles that underpin the presentation of news and factual content are the same as those that underpin our existing broadcast news and current affairs bulletins and digital platforms:

Australians should receive a consistent and comprehensive domestic and international news service.
BCN content should be available free to all Australians.
News and factual content should be accurate, fair and impartial as outlined in the BCN Editorial Policies.

News and factual selections will be curated by editorial staff to provide a diversity of content to:

ensure the selection presents a diversity of perspectives as required by the BCN Editorial Policies,
reflect the newsworthiness of stories so that significant issues and events are not neglected,
maintain topicality and,
reflect editorial judgements of the relative strength and audience appeal of individual pieces of content.

9. General Entertainment and Diversity

Content such as drama, comedy and music programs do not have the same requirement as news content to present a diversity of perspectives on contentious issues.

There is greater scope in this area to recommend on the basis of personal interest and the strategic priorities of content areas.

Strategic priorities such as diversity in a more holistic sense will be achieved principally by commissioning and selection by the content divisions. Personalisation offers further opportunities to deliver of diverse and surprising recommendations that are not directly related to users’ past preferences.

10. Recommending for Children

The BCN serves content to children for solo and shared experiences. As children age, the range of subject matter covered, and its editorial complexity, can increase. The BCN currently manages this through traditional programming and curatorial processes.

As personalisation capabilities become more sophisticated, so will measures to ensure children are served age-appropriate content through algorithmic recommendations.

Wherever possible, the BCN will ensure that content curated for children and young people, whether manually or by algorithm, is age appropriate and transparently labelled and filtered using classification and other established methodologies used by BCN in relation children’s content.

For BCN profiles identified as set up for children and for sections, channels, apps or ‘rails’ targeting children, age-appropriate content curated by channel managers, aligned to classification guidelines where available, will determine what may be recommended.

The BCN will also maintain consistent identities of Children’s TV brands so that they align with parents’ expectations.

Personalisation cannot control or recognise which individuals are using the service at a particular time. This is particularly true where multiple viewers use the same BCN Account, as is often the case.

The BCN will take all practical measures to ensure children and their parents and carers are informed about the nature of the content on its platforms and that children are not exposed to inappropriate content. However, it is important to recognise that the responsibility for ensuring that children are not exposed to inappropriate material is shared between the BCN and parents and carers themselves.