During the second half of 2021, the Journalism and Media Ethics program at the Markkula Center initiated a set of virtual roundtable meetings around the term "News Distribution Ethics". We invited product, policy, and program leaders from the news platforms (social media and search companies) as well as from AI Ethics and journalism organizations. This summer, we drafted our initial set of recommendations for ethical news distribution after defining the problem, and outlining our key terminology and an applied ethics approach. We are recommending that news platforms and aggregators make the workings of their distribution mechanics more transparent along a set of guidelines drawn from three related and distinctive paradigms: rights, harms, and discourse.
Update – February, 2023: Media Policy Scholars Provide Input to News Distribution Ethics Recommendations
View or download "News Distribution Ethics Roundtable Key Principles and Recommendations" in PDF format.
Executive Summary
This document is the first set of recommendations from a cross-functional convening of leaders from technology, journalism, and academia whose common interest is that both news and its distribution must better serve democracy. It is intended to be a living document and suitable for shepherding through a living process of public input run by a cross-functional multi-stakeholder group.
News is a public good with both utility and significance for citizens and communities worldwide. Since social media, search and news aggregator products distribute (push and pull) the news, sometimes personalized or influenced by the user’s network or location, news distribution both carries and shapes public discourse and hence impacts democracy. It is through distribution that we can (sometimes frictionlessly) discover both known and new news publishers (or sources), and access and read stories and perspectives without visiting myriad sites or channels or streams separately during every news-consuming episode of our day. Distribution can impact outcomes for public policy and democratic life downstream of it.
Technology companies are interested in actionable ethics guidelines for news distribution that focuses on publisher-related policies, human and machine curation, and recommendation systems. Having an ethics baseline or starting point for news distribution, particularly for democracies, is as critical as the evolution of content moderation policies for User Generated Content (UGC).
Key takeaways: This document offers a number of key takeaways for the target audience in both technology companies (product managers, curators, policy and standards leaders) and media analysis initiatives in civil society.
- The case that the ethics of news distribution is a worthy focus area, both related to and separate from the ethics that guide journalism and news production itself.
- Three paradigms to motivate and clarify the recommendations for news distribution ethics: rights, harms, and obligations regarding democratic discourse.
- The case for transparency as the first and common operating principle for news distribution entities. Transparency will help assess in measurable ways how news distribution works today across various types of tech platforms.
- The key takeaways: a scaffolding of six detailed transparency recommendations drawn in alignment with rights, harms, and obligations to discourse.
- Finally, a non-exhaustive set of examples of qualitative disclosures for two out of six recommendations, drawn from various web pages and communications of Google News, Twitter, and The Factual. This shows that a starting point for news distribution ethics already exists, and it needs to be expanded with structure and a common vocabulary.
- We are releasing this document for public input in October 2022.
Audience for this document
Organizations
- News aggregators/distributors (platforms, startups, etc.)
- Public oversight entities
- Researchers who want to understand and monitor news distribution
People at Platforms/Aggregators:
- Product managers and their teams (engineers working on recommenders included)
- News curators (often journalists) working within tech firms
- News curation standards leaders
- Policy staff handling news publishers
- Managers handling relationships with fact-checking entities, internal validation/assessments of news veracity, whether in policy or program or product teams.
News is a public good with both utility and significance for citizens and communities worldwide. In democracies especially, news organizations see their role as not merely releasing information, but also in agenda-setting, such as the legitimization of public issues and actors, and in defining the polity, by either including or excluding groups into or from the mainstream of discourse and narratives. Discourse refers to processes of argumentation and dialogue in which claims can be tested for their correctness or authenticity.1
Since social media, search and news aggregator products distribute (push and pull) the news, sometimes personalized or influenced by the user’s network or location, news distribution2 both carries and shapes public discourse and hence impacts democracy. Online news distribution is uniquely different from news. It is through distribution that we can (sometimes frictionlessly) discover both known and new news publishers (or sources), and access and read stories and perspectives without visiting myriad sites or channels or streams separately during every news-consuming episode of our day. But access is not all. Distribution technology is participative. Both the level of exposure and consumption of newsy content are at least partly determined with input from user signals and virality.
Discourse is a critical part of participatory and deliberative models of democracy3, where citizens take an interest to inform and express themselves and assert their agency on ongoing public matters and systemic questions. Both news and news distribution have significant impacts on discourse. While news itself significantly impacts and drives discourse by informing, engaging, catalyzing, legitimizing, polarizing, invisibilizing, radicalizing, and disinforming, news distribution mediates between us and the news. It operates through product “surfaces” we access news on. Those surfaces themselves are various combinations of push or pull with personalized, unpersonalized, or stateful designs. See the primer section for more. Distribution can impact outcomes for public policy and democratic life downstream of it.
In a democratic context, what policies and ethics should guide news distribution design and decision-making? One complication is the boundaryless nature of journalism and news, which has ensured long-enduring questions such as who is a “news publisher” that do not lend themselves easily to technical guidelines or a binary categorization framework. Another complication is that digital platforms, by facilitating information creation and distribution/publishing directly have created a second type of boundarylessness. Examples abound of acts of journalism that people (who are not traditional or professional journalists) release directly on YouTube, Twitter, Tiktok, etc. The platforms have facilitated such content creators to exist and find a means of income. When these videos (often explainers) gain circulation during a time of public attention on a pressing issue or controversy, they add to discourse by expanding the viewing’s understanding. The platforms’ increasing role as a source of information to society has blurred the lines with legacy journalistic outlets, which had been the source for "news".
Hence having an ethics baseline or starting point for news distribution is as critical as the evolution of content moderation policies for UGC. For their part, tech companies are interested in actionable ethics guidelines for news distribution that focuses on publisher-related policies, human and machine curation, and recommendation systems.
News Distribution is an umbrella term we use in this document to describe a variety of engineered digital “surfaces” on search, social media, and news aggregator applications that enable people to access news online. These surfaces might be available in news apps, news websites, or on search results with news. A number of search, social media, and news aggregator companies build these surfaces. An understanding of news distribution requires describing how surfaces are put together and what they assume about their human users.
Surfaces types: push or pull
- Push: The human user is sent something. For example, recommendations based on what they just engaged with.
- Pull: The viewer requests something, for example in search via a query, and the results are displayed.
Surfaces may be either unpersonalized, stateful, or personalized.
- Unpersonalized: The product knows nothing about you and you'd have the same experience as anyone else in incognito mode. However, the product may include agnostic information about the user such as location and language, since a query from the user itself will come in a particular language.
- Stateful: The product is showing you something based on where you left off with a story, or what you last searched, or where you currently are. There is no inference in your implicit or explicit interests, just recalling from a log what you've done and reflecting it back to you.
- Personalized: There is some kind of inference in your interests being made and stored in a database somewhere. For example, “User X has an affinity towards {topic Y}. Personalization can be inferred in different ways:
- Activity: Implicit or explicit understanding of interest in topics, people, etc.
- Monodirectional follows: For example, inferred from who you follow (person, group, community) and the content the followee posts/interacts with.
- Bidirectional relationships (for example, social network friends)
- Intersections of the above: Sometimes interests are inferred based on the content you interact with as a result of the friends you have and the content they post.
- Inferences that inform recommendations have varying degrees of impact on people’s usage. Every person’s graph of friends or followers is different from another’s.
Surfaces are often combinations:
- Within a single distribution service, there can be many combinations of the above matrix (push-pull; unpersonalized, stateful, personalized).
See Appendix A.2 for rough examples.
We rely on three paradigms to motivate and clarify our recommendations: rights, harms, and obligations regarding democratic discourse. Human rights are often used to motivate ethical approaches around original principles that have broad global consensus. Harm-based standards, legal or regulatory requirements, policies, and design are widely adopted, for instance, in online content moderation. As noted earlier, the content, quality and tone of discourse is greatly determined by the news (and its ethics) that publishers produce and its distribution by aggregators and platforms; what ethical obligations exist around discourse? These three paradigms are interconnected and can be in tension with one another, for example, defining the line between someone’s right to expression and harm to another person.
Rights
What rights do I/we have in public conversation? Article 1 of both the International Covenant for Civil and Political Rights and the International Covenant for Economic, Social, and Cultural Rights says everyone has the right to self-determination, to “freely determine their political status and freely pursue their economic, social and cultural development.” International covenants further enshrine the right to freedom of opinion and expression, freedom of assembly, and access to information. A simpler frame to encapsulate these rights is the right to autonomy which is fundamental in classic liberalism. In this context, autonomy implies that citizens can actively participate in discourse (participatory and deliberative models of democracy) based on access to reliable information, and without coercive influence. The question for news distribution is whether or not a given platform’s distribution algorithms and curation support or hinder this fundamental right.
Harms
When does the distribution of news harm communities themselves or public discourse? We recognize here again that upstream of distribution, unethical journalism has a long history of inflicting harm on marginalized communities through elitism, exclusion, erasure, and disinformation. News distribution can perpetuate and exacerbate those harms, which are sometimes direct and sometimes through their effect upon larger discourse. First, by offering rapid reach to disinformation and misinformation, news distribution allows powerful anti-democratic actors to sow confusion. Disinformation, lies, and propaganda can deprive individuals of the ability to judge4 matters regarding their health and well-being, or harm them economically, reputationally, and socially; at scale, this deprivation can create harms at the societal level. Second, by offering reach to illegitimate controversies, news distribution perpetuates and amplifies the power of controversy-stoking actors that affect our collective decision making around issues with wide-ranging effects. While journalistic practice lacks clearly defined ethics around newsworthiness, the term illegitimate controversy encapsulates the tension around promoting content that makes for interesting reading and but gives equal weight to views, disputes, allegations that are not based on facts or grounded knowledge.5 When news distribution exacerbates illegitimate controversies around climate science, public health, or legitimacy of the democratic process, it is harmful to discourse.
Obligations
To foster democratic discourse, what obligations exist for news distributors? For democracies to work, discourse cannot merely be an episodic exchange of views between publics and leaders through different forms of media. Discourse should center on systemic issues in ways that are non-elitist, inclusive, truth-seeking and acknowledging, accommodative of expertise, foster understanding, and rooted in dignity and democratic agency. Discourse allows disagreement while still committing to our common humanity. It is worth noting that the mechanisms of democracy are intended to allow for peaceful, productive conflict.6 An ethical role for news distribution in this context can include fostering public attention to stories that:
- raise systemic issues (e.g. housing, health, poverty, opioid crisis, working-class jobs)7,
- bring visibility to historically excluded, socially and economically marginalized voices and their solutions to systemic issues8,
- move beyond individualized “episodic framing,” 9
- elevate acts of journalism directly from people with professional or experiential knowledge.10
Consideration
We note here that it is not possible to make recommendations on ethical news distribution without a measurable way to assess how news distribution works today across various types of tech platforms. Second, any effort to recommend must also have the assurance that ongoing measures will help test the efficacy of said recommendations. With this context, we are proposing the following principles and recommendations for news distribution ethics.
Principle 1: Platform Transparency for Accountability in Shaping Discourse
Why: Transparency in news distribution will help researchers and ethicists understand the role news distribution systems are playing in online discourse separate from the role played by news publishers and individual news actors. For example, data may help researchers and media analysts understand which narratives (pathways along which stories are told)11 are leading or prominent in the discourse on a politically-charged topic. The narratives may be grounded in reality or misleading or disinforming. This in turn may allow research on the connections between discourse on the platforms, widely-held beliefs or misbeliefs,12 and citizens being able to make informed decisions about themselves and their polities. Examples apart, simply stated, transparency will provide the data with which to ask better questions. The data will help empirically observe how coverage of systemic issues fares in prominence in distribution with hot-button political stories news publishers are releasing at particular epochs. With a deeper understanding of the news distribution system and its role, deliberative oversight bodies may be able to hold platforms accountable around specific harms and rights.
Reasons: This disclosure will help a) identify potential questions of exclusion or inclusion for specific types of traditional or non-traditional news sources in different news markets, b) assess whether some definitions in themselves bear upon discourse, harm, or the right of people to receive information from specific types of sources and c) assess the similarities and dissimilarities in definitions across different platforms.
Disclosure recommendations:
- How does your platform define “news” for both curation and algorithmic classification purposes?
- How does your platform currently accord “News” site or “News publisher” status to any entity?
- How does that designation affect access for (or promotion of) news publishers
- Access: Access to platform policy, exemptions, resources, business partnerships, etc. For e.g. some platforms are offering news publishers exemptions of particular kinds for boosting/advertising of content (ad policy) otherwise deemed as "political". Another access example is just registering formally as a publisher to be included in listings like News Tab or Showcase. Some news publishers get beta access to features/other special access to distribution measures.
- Does your platform define “who is a journalist” for curation and algorithms and if so, what is the definition or standard
- Definitions for the following as used by the platform. “Curation”, “Location”, “Topics”, “Trends/Trending” , and “News categories” used on feeds.
- How does your platform define and/or detect “Opinion” from news publishers?
Reasons: These disclosures help clarify the role and scope of discourse-directing leverage that curation has at the platforms, including whether curation is being done both at the piece of content level (during breaking news cycles or trending conversations for e.g.) or also at the news source level. This in turn will help external oversight bodies determine the scope of accountability for curation in mitigating harms and enabling the participation of marginalized or excluded communities.
Disclosure recommendations:
Note: By ‘curation’ we specifically refer to human beings on the platform side, employed or contracted, choosing and/or ordering articles as well as news sources, for particular surfaces.
- Disclose surfaces
- Disclose surfaces and ways where human curation is a part of the platform
- Disclose surfaces and ways where humans training models are used for curation
- Disclose principles and policies that guide curation
- Disclose list of curated topics in each <period/reporting cycle>
- g. of “Topic”: Racial Justice, Housing, Food, etc. When you click those links it shows all stories in that topic. Sometimes a term represents a developing story itself like "Infant formula" but those go away after a while. Some terms are standard and product people are posting them on top. E.g. “COVID-19” is top on many news product surfaces.
- Human curators
- Which news stories did human curators give prominence to (display at the top of page/feed) per topic/moment/event/#trend each <period/cycle> without algorithmic assistance?
- Which news stories did human curators give prominence to (as in a. above) but with algorithmic assistance?
Reason: These disclosures will bring to the surface how news publishing entities – large, medium, small, and boutique -- are connected into a platform distribution system. This data may be usable to cross-interrogate data disclosed on other sections to build a holistic picture. It will help to understand the extent to which mainline or mass news media outlets are contributing to discourse vs. other niche outlets that often cover systemic issues continuously.
Disclosure recommendations:
Note: We recognize that any release of lists could be weaponized. The generation of such lists needs to be implemented with guardrails to mitigate disclosure risks. Each platform may have its own approach to this and we recommend they document it as part of this disclosure. For instance, some platforms may document their appeals process for decisions and disclose that along with these lists.
By market (US, Europe/countries, Asia/countries, US-local markets, etc.
- List of entities accorded news publisher status on curated products or by human curators.
- List of entities whose content was recommended algorithmically as “News”
- List of publishers whose stories (claims in stories) were disputed and failed third-party fact-checkers (false or pants on fire..)
- List of publishers who violated content policies and details
- List of publishers who were demonetized and grounds
- List of publishers whose accounts were suspended or deplatformed
- List of publishers whose accounts were restored after suspension
- List of publishers whose content triggered top #trends data (per cycle)
- List of publishers who are labeling News v Opinion
- List of entities whose content is not curated or not recommended as “News” but who have asked for News Publisher status (long tail), i.e. entities who have asked for, but never heard back and were not denied
- List of entities who requested news publisher status by formal means but not granted with grounds (junk publishers, deception, imposters, etc..), i.e. news entities that have been explicitly denied status
Reasons: These disclosures will help analyze the types of stories, posts, entities and accounts that are prominent in the news cycles of a given period. It will help review whether the types of coverage we outlined in the "Obligations to foster discourse" section are prominent or not, and under what conditions.
Disclosure recommendations:
Note:
- The terms used below may not carry the same specific meaning or apply to the distribution system the same way per platform. To avoid becoming platform-specific, we recommend that platforms revise these terms to the closest valid interpretations. For instance in the list below, “prominent” may mean displayed on top of a feed for that topic, trend, event, or moment, for one platform. For other platforms, the appropriate measure of prominence may be “impressions”. A separate but related measure of consumption for video is “views”. Alternatively, each adopting platform may simply customize a list using this as a reference, and add to this recommendation item.
- The usage of “topics”, “trends”, and wordings in hashtags for “trends” are not standardized - they are partly dynamic-cultural, partly taxonomically intuited. We recommend that platforms release along their default schemas for categorization that is publicly visible on their surfaces.
By curated topics/trends/moments/events (e.g. for Twitter) and/or
By market (US, Europe/countries, Asia/countries, US-local markets)
- List the top
- List the top #n most prominent recommendations
- List the top #n impressions
- List the top #n most viewed
- (a,b,c) List of top #n most prominent recommendations/impressions/views only from news publishers (entities that have identified themselves as news publishers, or that the platform otherwise considers as a news publisher)
- (a,b,c) List of top #n most prominent recommendations/impressions/views only from organic human accounts (exclude accounts of journalists connected with one of the news publisher entities)
- List the top #n most shared posts (retweets/shares)
- List of top #n most shared posts only from news publishers
- List of top #n most shared posts only from organic human accounts (exclude accounts of journalists connected with one of the news publisher entities)
- Does the platform maintain a quantitative measure of the percentage of content that got into distribution from the curation side vs the algorithmic recommendation side? If so, define the measure and disclose numbers for recent measurement periods.
(Note: We recognize that there are going to be standardization challenges for terminology such as topics, trends, moments, events, etc., that will make normalized comparisons difficult across platforms. One of the goals of this effort would be to move to standardization after some cycles of disclosure and learning have taken place.)
Reasons: Platforms take remedial actions every day. For news posts violating policy, remedial actions may include demonetization, demotion in ranking, exclusion from prominence in lists, de-platforming of a post itself, and so forth. How long particular false claim-laden stories propagate, especially on systemic issues, before remedial actions are taken, may have significance for rights, harms as well as discourse. Early impressions are harder to reverse in general even as more accurate stories surface from other news publishers. Disclosures about remedial actions taken within distribution systems about news will help oversight entities understand the current role of such actions in relation to harms, rights, and discourse. It may also identify potential gaps and opportunities.
For stories that news publishers may have themselves corrected upstream, remedial action may involve detecting the correction first. The question of whether platforms can re-distribute the corrected story without loss of prominence is somewhat moot since originating corrections have to be detected. (We recognize this is yet another major problem area in the news industry to standardize and signal corrections consistently.)
Disclosure recommendations:
Below, per day/week/per market:
- List of news posts/articles referred to fact-checkers and flagged as disputed and their retweet/share #n counts after the flagging happened.
- List of news posts/articles that violated content policies and were acted on in any way - engagement limited, labeled, demonetized, deplatformed, etc. (*Note: Items 3 and 4 below tagged “News Publisher Corrections”, are related to corrections in news articles at the publishers’ end. Currently, news aggregators and platforms do not have standardized visibility into data showing corrections in real time. We acknowledge these two items are futuristic.)
- *News Publisher Corrections: List of most prominent recommendations of news items that were later corrected by publishers, and the relative distribution data for those posts (This is to capture post-correction distribution (tweets, retweets) for the same story. Let's say WaPo posted the first story at 901am 5/16. And it got {x-impressions, y-shares/retweets, z-something} distribution metrics. At 12:02 pm they made a correction and posted that. Disclose the distribution for the correction in the same {x, y, z} terms)
- *News Publisher Corrections: List of most shared posts of news items that were later corrected by publishers, and the relative distribution data for those posts (same as above, except engagement metric changes from impressions to sharing)
Reason: Opinion journalism – which can often include reporting – is not only engaging, it has an independent strength to draw people into discourse that reportage alone does not. Livestreams may contain both reportage and opinion for instance. Crucial explanatory journalism at critical times by individuals can often elevate the discourse to a shared understanding where the distinction between opinion and reportage seems even unnecessary to some. But in general, when observing on-platform discourse, the role played by opinion journalism (required by ethics for arguments to be based on facts) as opposed to political propaganda (which uses lies, false narratives, and “othering”) is not clear.
The news industry itself (in most markets if not every) has not built a consistent practice of separating and signaling opinion journalism. Broadcast journalism blurred the lines between commentary and reportage before digital journalism arrived. Even when opinion labels exist, it isn’t clear whether they are accurate and whether news organizations are incentivized to mark them accurately. Disclosures will help oversight entities understand how otherwise robust opinion journalism from news publishers is passing through distribution systems and its role in the discourse on systemic issues, and whether distribution policies and design are determining any aspects of the role.
Disclosure recommendations:
Policy
- Are you automatically detecting opinion as distinct from reporting? If so is your system itself designating an opinion label?
- Is your platform separating opinion feeds for topics/developing stories/etc.?
- Is your platform passing opinion label metadata (such as schema.org labels, when present) downstream to the news feed/discovery surfaces for users to see as a label?
- How is your platform handling conflict between a publisher-supplied label and your own auto-detection? (E.g. Publisher says “reportage or news”, but your detection says “opinion”)
- Is your platform subjecting opinion journalism to any form of vetting? If so, what? (e.g. internal validation/assessment for factual basis, or fact-checking..)
Data
By curated topics/trends/moments/events AND/OR
By market (US, Europe/countries, Asia/countries, US-local markets)
- List of top#n most prominent recommendations/impressions/views of opinion posts from news publishers
- List of top#n most shared opinion posts from news publishers
- List of news publishers labeling opinion content formally whose stories you are distributing.
Roundtable Members
Nick Diakopoulos, Associate Professor of Communication and Computer Science, Northwestern University
Jen Granito, Product Manager, Google News
Claire Leibowicz, Head, AI and Media Integrity, Partnership on AI
Arjun Moorthy, CEO, The Factual
Aubrey Nagle, Director of Practice Change, Reframe, Resolve Philly
Mutale Nkonde, CEO, AI For The People
Aviv Ovadya, Harvard-Belfer Fellow, Harvard Technology and Public Purpose Project
Kristy Roschke, Managing Director, News Co/Lab, Walter Cronkite School of Journalism and Mass Communication
Geoff Samek, Product Manager, YouTube
Connie Moon Sehat, Researcher at Large, Hacks/Hackers and Director, News Quality Initiative
Bill Skeet, Chief Product Officer, NOBL Media
Jonathan Stray, Senior Scientist, Berkeley Center for Human-Compatible AI (CHAI)
Cheryl Thompson-Morton, Black Media Initiative Director, Center for Community Media, City University of New York
Subramaniam Vincent, Director Journalism and Media Ethics, Markkula Center for Applied Ethics (Convener)
Eric Han/Katie Joseff/Tara Wadhwa, US Trust & Safety, Tiktok
Jennifer Wilson, Curation Standards Lead, Twitter
This roundtable includes members from current News Distribution organizations (platforms and aggregators). Their participation has been to give input and ideas as stakeholders. Participation does not constitute an endorsement of the recommendations as a whole or in parts by their organizations.
Appendix A.1 Examples for Definitions (1.1) and Curations Policy (1.2) Disclosures
Note: The tables below are examples of what disclosures may look like in the future. We have curated/synthesized a range of current public disclosures/communications from participant platforms and aggregators as responses for 1.1 and 1.2. They are illustrative only, and neither comprehensive nor meant to be a substitute for formal disclosures from a news distributor who implements these recommendations.
|
NDE item No. |
Item |
|
|
1.1 Disclose Definitions |
||
|
1.1.1 |
How does your platform define “news” for both curation and algorithmic classification purposes? |
News is what's happening Mission: We serve the public conversation. |
|
1.1.2 |
How does your platform currently accord “News” site or “News publisher” status to any entity? |
Not publicly disclosed currently. Further: *can be disclosed with some lead time |
|
1.1.3 |
How does that designation affect access for (or promotion of) news publishers *Access to platform policy, exemptions, resources, business partnerships, news surfaces, etc. |
Not available publicly. *Resources and tools for the news industry are listed here. |
|
1.1.4 |
Does your platform define “who is a journalist” for curation and algorithms and if so, what is the definition or standard? |
No. The Curation Style Guide (“Guiding Principles”) under Platform Use Guidelines discusses standards for sourcing, vocabulary use, and curation decisions. It refers to several journalistic standards. |
|
1.1.5 |
Definitions for the following as used by the platform. “Curation”, “Location”, “Topics”, “Trends/Trending”, and “News categories” used on feeds. |
Curation: Manual selection of content. Curators don’t act as reporters or creators of original work, they organize and present compelling content that already exists on Twitter. This content appears in Moments, explanatory content on Trends, in lists and more. The Curation team is responsible for highlighting and contextualizing the best events and stories that unfold on Twitter. Curation works across multiple product surfaces, including Topics, Trends descriptions, and Moments. Our curators are a global, multilingual team looking for the best things happening on Twitter across news, sports, entertainment, and fun. We currently serve 16 markets in five languages (English, Japanese, Arabic, Spanish and Portuguese). Curators receive regular training on accuracy, impartiality, and identifying high-quality content. Trends: “Trends are determined by an algorithm and, by default, are tailored for you based on who you follow, your interests, and your location. This algorithm identifies topics that are popular now, rather than topics that have been popular for a while or on a daily basis..” https://help.twitter.com/en/using-twitter/twitter-trending-faqs Moments: Moments surface the best of what’s happening on Twitter. Moments are created in multiple ways. Some Moments, such as those covering sporting events or TV shows, are created algorithmically to reflect the unfolding conversation while the event is happening. Others are prepared manually by our Curation team. Topics: Topics are a followable product. “We create followable terms (Topics) that help Twitter users find what’s most relevant to them on our platform.” Selection of Topics: https://help.twitter.com/en/using-twitter/follow-and-unfollow-topics |
|
1.1.6 |
How does your platform define and/or detect “Opinion” from news publishers? |
No detection. Brief discussion of opinion and sourcing of viewpoints in Curation Style Guide. |
|
1.2 Disclose Data and Policies About News Curation |
||
|
1.2.1 |
Disclose surfaces and ways where human curation is a part of the platform |
Curation is used for Trends, Moments and Topics. (products) |
|
1.2.2 |
Disclose surfaces and ways where humans training models are used for curation |
Not currently available but planned. |
|
1.2.3 |
Disclose principles and policies that guide human curation |
Curation guiding principles summary Curation Style Guide (detailed) Moments guidelines and principles |
|
1.2.4 |
Disclose surfaces and ways where algorithmic curation is a part of the platform |
Algorithmic curation used in Moments, Trends, Topics. [Detailed info not available yet.] https://help.twitter.com/en/using-twitter/twitter-trending-faqs) |
|
1.2.5 |
Disclose principles and policies that guide algorithmic curation |
Not called out specifically. Largely the same as that which guides human curation. |
|
NDE item No. |
Item |
Google News / Search (Top Stories & News) / Discover .. |
|
1.1 Disclose Definitions |
||
|
1.1.1 |
How does your platform define “news” for both curation and algorithmic classification purposes? |
News informs users about recent or important events. Original news reporting: “provides information that would not otherwise have been known had the article not revealed it. Original, in-depth, and investigative reporting requires a high degree of skill, time, and effort.” (Elevating original reporting announcement) (Note: Google indirectly defines news and original reporting this way in its search rater guidelines document. Input from search raters are applied to algorithms.) |
|
1.1.2 |
How does your platform currently accord “News” site or “News publisher” status to any entity? |
A “site is automatically considered for Google News and news surfaces in Search—no application required.” News articles can show up in Google News or News in search results without a news publisher registering in Google’s News Publisher Center. There are compliance requirements for news sites (and publishing entities behind the sites) through Google’s News Publisher Transparency policy and Google News Policies (content policies). Transparency: News publishers are asked to provide: a) Clear dates and bylines b) Information about the authors, publication, and publisher c) Information about the company or network behind the content and d) Contact information. Content: Google says when it finds news articles not meeting content policies (Dangerous content, Deceptive practices, Harassing content, Hateful content, Manipulated media, Medical content, Terrorist content, Sexually explicit content, Violence & gore, and Vulgar language & profanity), “we may remove the content from our news surfaces. In cases of repeated or egregious violations, a site may be no longer eligible to appear on our news surfaces.” |
|
1.1.3 |
How does that designation affect access for (or promotion of) news publishers *Access to platform policy, exemptions, resources, business partnerships, news surfaces, etc. |
Google recommends publishers set up a publication in Google News here. It provides publishers the following:
|
|
1.1.4 |
Does your platform define “who is a journalist” for curation and algorithms and if so, what is the definition or standard? |
Not directly. In the context of what Google terms “Expertise, Authoritativeness, and Trustworthiness” or EAT of news articles, Google’s search rater guidelines indirectly define “journalistic professionalism” as: “Should contain factually accurate content presented in a way that helps users achieve a better understanding of events.” (Input from search raters is applied to algorithms.) |
|
1.1.5 |
Definitions for the following as used by the platform. “Curation”, “Location”, “Topics”, “Trends/Trending”, and “News categories” used on feeds. |
Topics: The search rater guidelines section on Your Money or Your Life (YMYL) topics has a detailed discussion of how Google handles topics around potential for harm based on accuracy standards. No definition for the word “topic” itself. Google News topics are U.S., World, Business, Technology, Entertainment, Sports, Science, Health. Google News Showcase (paid publisher partnership program being launched per country) allows publishers to curate their news panels in Showcase as well as in Google News. Story curation is done by the publisher. (All stories on Google News or other news surfaces are algorithmically selected, there is no curation.) |
|
1.1.6 |
How does your platform define and/or detect “Opinion” from news publishers? |
Not disclosed. |
|
1.2 Disclose Data and Policies About News Curation |
||
|
1.2.1 |
Disclose surfaces and ways where human curation is a part of the platform |
All stories on Google News or other news surfaces are algorithmically selected, there is no curation. Google News Showcase is a publisher partnership product (per country) where publishers are paid and do the curation. |
|
1.2.2 |
Disclose surfaces and ways where humans training models are used for curation |
Search rate guidelines are used by human raters to provide input to algorithms. |
|
1.2.3 |
Disclose principles and policies that guide human curation |
N/A |
|
1.2.4 |
Disclose surfaces and ways where algorithmic curation is a part of the platform |
Google News, Google Search (News results and Top Stories), Discover, and YouTube |
|
1.2.5 |
Disclose principles and policies that guide algorithmic curation |
The primary content quality principles document, which includes news quality is the Google search rater guidelines. |
|
NDE item No. |
Item |
The Factual |
|
1.1 Disclose Definitions |
||
|
1.1.1 |
How does your platform define “news” for both curation and algorithmic classification purposes? |
We have a list of ~2000 news sites. Any articles from these sites that are not paid are considered news. |
|
1.1.2 |
How does your platform currently accord “News” site or “News publisher” status to any entity? |
To be a news site you must have at least one editor in addition to a journalist. While we don't actively enforce it, you must have an about page disclosing ownership. |
|
1.1.3 |
How does that designation affect access for (or promotion of) news publishers? *Access to platform policy, exemptions, resources, business partnerships, news surfaces, etc. |
We exclude single-person blogs from our index. |
|
1.1.4 |
Does your platform define “who is a journalist” for curation and algorithms and if so, what is the definition or standard? |
No. |
|
1.1.5 |
Definitions for the following as used by the platform. “Curation”, “Location”, “Topics”, “Trends/Trending”, and “News categories” used on feeds. |
Topics: 3 or more articles that are thematically related (usually at a headline level). Trending: Volume of articles in the last 48 hrs on a topic. Categories (those visible to the user): standard ones used by most news sites, i.e. US, World, Business, Science, Health, Culture, Sports. Hundreds of categories not visible to the user are used by our algorithm for classification and rating. Curation: We do not define this. |
|
1.1.6 |
How does your platform define and/or detect “Opinion” from news publishers? |
All articles are graded for how opinionated they are. We don't distinguish between news and opinion pieces as the line is blurred and in many publications, the label is not even applied. |
|
1.2 Disclose Data and Policies About News Curation |
||
|
1.2.1 |
Disclose surfaces and ways where human curation is a part of the platform |
A newsletter: The Factual produces a daily newsletter of the most informative articles from across the political spectrum on trending news topics. An iOS app and Android app - Combines the daily newsletter briefing and our website in a simple app. |
|
1.2.2 |
Disclose surfaces and ways where humans training models are used for curation |
N/A |
|
1.2.3 |
Disclose principles and policies that guide human curation |
|
|
1.2.4 |
Disclose surfaces and ways where algorithmic curation is a part of the platform |
|
|
1.2.5 |
Disclose principles and policies that guide algorithmic curation |
Each article receives a grade between 1-100% based on four metrics:
These four metrics combine to give a single percentage grade, which we interpret as the probability of the article being informative. Grades above 75% are regarded as highly likely to be informative while grades below 50% are less likely to be informative. |
Appendix A.2 - News Distribution Surfaces Examples
In the table below are rough examples from Social Media, Search, and News Aggregators for news distribution combinations as discussed in the Primer section of this document. These are not meant to be comprehensive and are for illustrative purposes only.
| Source | Detail |
|
|
|
|
|
|
|
|
|
|
|
|
|
The Factual |
|
Acknowledgment
The Markkula Center for Applied Ethics thanks Craig Newmark Philanthropies for support of this project.
Notes
1 Discourse meanings: 1. “Processes of argumentation and dialogue in which the claims implicit in the speech act are tested for their rational justifiability as true, correct or authentic” – Stanford Encyclopedia of Philosophy, page on Habermas’ Discourse Ethics Theory. 2. Common meaning: Verbal exchange of ideas in extended communication. (Miriam-Webster, Oxford.)
2 In this document we restrict the scope of news distribution to products from third party entities such as big tech platforms, aggregators, etc. We do not include native newsletters and websites of the news publishers themselves, which do distribute some of the news to the public.
3 “Democratic Role of News Recommenders” and the “Relevance of Algorithms”
4 Report of the Special Rapporteur on the promotion and protection of the right to freedom of opinion and expression, David Kaye, submitted to the Human Rights Council pursuant to Council resolution 34/18. Cites Hannah Arendt’s work on “Lying in Politics”, connecting lies to judgment.
5 In 1987, in his book on media coverage of the Vietnam War, media scholar Daniel Hallin offered a descriptive model for media objectivity. He drew three spheres, which have since been called Hallin’s spheres. He called one of these as the sphere of legitimate controversy. He points that when journalists decide that an issue falls into this sphere, they cover all sides.
6 See for example Democracy and Armed Conflict
7 Stories on systemic issues and visibility to marginalized voices are sourced from Dr Anita Varma’s Solidarity Journalism Initiative’s tracker (UT-Austin’s Center for Media Engagement. See for example Inflation and rent increases are making homelessness worse - The Washington Post; 'Black and Asian unity': attacks on elders spark reckoning with racism's roots - The Guardian
8 How The Insurrection At The Capitol Affected DC Essential Workers | DCist; The Roadblocks to Relief. Latino immigrants sound off on the… | by Madeleine Bair | El Tímpano | Medium
9 The Influence of Episodic and Thematic Frames on Policy and Group Attitudes: Mediational Analysis; for story examples see Miami Is Tearing Itself Apart Over Bonkers Plan to Move Homeless to Island; One man killed, another man injured in shooting on Indy's east side | IndyStar; 1 dead after shooting at north side gas station | Fox59
10 These stories, including from those who don’t call themselves journalists, can add clarity, foster curiosity, and dispel misbeliefs or misunderstandings. See for example What Really Happened During the Texas Power Grid Outage?, Practical Engineering#, YouTube; Darnella Frazier’s Pulitzer Prize “for courageously recording the murder of George Floyd.”
11 Bryan Stevenson of the Equal Justice Initiative used these words to define the word “narrative” in a podcast in 2019.
12 There are examples of research exploring causality between misinformation, misbeliefs and behavior.
View or download "News Distribution Ethics Roundtable Key Principles and Recommendations" in PDF format.
View directly, or right-click to save a copy.