Being the audience: why am I seeing this?
Why am I seeing this, and what does the sender think they know about me?
This chapter explains how influence operations choose and study their audiences, and how that choice looks from the receiving end. It helps you work out by what route something reached you, what it assumes about you, and whether your own community is being courted or impersonated, without turning suspicion on your neighbours.
All stages
The mundane answer comes first
The honest first answer to why you are seeing something is usually ordinary. A friend shared it, a recommendation system judged that people like you linger on posts like this, or someone paid to place it in front of residents of your area. None of that is covert. All advertising selects an audience, and most platforms offer a panel explaining why a sponsored post was shown to you and a public archive of adverts. Use both before drawing any conclusion. The first public investigation of Doppelganger relied partly on a public advertising archive.
The stage question has a second half: what does the sender think they know about you? Read the material for its assumptions. It may presume your politics, your worries about money or safety, your faith, your age, where you live, or which neighbours you resent. Those assumptions are the audience analysis, visible from the receiving end.
Operations go where the grievance already is
Documented operations rarely invent a division. They study existing ones and move in. The Internet Research Agency built separate personas for distinct American communities. Arif, Stewart and Starbird examined its accounts in the conversation around Black Lives Matter and found them embedded on both sides of it, each cluster performing the voice its audience expected. Linvill and Warren found a similar division of labour across the agency's Twitter accounts, with distinct account types addressing left-leaning and right-leaning audiences.
Two lessons follow. First, a grievance that is being exploited is still a real grievance; operators choose it because it is real. Second, check the other side of the divide. If your community is being courted with one message, the people you disagree with may be receiving its mirror image, and each side is then shown the other at its worst.
Data voids
Golebiewski and boyd use the term data void for a search query that returns little or no reliable material because few trustworthy sources have written about it. Voids open around breaking news, newly coined phrases and niche subjects. Local affairs are full of them: a planning dispute in a small town may have no authoritative page at all. Whoever publishes first, in the words people actually type, owns the results.
From the receiving end the tell is a search that returns unfamiliar sites, forum threads and videos but nothing from the body that holds the facts. The remedy sits with institutions and newsrooms: publish a plain, findable account early, in the public's own vocabulary.
When your own community is courted or impersonated
Some signs are worth noting, though none is decisive: voices that arrived recently and speak for the group as we and us; praise for the community paired with contempt for a named out-group; positions a notch more extreme than the membership holds; small errors of local knowledge; and pressure to move the conversation to another site or channel.
Every one of these has an innocent reading. Communities gain new members, people who moved away still care, and second-language speakers and fervent converts exist in every movement. The attribution guide's rule applies with full force: describe the pattern and do not name the person. Moderators can respond to conduct such as rule-breaking, link dumping and harassment without ruling on anyone's authenticity. Members can simply ask whether a claim is true, whoever made it.
Techniques you will meet here, and the tell
The operation publishes into a topic that reliable sources have not covered, so its material is what a curious person finds.
The tell: A search on the exact phrase used in the posts returns only unfamiliar sites and nothing from the institution that holds the facts.
Names, hashtags and search terms a community would naturally use are occupied early by someone outside it.
The tell: The obvious handle, page name or hashtag for a local cause belongs to an account nobody in the cause recognises.
Accounts present themselves as members of the audience they are addressing: a local parent, a veteran, a congregation member.
The tell: The persona's whole identity is the group label, with little ordinary life around it and no one who knows them offline.
The voice is tuned to sound like the audience, borrowing its slang, complaints and humour to earn a hearing.
The tell: The folksiness is generic: it fits any town or any movement, and local detail is missing or slightly wrong.
Material is framed so that accepting it feels like loyalty to your group and doubting it feels like betrayal.
The tell: The post tells you what people like us believe before it offers any evidence for the belief.
Audience analysis looks for the neighbouring group a community already distrusts and builds the message around that line.
The tell: Your side is described in detail and with warmth; the other side appears only as a caricature.
Paid placement lets a sender show different messages to narrow slices of a population. This is routine in commercial and political advertising and is used by covert campaigns as well.
The tell: The sponsor named in the advert panel is a page with no history, and your neighbours in a different group are seeing a different advert from it.
Worked example
Priya has a working reading that the Alder Bay footbridge posts aim at trust more than at the bridge. Now she asks who is being addressed, by what route the posts reach them, and what the posts seem to assume about the people reading.
- 1She maps where the posts appear. Most are in two places: a Harbourside traders' noticeboard group and a Hillcrest families group. Traders are told the bridge will kill parking and footfall. Parents are told that traders would put takings before children's safety.
- 2Several of the loudest voices introduce themselves as lifelong residents but get small things wrong. One mentions catching the Sunday ferry, which stopped running years ago. Priya notes the innocent explanations beside each: people who moved away, newcomers, a slip of memory.
- 3She searches the phrase a resident would type, Alder Bay footbridge cost. The council has published only a long committee paper. The top result is a site she has not seen before, laid out like a local newspaper.
- 4A sponsored post about the bridge appears in her own feed. The platform panel says it was shown to people near Alder Bay who follow local news, and the sponsor is a page created the previous month. She records this and draws no conclusion.
- 5She speaks privately to the two group administrators, both people she knows. They confirm that a number of the newer members joined in the same fortnight. All three agree not to accuse anyone in the groups.
Outcome. Priya now knows that two communities are being addressed separately with opposing messages, and that a search on the bridge leads away from any reliable source. She still does not know who is responsible. She asks the council clerk for a one-page plain summary of the costings and schedules a Courier explainer written in the words residents are searching for.
Audience check
Fill this in about one item or a small set of items that reached you or your community. The aim is to record the route and the assumptions, with the innocent explanations alongside. Saved in this browser only.
Name the platform, group, channel or search that put it in front of you.
Tick every route that applies; use the advert panel or archive if it was sponsored.
List the beliefs, worries, loyalties or resentments the material takes for granted in its reader.
Describe the real concern the material speaks to, in terms a fair-minded member of the community would accept.
Look at what people on the other side of the dispute are being shown, then choose.
Search the key phrase and note whether a reliable source appears, and which sources fill the page if none does.
Describe any pattern among voices claiming to speak for the community, and write the ordinary explanation next to each observation.
Decision check: do I understand why this reached me?
0/6If any of these is not true: Do not act on a hunch about who belongs. Go back and record the route and the search result, which are checkable, and leave the question of authenticity open. If a void exists, ask the body that holds the facts to fill it. If conduct in a group breaks its rules, raise the conduct with the moderators and say nothing about identity.
Common mistakes
- ×Gatekeeping. Treating every newcomer, convert or clumsy speaker as an infiltrator damages a community more reliably than any operation could.
- ×Treating targeting itself as sinister. Every advertiser and every campaign selects an audience; what matters is whether the sender is honest about who they are.
- ×Assuming only the other side is being courted. Communities you belong to and agree with are at least as attractive to an operator, because you are less likely to check.
- ×Over-attribution: reading a pattern of new, like-minded voices as an operation when a local controversy has simply brought people in.
- ×Leaving the data void open. Complaining about unreliable search results while no reliable account exists hands the topic to whoever writes one first.
Who exactly are you trying to reach, and what do they already believe and need?
Related
References
- Golebiewski, M., & boyd, d. (2018). Data Voids: Where Missing Data Can Easily Be Exploited. Data & SocietyThe definition of a data void and the conditions under which search results can be captured.
- Arif, A., Stewart, L. G., & Starbird, K. (2018). Acting the Part: Examining Information Operations Within #BlackLivesMatter Discourse. Proceedings of the ACM on Human-Computer Interaction (CSCW)Internet Research Agency accounts impersonated participants on both sides of a divided conversation.
- Linvill, D. L., & Warren, P. L. (2020). Troll Factories: Manufacturing Specialized Disinformation on Twitter. Political CommunicationDistinct account types were assigned to distinct audiences, including left-leaning and right-leaning ones.
- DiResta, R., Shaffer, K., Ruppel, B., Sullivan, D., Matney, R., Fox, R., Albright, J., & Johnson, B. (2018). The Tactics & Tropes of the Internet Research Agency. New Knowledge, report for the US Senate Select Committee on IntelligenceAudience segmentation by community and identity in a documented operation.
- Starbird, K., Arif, A., & Wilson, T. (2019). Disinformation as Collaborative Work: Surfacing the Participatory Nature of Strategic Information Operations. Proceedings of the ACM on Human-Computer Interaction (CSCW)Operations blend into real communities, whose sincere members do much of the work.