“Shadowbanning is not a thing”: black box gaslighting and the power to independently know and credibly critique algorithms
Open accessAbstract
Efforts to govern algorithms have centerd the ‘black box problem,’ or the opacity of algorithms resulting from corporate secrecy and technical complexity. In this article, I conceptualize a related and equally fundamental challenge for governance efforts: black box gaslighting. Black box gaslighting captures how platforms may leverage perceptions of their epistemic authority on their algorithms to undermine users’ confidence in what they know about algorithms and destabilize credible criticism. I explicate the concept of black box gaslighting through a case study of the ‘shadowbanning’ dispute within the Instagram influencer community, drawing on interviews with influencers (n = 17) and online discourse materials (e.g., social media posts, blog posts, videos, etc.). I argue that black box gaslighting presents a formidable deterrent for those seeking accountability: an epistemic contest over the legitimacy of critiques in which platforms hold the upper hand. At the same time, I suggest we must be mindful of the partial nature of platforms’ claim to ‘the truth,’ as well as the value of user understandings of algorithms.
AI snapshot
Written from the paper’s full text.
TL;DR
Cotter introduces "black box gaslighting" to describe how platforms leverage their perceived epistemic authority over opaque algorithms to undermine users' confidence in what they know and destabilize credible criticism. She develops the concept through a case study of the Instagram "shadowbanning" dispute, arguing that platforms can render outside critiques irrational even though their own claim to "the truth" is only partial.
Key findings
- Instagram's denials of shadowbanning were often wholesale and used definitions that did not match influencers' (e.g., CEO Adam Mosseri framing shadowbanning as content takedown), while the platform's own policies on demotion, non-recommendation, and restricted hashtags "correlate with influencers' understanding of shadowbanning."
- Instagram offered three alternative explanations for what influencers called shadowbanning: glitches/bugs, poor content strategy, and serendipity ("Sometimes you get lucky, sometimes you won't").
- The platform's epistemic authority induced self-doubt: influencers reported reinterpreting their own experiences as glitches or personal content failures, and some deferred to Instagram's statements as definitive (e.g., Christina: "I would much rather take the word of Instagram directly").
- Black box gaslighting is not always effective: influencers who had built algorithmic knowledge over time resisted it (e.g., "the official line from Instagram is that shadowbans don't even exist. We all know they do").
Methods
- Qualitative case study of the Instagram shadowbanning dispute, drawing on a broader project on how influencers learn about algorithms.
- Data: semi-structured interviews with Instagram influencers (n = 17, mostly May–November 2018, ~1 hour each) plus online discourse materials (social media posts, blog posts, videos, news articles) collected September 2017–May 2020, with Nexis Uni media coverage searched through July 2021.
- Analysis: inductive qualitative approach informed by constructivist grounded theory, coding actions related to knowledge production, synthesizing themes, and comparing points of tension between influencers' and Instagram's statements.
Limitations
- Stated by the paper: the qualitative methods "do not lend themselves to establishing the prevalence of opinions or beliefs"; findings present a simplified view because influencers' beliefs vary (e.g., by career tenure), which the author flags as an opportunity for future work.
- Evident in the text: the study is a single-platform, single-controversy case, so the concept of black box gaslighting is explicated through one dispute rather than tested across platforms; the author also notes it is "difficult, if not impossible, to know the intentions behind statements made by Mosseri and Instagram surrogates."
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
0 of 81 cited sources checked · 0 supported · 0 flagged
- Findingown data§ Method
Instagram and Facebook's official documents only refer to shadowbanning on Instagram in two blog posts.
“While Instagram and Facebook’s official documents (blog, help, and policy pages) only refer to shadowbanning on Instagram in two blog posts”
Quote in the paperNot a substantial claim - Findingown data§ Method
Influencers typically became aware of a shadowban upon noticing stark drops in engagement anomalous to their average level of engagement.
“Specifically, influencers typically became aware of a shadowban upon noticing stark drops in engagement (e.g., reach, likes, clicks) anomalous to their average level of engagement (e.g., see ).”
Quote in the paperNot a substantial claim - Findingreasoning§ ABSTRACT
In some cases, users' experiences do not match platforms' assertions.
“In some cases, as in the case study that follows, users’ experiences do not match platforms’ assertions.”
Quote in the paperNot a substantial claim - Findingown data§ Method
In spite of Instagram's statement, some influencers persisted in their belief that shadowbanning was real.
“Yet, I observed that, in spite of Instagram’s statement, some influencers persisted in their belief that shadowbanning was real.”
Quote in the paperSubstantial claim - Findingnone given§ Method
Instagram has issued a variety of public statements that depict shadowbanning as a kind of urban myth.
“Instagram has issued a variety of public statements that depict shadowbanning as a kind of urban myth.”
Quote in the paperSubstantial claim
- Findingcitation§ Method
Influencers typically determined they had been shadowbanned by attending to visibility metrics.
“They typically determined they had been shadowbanned by attending to visibility metrics, which is a central practice in influencers’ work, with metrics serving as indicators of status and economic value in the broader landscape of social media influence (Christin & Lewis, Citation2021).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Christin & Lewis, 2021
Not checked in the free run.
- Findingown data§ Method
About half of the influencers interviewed talked about experiencing shadowbanning themselves.
“About half of the influencers I interviewed talked about experiencing shadowbanning themselves.”
Quote in the paperNot a substantial claim - Findingcitation§ Method
In a blog post in June 2020, without refutation, Mosseri explicitly acknowledged users' concerns about shadowbanning - specifically, concerns that Black creators were being disproportionately affected as they engaged in Black Lives Matter activism on the platform.
“In a blog post in June 2020, without refutation, Mosseri explicitly acknowledged users’ concerns about shadowbanning – specifically, concerns that Black creators were being disproportionately affected as they engaged in Black Lives Matter activism on the platform (see ; Bowenbank, Citation2020).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Bowenbank, 2020
Not checked in the free run.
- Findingreasoning§ Method
When Instagram's denials occasionally offer some definition of shadowbanning, they often use a definition that does not fully match that of influencers.
“When Instagram’s denials occasionally offer some definition of shadowbanning, they often use a definition that does not fully match that of influencers.”
Quote in the paperSubstantial claim - Findingreasoning§ Method
From a position of doubt, many influencers have re-interpreted their insights about algorithmic moderation to bring them in line with Instagram's narrative of shadowbanning-as-myth.
“From this position of doubt, many influencers have re-interpreted their insights about algorithmic moderation to bring them in line with Instagram’s narrative of shadowbanning-as-myth.”
Quote in the paperSubstantial claim - Findingown data§ Method
Influencers often look to Instagram to verify details about the platform's algorithms, which affirms the company's perceived epistemic authority in the shadowbanning dispute.
“Influencers often look to Instagram to verify details about the platform’s algorithms, which affirms the company’s perceived epistemic authority in the shadowbanning dispute.”
Quote in the paperSubstantial claim - Findingreasoning§ Discussion
Even influencers who continued to resolutely assert that shadowbanning was real often saw Instagram as being in possession of, but perhaps reluctant to acknowledge, 'the truth.'
“Further, even influencers who continued to resolutely assert that shadowbanning was real often saw Instagram as being in possession of, but perhaps reluctant to acknowledge, ‘the truth.’”
Quote in the paperSubstantial claim - Findingreasoning§ Discussion
While many influencers expressed considerable distrust towards Instagram in general, this distrust did not always render influencers immune to perceptions of Instagram's epistemic authority.
“While many influencers expressed considerable distrust towards Instagram in general, this distrust did not always render influencers immune to perceptions of Instagram’s epistemic authority.”
Quote in the paperSubstantial claim - Interpretationreasoning§ ABSTRACT
The power dynamic between platforms and users underlies the legitimization of knowledge claims about algorithms.
“The problem emerges from the power dynamic between platforms and users that underlies the legitimization of knowledge claims about algorithms.”
Quote in the paperNot a substantial claim - Interpretationreasoning§ ABSTRACT
Black box gaslighting presents a formidable deterrent for those seeking accountability: an epistemic contest over the legitimacy of critiques in which platforms hold the upper hand.
“I argue that black box gaslighting presents a formidable deterrent for those seeking accountability: an epistemic contest over the legitimacy of critiques in which platforms hold the upper hand.”
Quote in the paperSubstantial claim - Interpretationreasoning§ ABSTRACT
Platforms can undermine users' and other stakeholders' perceived capacity to autonomously know algorithms independently of platform-sanctioned 'truth.'
“I argue that platforms can undermine users’ and other stakeholders’ perceived capacity to autonomously know algorithms independently of platform-sanctioned ‘truth.’”
Quote in the paperNot a substantial claim - Interpretationreasoning§ ABSTRACT
Black box gaslighting interferes with efforts to mitigate vulnerabilities in the platform economy and, more broadly, to ensure algorithms operate in the public's best interest.
“I argue that black box gaslighting interferes with efforts to mitigate vulnerabilities in the platform economy and, more broadly, to ensure algorithms operate in the public’s best interest.”
Quote in the paperNot a substantial claim- Rests on an assumption: Credible criticism by users is necessary for efforts to mitigate vulnerabilities in the platform economy and ensure algorithms operate in the public’s best interest.
- Interpretationreasoning§ ABSTRACT
Instagram's denials have provoked considerable confusion.
“Consequently, Instagram’s denials have provoked considerable confusion.”
Quote in the paperSubstantial claim - Interpretationreasoning§ ABSTRACT
By retaining exclusive access to certain information about their algorithms, as well as user data, platforms have achieved a position of epistemic authority.
“By retaining exclusive access to certain information about their algorithms, as well as user data, platforms have achieved a position of epistemic authority.”
Quote in the paperSubstantial claim - Premisecitation§ ABSTRACT
The power dynamic between platforms and users grows, in part, from the information asymmetry between platforms and users, as platforms withhold, obscure, and strategically disclose details about their algorithms.
“This power dynamic grows, in part, from the information asymmetry between platforms and users, as platforms withhold, obscure, and strategically disclose details about their algorithms (Burrell, Citation2016; Pasquale, Citation2015).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Burrell, 2016
Not checked in the free run.
- cites Pasquale, 2015
Not checked in the free run.
- Interpretationreasoning§ ABSTRACT
Understanding how algorithms function, why, and with what effect helps influencers ground and give weight to their critiques.
“Understanding how algorithms function, why, and with what effect helps influencers ground and give weight to their critiques.”
Quote in the paperNot a substantial claim - Premisenone given§ ABSTRACT
Machine learning algorithms often operate independently of human oversight.
“Further, machine learning algorithms often operate independently of human oversight.”
Quote in the paperSubstantial claim - Premisecitation§ ABSTRACT
Even when companies can trace algorithms' work, it still may not be possible to explain why they produce particular models or outcomes.
“Even when companies can trace algorithms’ work, it still may not be possible to explain why they produce particular models or outcomes (Ananny & Crawford, Citation2016; Burrell, Citation2016; Kroll, Citation2018).”
Quote in the paperSubstantial claim0 of 3 sources checked- cites Ananny & Crawford, 2016
Not checked in the free run.
- cites Burrell, 2016
Not checked in the free run.
- cites Kroll, 2018
Not checked in the free run.
- Premisecitation§ ABSTRACT
Platforms have insulated themselves from public scrutiny by sharing little information about their algorithms.
“Platforms have insulated themselves from public scrutiny by sharing little information about their algorithms (Flyverbom, Citation2016; Pasquale, Citation2015).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Flyverbom, 2016
Not checked in the free run.
- cites Pasquale, 2015
Not checked in the free run.
- Premisecitation§ ABSTRACT
While influencers are essential to the success of platforms, they have little say in establishing the conditions of their labor beyond collective organizing and issuing public complaints and critiques.
“While influencers are essential to the success of platforms, they have little say in establishing the conditions of their labor beyond collective organizing and issuing public complaints and critiques (Cunningham & Craig, Citation2019; Duffy, Citation2017).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Cunningham & Craig, 2019
Not checked in the free run.
- cites Duffy, 2017
Not checked in the free run.
- Premisecitation§ ABSTRACT
Greater transparency does not necessarily make algorithms knowable.
“In reality, greater transparency does not necessarily make algorithms knowable (Ananny & Crawford, Citation2016; Kemper & Kolkman, Citation2019).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Ananny & Crawford, 2016
Not checked in the free run.
- cites Kemper & Kolkman, 2019
Not checked in the free run.
- Premisecitation§ ABSTRACT
Algorithmic precarity motivates influencers and other creative laborers on platforms to optimize their practices and content according to their understanding of algorithms.
“This ‘algorithmic precarity’ (Duffy, Citation2020) motivates influencers and other creative laborers on platforms to optimize their practices and content according to their understanding of algorithms (Bishop, Citation2018; Cotter, Citation2019; Stuart, Citation2020).”
Quote in the paperSubstantial claim0 of 3 sources checked- cites Duffy, 2020
Not checked in the free run.
- cites Bishop, 2018
Not checked in the free run.
- cites Cotter, 2019
Not checked in the free run.
- Premisecitation§ ABSTRACT
Algorithms commonly process enormous datasets with a large number of heterogenous properties, and integrate a series of complex statistical and computational techniques, which limits our capacity to comprehend what they do procedurally.
“Algorithms commonly process enormous datasets with a large number of heterogenous properties, and integrate a series of complex statistical and computational techniques, which limits our capacity to comprehend what they do procedurally (Ananny & Crawford, Citation2016; Burrell, Citation2016; Seaver, Citation2014).”
Quote in the paperSubstantial claim0 of 3 sources checked- cites Ananny & Crawford, 2016
Not checked in the free run.
- cites Burrell, 2016
Not checked in the free run.
- cites Seaver, 2014
Not checked in the free run.
- Premisecitation§ ABSTRACT
Platforms have gained substantial market power primarily via 'ownership of the means of behavioral modification'.
“Indeed, platforms have gained substantial market power primarily via ‘ownership of the means of behavioral modification’ (Zuboff, Citation2015, p. 82).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Zuboff, 2015
Not checked in the free run.
- Premisecitation§ ABSTRACT
The opacity resulting from corporate secrecy forms the basis of characterizations of algorithms as black boxes.
“The opacity resulting from such corporate secrecy forms the basis of characterizations of algorithms as black boxes (Pasquale, Citation2015).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Pasquale, 2015
Not checked in the free run.
- Interpretationreasoning§ ABSTRACT
Learning about algorithms does not guarantee visibility or success.
“Yet, learning about algorithms does not guarantee visibility or success.”
Quote in the paperSubstantial claim - Premisecitation§ ABSTRACT
Algorithms are dynamic.
“Further, algorithms are dynamic (Kitchin, Citation2017).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Kitchin, 2017
Not checked in the free run.
- Premisecitation§ ABSTRACT
Algorithms tend to exacerbate conditions of instability and inequity in media and cultural industries.
“In fact, algorithms tend to exacerbate conditions of instability and inequity in media and cultural industries (Duffy, Citation2020).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Duffy, 2020
Not checked in the free run.
- Premisecitation§ ABSTRACT
Organizing or speaking out carries risks, as doing so may jeopardize current or future access to benefits and privileges afforded by relationships and formal partnerships with platforms.
“Moreover, organizing or speaking out carries risks, as doing so may jeopardize current or future access to benefits and privileges afforded by relationships and formal partnerships with platforms (Caplan & Gillespie, Citation2020).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Caplan & Gillespie, 2020
Not checked in the free run.
- Interpretationreasoning§ ABSTRACT
To the degree that algorithms are constantly in flux, it is difficult to draw stable conclusions about them.
“Thus, to the degree that algorithms are constantly in flux, it is difficult to draw stable conclusions about them.”
Quote in the paperSubstantial claim - Premisecitation§ ABSTRACT
Platforms wield significant power over work and employment through the induction of a new mode of management: the 'algorithmic boss'.
“Platforms particularly wield significant power over work and employment, which they accomplish through the induction of a new mode of management: the ‘algorithmic boss’ (Duffy, Citation2020; Kellogg et al., Citation2020; Rosenblat & Stark, Citation2016).”
Quote in the paperSubstantial claim0 of 3 sources checked- cites Duffy, 2020
Not checked in the free run.
- cites Kellogg et al., 2020
Not checked in the free run.
- cites Rosenblat & Stark, 2016
Not checked in the free run.
- Premisecitation§ Method
Instagram restricts certain hashtags in ways that limit the visibility of posts using them.
“Instagram also restricts certain hashtags (e.g., #tagsforlikes, #sunbathing, #ass) in ways that limit the visibility of posts using them (Instagram, Citationn.d.-b).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Instagram, n.d.-b
Not checked in the free run.
- Premisecitation§ ABSTRACT
The routine experience of such self-doubt can cause victims to begin to question their own ability to independently and reliably understand their realities clearly.
“The routine experience of such self-doubt can cause victims to begin to question their own ability to independently and reliably understand their realities clearly (Spear, Citation2019).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Spear, 2019
Not checked in the free run.
- Interpretationreasoning§ ABSTRACT
The information asymmetry between platforms and users allows platforms to credibly extend the boundaries of their authority to assert certain 'facts' about their algorithms, 'facts' not easily rebutted, given the complex, dynamic nature of algorithms.
“The information asymmetry between platforms and users allows platforms to credibly extend the boundaries of their authority to assert certain ‘facts’ about their algorithms, ‘facts’ not easily rebutted, given the complex, dynamic nature of algorithms.”
Quote in the paperNot a substantial claim - Interpretationreasoning§ Method
Stark drops in visibility metrics offer influencers a manifest indication of a change in the normal state of affairs.
“Stark drops in visibility metrics offer influencers a manifest indication of a change in the normal state of affairs.”
Quote in the paperSubstantial claim - Premisecitation§ Method
Instagram uses algorithms to limit the reach of 'borderline' content that does not violate the platform's Community Guidelines, but is otherwise deemed in 'bad taste, lewd, violent or hurtful'.
“Instagram also uses algorithms to limit to reach of ‘borderline’ content that does not violate the platform’s Community Guidelines, but is otherwise deemed (usually algorithmically) in ‘bad taste, lewd, violent or hurtful’ (Constine, Citation2019).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Constine, 2019
Not checked in the free run.
- Interpretationreasoning§ Method
Instagram restricts hashtags when many posts using a hashtag exhibit signals that suggest (to algorithms) that they violate content rules.
“In other words, Instagram restricts hashtags when many posts using a hashtag exhibit signals that suggest (to algorithms) that they violate content rules.”
Quote in the paperSubstantial claim - Interpretationreasoning§ ABSTRACT
The tension between victims' trust of their gaslighters and self-trust founds the basis for self-doubt that inevitably arises from gaslighting.
“The tension between victims’ trust of their gaslighters and self-trust founds the basis for self-doubt that inevitably arises from gaslighting.”
Quote in the paperSubstantial claim - Premisecitation§ Method
A search for restricted hashtags displays only the 'Top posts' and a message stating 'Recent posts from [hashtag] are currently hidden because the community has reported some content that may not meet Instagram's community guidelines.'
“A search for restricted hashtags displays only the ‘Top posts’ and a message stating ‘Recent posts from [hashtag] are currently hidden because the community has reported some content that may not meet Instagram’s community guidelines.’”
Quote in the paperSubstantial claim - Premisecitation§ ABSTRACT
Gaslighting is effective in altering individuals' conception of reality because it is predicated on relationships in which a victim trusts and/or grants authority to their gaslighter.
“Gaslighting is effective in altering individuals’ conception of reality because it is predicated on relationships in which a victim trusts and/or grants authority to their gaslighter (Abramson, Citation2014; Sweet, Citation2019).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Abramson, 2014
Not checked in the free run.
- cites Sweet, 2019
Not checked in the free run.
- Premisecitation§ ABSTRACT
When gaslighters occupy a position of authority, they can further mobilize their authority 'as leverage to demand they be treated with unjustified degrees of credence'.
“When gaslighters occupy a position of authority, they can further mobilize their authority ‘as leverage to demand they be treated with unjustified degrees of credence’ (Abramson, Citation2014, p. 21).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Abramson, 2014
Not checked in the free run.
- Premisecitation§ Method
It is not possible for influencers or platforms to pinpoint which behaviors or content features trigger individual cases of shadowbanning.
“In fact, it is not possible for influencers or platforms to pinpoint which behaviors or content features trigger individual cases of shadowbanning (Myers West, Citation2018).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Myers West, 2018
Not checked in the free run.
- Premisecitation§ Method
Observing and reflecting on visibility metrics represents an important means by which people learn about platform algorithms.
“Observing and reflecting on visibility metrics (e.g., reach, engagement), represents an important means by which people learn about platform algorithms (Cotter, Citation2020).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Cotter, 2020
Not checked in the free run.
- Interpretationcitation§ Method
As influencers exchange insights in communities of practice, patterns across their experiences help crystalize the phenomenon of shadowbanning.
“Further, as influencers exchange insights in communities of practice, via Facebook groups, forums, and group chats (Cotter, Citation2020), patterns across their experiences help crystalize the phenomenon of shadowbanning.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Cotter, 2020
Not checked in the free run.
- Interpretationcitation§ Method
Instagram has been successful in convincing some influencers that they have fallen victim to misinformation and have misinterpreted their experiences in line with the 'myth' of shadowbanning.
“As Are suggested, Instagram has been successful in convincing (some) influencers’ that they have fallen victim to misinformation and have misinterpreted their experiences in line with the ‘myth’ of shadowbanning.”
Quote in the paperSubstantial claim - Interpretationcitation§ Method
Such concerns shape the knowledge influencers construct, particularly that constructed by Instagram strategy gurus whose income depends upon maintaining credibility.
“Such concerns shape the knowledge influencers construct, particularly that constructed by Instagram strategy gurus whose income depends upon maintaining credibility (Cotter, Citation2020).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Cotter, Citation2020
Not checked in the free run.
- Premisecitation§ Discussion
Users have valuable insights to share about algorithms, because algorithmic outcomes arise from highly contextual data inputs that become visible through located, embodied experiences.
“However, users have valuable insights to share about algorithms, because algorithmic outcomes arise from highly contextual data inputs that become visible through located, embodied experiences (Bishop, Citation2019; Cotter, Citation2020).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Bishop, Citation2019
Not checked in the free run.
- cites Cotter, Citation2020
Not checked in the free run.
- Interpretationreasoning§ Method
Instagram has prompted many influencers to question their own judgment and memory of firsthand experiences with and knowledge building around algorithmic moderation.
“In this sense, Instagram has prompted many influencers to question their own judgment and memory of firsthand experiences with and knowledge building around algorithmic moderation.”
Quote in the paperSubstantial claim - Interpretationown data§ Method
In its unequivocal denials, Instagram gave influencers like Christina and Cameron little reason to continue believing shadowbanning was real.
“In its unequivocal denials, Instagram gave influencers like Christina and Cameron little reason to continue believing shadowbanning was real.”
Quote in the paperSubstantial claim - Premisecitation§ Discussion
Discerning the consequences and affordances of algorithms, even for platform owners, often requires more than mere access to source code or design specifications.
“At the same time, it must be said that discerning the consequences and affordances of algorithms, even for platform owners, often requires more than mere access to source code or design specifications (Kemper & Kolkman, Citation2019; Kroll, Citation2018).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Kemper & Kolkman, Citation2019
Not checked in the free run.
- cites Kroll, Citation2018
Not checked in the free run.
- Interpretationreasoning§ Discussion
Transparency practices and black box gaslighting are interrelated.
“Transparency practices and black box gaslighting are interrelated.”
Quote in the paperSubstantial claim - Interpretationreasoning§ Discussion
Black box gaslighting suggests a deterrent for those seeking accountability: an epistemic contest over the legitimacy of critiques in which platforms hold the upper hand.
“Black box gaslighting suggests a deterrent for those seeking accountability: an epistemic contest over the legitimacy of critiques in which platforms hold the upper hand.”
Quote in the paperSubstantial claim - Interpretationreasoning§ Method
Even influencers who do not trust Instagram and feel comfortable publicly criticizing the platform may worry others will not deem them credible.
“Even influencers who do not trust Instagram and feel comfortable publicly criticizing the platform may worry others will not deem them credible.”
Quote in the paperNot a substantial claim - Interpretationreasoning§ Discussion
Through careful statements that leveraged perceptions of epistemic authority, Instagram prompted many influencers to second guess what they knew about algorithmic moderation on the platform.
“Through careful statements that leveraged perceptions of epistemic authority, Instagram prompted many influencers to second guess what they knew about algorithmic moderation on the platform.”
Quote in the paperSubstantial claim - Interpretationreasoning§ Discussion
Black box gaslighting is not always effective.
“It should also be noted that black box gaslighting is not always effective.”
Quote in the paperSubstantial claim - Interpretationreasoning§ Method
Those who make claims that counter Instagram's public statements risk being seen as irrational or uninformed in the face of perceptions that the company is the authority on its own algorithms.
“Yet, it is also true that those who make claims that counter Instagram’s public statements risk being seen as irrational or uninformed in the face of perceptions that the company is the authority on its own algorithms.”
Quote in the paperSubstantial claim - Interpretationown data§ Method
As influencers received Instagram's statements refuting shadowbanning, the epistemic authority granted to Instagram induced confusion and self-doubt, a parallel sense of not being in a position to properly judge the credibility of claims about shadowbanning.
“As influencers have received Instagram’s statements refuting shadowbanning, the epistemic authority granted to Instagram has induced confusion and self-doubt, a parallel sense of not being in a position to properly judge the credibility of claims about shadowbanning.”
Quote in the paperSubstantial claim - Interpretationreasoning§ Discussion
In the context of the shadowbanning dispute, influencers were not without information; they just had access to different information to support the conclusions they drew.
“In the context of the shadowbanning dispute, this means that influencers were not, as Mosseri suggested in recent statements, without information; they just had access to different information to support the conclusions they drew.”
Quote in the paperSubstantial claim - Interpretationcitation§ Discussion
The allowance of an acceptable level of opacity, in addition to practices of strategic obfuscation and visibility management, creates a space for platforms to engage in black box gaslighting, to convincingly suggest that their critics have drawn faulty conclusions as a result of the information they lack, but platforms possess.
“The allowance of an acceptable level of opacity, in addition to practices of strategic obfuscation (Pasquale, Citation2015) and visibility management (Flyverbom, Citation2016), creates a space for platforms to engage in black box gaslighting, to convincingly suggest that their critics have drawn faulty conclusions as a result of the information they lack, but platforms possess.”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Pasquale, Citation2015
Not checked in the free run.
- cites Flyverbom, Citation2016
Not checked in the free run.
- Interpretationreasoning§ Method
Accepting shadowbanning as a myth preserves perceived meritocratic status hierarchies among influencers.
“Thus, accepting shadowbanning as a myth preserves perceived meritocratic status hierarchies among influencers – i.e., ‘I made it of my own accord; others failed not because of platform intervention, but because they were not as good.’”
Quote in the paperNot a substantial claim - Interpretationreasoning§ Discussion
Platforms can subtly neutralize criticism through a technique termed black box gaslighting.
“a technique by which platforms can subtly neutralize criticism, which I termed black box gaslighting.”
Quote in the paperSubstantial claim- Rests on an assumption: The black box nature of algorithms results from corporate secrecy and technical complexity.
- Interpretationreasoning§ Discussion
In the present case study, the black box nature of algorithms granted Instagram ground upon which to stake plausible denials of shadowbanning.
“In the present case study, this nature granted Instagram ground upon which to stake plausible denials of shadowbanning.”
Quote in the paperSubstantial claim- Rests on an assumption: The black box nature of algorithms results from corporate secrecy and technical complexity.
- Interpretationown data§ Method
Instagram comes to be seen as the principal (or only) actor in a position to judge the credibility of claims about its algorithms.
“Instagram, then, comes to be seen as the principal (or only) actor in a position to judge the credibility of claims about its algorithms.”
Quote in the paperSubstantial claim - Interpretationreasoning§ Discussion
Black box gaslighting is not merely a means of shouting down criticism or impeding the ability to speak out; it is a means of configuring users (and other stakeholders) as incapable of assessing algorithms independently of what platforms say about them.
“In this, black box gaslighting is not merely a means of shouting down criticism or impeding the ability to speak out. It is a means of configuring users (and other stakeholders) as incapable of assessing algorithms independently of what platforms say about them.”
Quote in the paperNot a substantial claim - Interpretationreasoning§ Discussion
What seems to matter more is the knowledge influencers built about Instagram's algorithms over time.
“What seems to matter more is the knowledge influencers built about Instagram’s algorithms over time.”
Quote in the paperSubstantial claim - Interpretationreasoning§ Discussion
A critical audience to challenge what algorithms do and how will not affect accountability if platforms can simply render challenges irrational.
“Yet, a critical audience to challenge what algorithms do and how will not affect accountability if platforms can simply render challenges irrational.”
Quote in the paperNot a substantial claim - Interpretationreasoning§ Discussion
Black box gaslighting rests on the black box nature of algorithms.
“Black box gaslighting rests on the black box nature of algorithms, which results from corporate secrecy and technical complexity.”
Quote in the paperSubstantial claim- Rests on an assumption: The black box nature of algorithms results from corporate secrecy and technical complexity.
- Premisecitation§ Discussion
Much of the decisions or outcomes algorithms produce are not planned, programmed, or anticipated.
“Much of the decisions or outcomes algorithms produce are not planned, programmed, or anticipated (Kemper & Kolkman, Citation2019).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Kemper & Kolkman, Citation2019
Not checked in the free run.
- Classified as probably not a substantial claim. 54
- May combine two claims. 28
Probably not a substantial claim: may not pass S4 (based).
Findingreasoning§ ABSTRACTShadowbanning has provoked considerable anxiety among influencers.
“For this reason, shadowbanning has provoked considerable anxiety among influencers.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingreasoning§ ABSTRACTInstagram has not been clear or consistent about precisely what it is denying or acknowledged that elements of influencers' claims ring true.
“Yet, as will be explained, the platform has not been clear or consistent about precisely what it is denying or acknowledged that elements of influencers’ claims ring true.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
Findingnone given§ ABSTRACTInstagram has denied the use of shadowbanning.
“Instagram, for its part, has denied the use of shadowbanning.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Findingown data§ MethodBased on their experiences, influencers tend to understand shadowbanning as resulting from engaging in activity and/or publishing content deemed undesirable by Instagram, which would degrade the user experience.
“Based on their experiences, influencers tend to understand shadowbanning as resulting from engaging in activity and/or publishing content deemed undesirable by Instagram, which would degrade the user experience.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Findingown data§ MethodNearly all statements explicitly referring to shadowbanning were communicated via social media posts or statements to journalists.
“Notably, nearly all statements explicitly referring to shadowbanning were communicated via social media posts or statements to journalists.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingcitation§ MethodAlthough influencers recognize shadowbanning serves a goal of delivering a (subjectively defined) 'good' user experience, they usually do not know the specific cause of their own shadowban.
“Notably, although influencers recognize shadowbanning serves a goal of delivering a (subjectively defined) ‘good’ user experience, they usually do not know the specific cause of their own shadowban.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing).
Findingcitation§ MethodInfluencers say shadowbanning is a means by which Instagram (algorithmically) enforces platform 'rules' to minimize behavior that could be seen as 'spamming' and particularly behavior indicative of bots.
“As Alex Tooby’s aforementioned definition exemplifies, influencers say shadowbanning is a means by which Instagram (algorithmically) enforces platform ‘rules’ to minimize behavior that could be seen a ‘spamming’ and particularly behavior indicative of bots.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
Findingnone given§ MethodInfluencers refer to shadowbanning as a shorthand for occasions when posts or accounts are downranked in or filtered out of Instagram's main feed; Explore, hashtag, and Reels pages; and/or searches without notice.
“Influencers refer to shadowbanning as a shorthand for occasions when posts or accounts are downranked in or filtered out of Instagram’s main feed; Explore, hashtag, and Reels pages; and/or searches without notice.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
Findingnone given§ MethodDuring the data collection period, Instagram first publicly refuted shadowbanning.
“During the data collection period, Instagram first publicly refuted shadowbanning.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Findingreasoning§ MethodIn this statement, Instagram referred specifically to one kind of behavior many influencers believed contributed to shadowbans.
“In this statement, Instagram referred specifically to one kind of behavior many influencers believed contributed to shadowbans.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Findingcitation§ MethodSecond, to a lesser extent, Instagram has suggested that what influencers understand as shadowbanning is simply their own failure to create engaging content.
“Second, to a lesser extent, Instagram has suggested that what influencers understand as shadowbanning is simply their own failure to create engaging content.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Findingcitation§ MethodFirst, Instagram has suggested glitches.
“First, Instagram has suggested glitches.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based). May combine two claims.
Findingnone given§ MethodIn most statements, Instagram offers wholesale denials without much explanation, which makes it difficult to judge exactly what the company is denying.
“However, in most statements, Instagram offers wholesale denials without much explanation, which makes it difficult to judge exactly what the company is denying.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
Findingcitation§ MethodInstagram has attempted to debunk the shadowbanning 'myth' by suggesting three alternative explanations.
“Instagram has attempted to debunk the shadowbanning ‘myth’ by suggesting three alternative explanations.”
Quote in the paperSubstantial claimMay combine two claims.
Findingcitation§ MethodThe first time Instagram explicitly referred to shadowbanning, the platform stated, as conveyed by TechCrunch: 'Shadowbanning is not a real thing, and Instagram says it doesn't hide people's content for posting too many hashtags or taking other actions'.
“The first time Instagram explicitly referred to shadowbanning, the platform stated, as conveyed by TechCrunch: ‘Shadowbanning is not a real thing, and Instagram says it doesn’t hide people’s content for posting too many hashtags or taking other actions’ (Constine, Citation2018).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Constine, 2018
Not checked in the free run.
May combine two claims.
Findingreasoning§ DiscussionIn spite of the confusion black box gaslighting provoked, what influencers knew about Instagram's algorithms helped many see the fault lines in the platform's renarrativization of shadowbanning, to understand statements as omitting or glossing over important information.
“In spite of the confusion black box gaslighting provoked, what influencers knew about Instagram’s algorithms helped many see the fault lines in the platform’s renarrativization of shadowbanning, to understand statements as omitting or glossing over important information.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Findingcitation§ MethodThird, Instagram has suggested that what seems like shadowbanning is beyond the platform's control, a matter of chance.
“Third, Instagram has suggested that what seems like shadowbanning is beyond the platform’s control, a matter of chance.”
Quote in the paperSubstantial claimMay combine two claims.
Findingcitation§ MethodInstagram CEO Adam Mosseri has referred to shadowbanning as having content taken down, which is not how most influencers define shadowbanning.
“For example, Instagram CEO Adam Mosseri has referred to shadowbanning as having content taken down (Facebook, Citation2020), which is not how most influencers define shadowbanning, as discussed above.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Facebook, 2020
Not checked in the free run.
Probably not a substantial claim: may not pass S4 (based). May combine two claims.
Findingnone given§ DiscussionIn the present case study, when Black Instagram influencers engaged in activism as part of the Black Lives Matter movement raised concerns about racial bias and shadowbanning, it prompted Instagram's most forthcoming statements about the phenomenon to date and a promise to address the issue.
“In the present case study, when Black Instagram influencers engaged in activism as part of the Black Lives Matter movement raised concerns about racial bias and shadowbanning (e.g., see ), as discussed, it prompted Instagram’s most forthcoming statements about the phenomenon to date and a promise to address the issue.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown data§ MethodMany influencers trust Instagram's public statements as the definitive source of information about the platform's algorithms and check their insights against details shared by the company.
“As these comments exemplify, many influencers trust Instagram’s public statements as the definitive source of information about the platform’ algorithms and check their insights against details shared by the company.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S2 (checkable).
Premisecitation§ ABSTRACTPlatforms shape social practices, values, and institutions through their ubiquitous presence in and mediation of everyday life.
“Platforms shape social practices, values, and institutions through their ubiquitous presence in and mediation of everyday life (Duffy et al., Citation2019; van Dijck et al., Citation2018).”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Duffy et al., 2019
Not checked in the free run.
- cites van Dijck et al., 2018
Not checked in the free run.
Classified as probably not a substantial claim.
Interpretationreasoning§ ABSTRACTBlack box gaslighting captures how platforms may leverage perceptions of their epistemic authority on their algorithms to undermine users' confidence in what they know about algorithms and destabilize credible criticism.
“Black box gaslighting captures how platforms may leverage perceptions of their epistemic authority on their algorithms to undermine users’ confidence in what they know about algorithms and destabilize credible criticism.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Interpretationreasoning§ ABSTRACTThe black box problem stands in the way of effecting greater accountability through oversight.
“The black box problem stands in the way of effecting greater accountability through oversight.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Premisecitation§ ABSTRACTInfluencers view learning about algorithms as a central component of their work.
“Thus, they view learning about algorithms as a central component of their work (Cotter, Citation2019; Duffy, Citation2020).”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Cotter, 2019
Not checked in the free run.
- cites Duffy, 2020
Not checked in the free run.
Classified as probably not a substantial claim.
Premisecitation§ ABSTRACTAn influencer's success is defined largely by how well they can achieve visibility in algorithmically-ranked feeds.
“An influencer’s success is defined largely by how well they can achieve visibility in algorithmically-ranked feeds (Bishop, Citation2019; Christin & Lewis, Citation2021; Cotter, Citation2019).”
Quote in the paperSubstantial claim0 of 3 sources checked- cites Bishop, 2019
Not checked in the free run.
- cites Christin & Lewis, 2021
Not checked in the free run.
- cites Cotter, 2019
Not checked in the free run.
May combine two claims.
Premisecitation§ ABSTRACTOn Instagram and other social media platforms, influencers and other creative laborers generate considerable value for platforms by engaging other users, which renders them subject to 'algorithmic control'.
“On Instagram and other social media platforms, influencers and other creative laborers generate considerable value for platforms by engaging other users (Craig & Cunningham, Citation2019), which renders them subject to ‘algorithmic control’ (Kellogg et al., Citation2020).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Craig & Cunningham, 2019
Not checked in the free run.
- cites Kellogg et al., 2020
Not checked in the free run.
May combine two claims.
Premisecitation§ ABSTRACTAlgorithms are developed iteratively and via experimentation (e.g., A/B testing).
“They are developed iteratively (Kroll, Citation2018) and via experimentation (e.g., A/B testing) (Seaver, Citation2014).”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Kroll, 2018
Not checked in the free run.
- cites Seaver, 2014
Not checked in the free run.
May combine two claims.
Premisecitation§ ABSTRACTCultivating knowledge about algorithms has helped influencers to resist and/or reign in the demands algorithms impose on their day-to-day work and creative processes, to advocate for fair remuneration, and to mobilize collectively.
“Nevertheless, cultivating knowledge about algorithms has helped influencers to resist and/or reign in the demand’s algorithms impose on their day-to-day work and creative processes, to advocate for fair remuneration, and to mobilize collectively (Bishop, Citation2019; Cunningham & Craig, Citation2019; O’Meara, Citation2019).”
Quote in the paperSubstantial claim0 of 3 sources checked- cites Bishop, 2019
Not checked in the free run.
- cites Cunningham & Craig, 2019
Not checked in the free run.
- cites O’Meara, 2019
Not checked in the free run.
Classified as probably not a substantial claim.
Interpretationreasoning§ ABSTRACTWith data extracted from users, platforms use algorithms to evaluate and hierarchize laborers in ways that ensure their own profitability.
“With data extracted from users, platforms use algorithms to evaluate and hierarchize laborers in ways that ensure their own profitability.”
Quote in the paperSubstantial claimMay combine two claims.
Premisecitation§ ABSTRACTKeeping tight-lipped helps platforms protect algorithms as their 'secret sauce' and prevent people from 'gaming the system'.
“Keeping tight-lipped helps platforms protect algorithms as their ‘secret sauce’ and prevent people from ‘gaming the system’ (Burrell, Citation2016; Pasquale, Citation2015; Ziewitz, Citation2019).”
Quote in the paperNot a substantial claim0 of 3 sources checked- cites Burrell, 2016
Not checked in the free run.
- cites Pasquale, 2015
Not checked in the free run.
- cites Ziewitz, 2019
Not checked in the free run.
May combine two claims.
Premisecitation§ ABSTRACTThe internal logic of machine learning algorithms develops in relation to data produced by user activity and alongside the broader system as interfaces, settings, capabilities, and user populations change.
“The internal logic of machine learning algorithms develops in relation to data produced by user activity (Burrell, Citation2016) and alongside the broader system as interfaces, settings, capabilities, and user populations change (Ananny & Crawford, Citation2016).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Burrell, 2016
Not checked in the free run.
- cites Ananny & Crawford, 2016
Not checked in the free run.
Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.
Premisecitation§ ABSTRACTGaslighting occurs under conditions of power asymmetries and entails prompting someone to question their reality and conform to the gaslighter's will.
“Gaslighting occurs under conditions of power asymmetries (Abramson, Citation2014; Sweet, Citation2019) and entails prompting someone to question their reality and conform to the gaslighter’s will.”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Abramson, 2014
Not checked in the free run.
- cites Sweet, 2019
Not checked in the free run.
Classified as probably not a substantial claim.
Premisecitation§ ABSTRACTPlatforms use algorithms to manage the quality of labor-outputs via 'soft control' intended to nudge laborers' behavior.
“Platforms use algorithms to manage the quality of labor-outputs via ‘soft control’ intended to nudge laborers’ behavior (Kellogg et al., Citation2020; Rosenblat & Stark, Citation2016).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Kellogg et al., 2020
Not checked in the free run.
- cites Rosenblat & Stark, 2016
Not checked in the free run.
May combine two claims.
Interpretationreasoning§ ABSTRACTUsers (reasonably) perceive platforms as the utmost epistemic authority on their own algorithms, because platforms alone have full access to details about their design and functionality, and they generally share very little of this information publicly.
“Users (reasonably) perceive platforms as the utmost epistemic authority on their own algorithms, because platforms alone have full access to details about their design and functionality, and they generally share very little of this information publicly.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Interpretationreasoning§ MethodIt would not be unreasonable to conclude, as reporter Jesselyn Cook did, that 'selective shadow banning is written into Instagram's rulebook.'
“With these details, it would not be unreasonable to conclude, as reporter Jesselyn Cook (Citation2020) did that ‘selective shadow banning is written into Instagram’s rulebook.’”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Cook, 2020
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
Premisecitation§ MethodInstagram algorithmically arranges content in the main feed, Explore, Reels, and hashtag pages according to predictions of what users 'care about most'.
“Instagram algorithmically arranges content in the main feed, Explore, Reels, and hashtag pages according to predictions of what users ‘care about most’ (Mosseri, Citation2021).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Mosseri, 2021
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
Premisenone given§ MethodEach post is ranked according to predicted likelihood of engaging with a post, with higher values resulting in more prominent placement in the different surfaces.
“Each post is ranked according to predicted likelihood of engaging with a post, with higher values resulting in more prominent placement in the different surfaces.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).
Interpretationreasoning§ MethodThe above details correlate with influencers' understanding of shadowbanning.
“The above details correlate with influencers’ understanding of shadowbanning.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
Interpretationreasoning§ MethodPolicy changes may result in an influencer's account or content facing new ranking 'penalties' (demotion or non-recommendation) overnight.
“Moreover, policy changes may result in an influencer’s account or content facing new ranking ‘penalties’ (demotion or non-recommendation) overnight.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductnone given§ MethodThe qualitative methods used do not lend themselves to establishing the prevalence of opinions or beliefs.
“the qualitative methods I used do not lend themselves to establishing the prevalence of opinions or beliefs.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
Premisecitation§ MethodCertain content is rendered non-recommendable.
“Alternately, certain content is rendered non-recommendable (Instagram, Citationn.d.-a), as explained in a blog post (Rosen & Lyons, Citation2019, n.p.):”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Instagram, n.d.-a
Not checked in the free run.
- cites Rosen & Lyons, 2019
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
Premisecitation§ MethodRanking values are specific to individual users and predicted from signals derived from both content features and user characteristics and activity.
“Ranking values are specific to individual users and predicted from signals derived from both content features (e.g., photo vs. video, keywords) and user characteristics and activity (e.g., age, user interactions).”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
Interpretationnone given§ MethodInfluencers may see their visibility drop significantly and without explanation for a variety of reasons, for example if they share content deemed objectionable.
“Influencers may see their visibility drop significantly and without explanation for a variety of reasons, for example if they share content deemed objectionable.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
Premisecitation§ ABSTRACTThose who gaslight may do so unconsciously and/or without an explicit intention of harm.
“Although gaslighting ultimately results in manipulation, those who gaslight may do so unconsciously and/or without an explicit intention of harm (Dorpat, Citation1996).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Dorpat, 1996
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
Premisecitation§ MethodAn algorithm predicts the likelihood that content is 'objectionable' based on signals like user reports or whether it contains certain keywords.
“For this, an algorithm predicts the likelihood that content is ‘objectionable’ based on signals like user reports or whether it contains certain keywords (Owens & Kacholia, Citation2019).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Owens & Kacholia, 2019
Not checked in the free run.
Probably not a substantial claim: may not pass S2 (checkable).
Premisecitation§ ABSTRACTThe purpose of gaslighting is not only to neutralize particular criticisms but to neutralize the very possibility of criticism by undermining the victim's conception of herself as an autonomous locus of thought, judgement, and action.
“the purpose of gaslighting is not only to neutralize particular criticisms that such individuals might lodge, but to neutralize the very possibility of criticism by undermining the victim’s conception of herself as an autonomous locus of thought, judgement, and action.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Spear, 2019
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
Study design and conductnone given§ MethodInfluencers with more years under their belts will have experienced various evolutions in platform governance and technical infrastructure that might render their judgements dissimilar from those of newcomers.
“For example, influencers with more years under their belts, will have experienced various evolutions in platform governance and technical infrastructure that might render their judgements dissimilar from those of newcomers.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Premisecitation§ MethodEngagement (liking, commenting, etc.) serves as a proxy for what users care about.
“Here, engagement (liking, commenting, etc.) serves as a proxy for what users care about (Mosseri, Citation2021).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Mosseri, 2021
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductnone given§ MethodCollecting multiple genres of data permits a degree of triangulation - the ability to trace the recurrence of themes across different sources.
“While collecting multiple genres of data permits a degree of triangulation – the ability to trace the recurrence of themes across different sources – the qualitative methods I used do not lend themselves to establishing the prevalence of opinions or beliefs.”
Quote in the paperNot a substantial claimMay combine two claims.
Interpretationreasoning§ ABSTRACTIn these situations, platforms may engage in black box gaslighting: they may leverage perceptions of their epistemic authority to undermine users' confidence in what they believed to be true.
“In these situations, platforms may engage in black box gaslighting: they may leverage perceptions of their epistemic authority to undermine users’ confidence in what they believed to be true.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Premisecitation§ Method'Reduction' can be accomplished by adjusting initial ranking values of posts in order to make them less visible.
“‘Reduction’ can be accomplished by adjusting initial ranking values of posts in order to make them less visible (Owens & Kacholia, Citation2019).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Owens & Kacholia, 2019
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
Premisecitation§ ABSTRACTThose who are the target of gaslighting may or may not realize they are being gaslit.
“Likewise, those who are the target of gaslighting may or may not realize they are being gaslit (Dorpat, Citation1996).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Dorpat, 1996
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
Premisecitation§ MethodThis policy is part of the 'reduce' component of Facebook's 'remove, reduce, and inform' strategy.
“This policy is part of the 'reduce' component of Facebook’s ‘remove, reduce, and inform’ strategy (Rosen & Lyons, Citation2019, n.p.).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Rosen & Lyons, 2019
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
Study design and conductreasoning§ MethodInfluencers' opinions and beliefs about shadowbanning vary for a variety of reasons and the findings necessarily present a simplified view.
“Moreover, influencers’ opinions and beliefs about shadowbanning vary for variety of reasons and the findings that follow necessarily present a simplified view.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable). May combine two claims.
Interpretationreasoning§ DiscussionWhile platforms do benefit from unparalleled access to certain information about their algorithms, this does not make them the sole authority on all knowledge claims about algorithms.
“While platforms do benefit from unparalleled access to certain information about their algorithms, this does not make them the sole authority on all knowledge claims about algorithms.”
Quote in the paperNot a substantial claimMay combine two claims.
Interpretationreasoning§ DiscussionWhile black box gaslighting can interfere with what people know about algorithms, what people know about algorithms may alternately help them recognize and resist black box gaslighting.
“While black box gaslighting can interfere with what people know about algorithms, what people know about algorithms may alternately help them recognize and resist black box gaslighting.”
Quote in the paperSubstantial claimMay combine two claims.
Interpretationreasoning§ MethodInstagram's statements about shadowbanning avoid obvious falsehoods but omit important clarifying information, such as a clear and consistent definition of shadowbanning, which permits Instagram's alternative explanations to take shape.
“while Instagram’s statements avoid obvious falsehoods, they omit important clarifying information, for example a clear and consistent definition of shadowbanning, which permits Instagram’s alternative explanations to take shape.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable).
Interpretationreasoning§ DiscussionBased on this article's findings, how we treat user understandings of algorithms has consequences for governance efforts.
“Based on this article’s findings, I suggest that how we treat user understandings of algorithms has consequences for governance efforts.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable). May combine two claims.
Interpretationreasoning§ DiscussionPlatforms' claim to 'the truth' is only partial, and we should be mindful of the incompleteness of their knowledge when platforms refute critiques.
“Although it is sometimes inconvenient for platforms to admit, their claim to ‘the truth’ is only partial, and we should be mindful of the incompleteness of their knowledge when platforms refute critiques.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.
Interpretationreasoning§ MethodThe carefully laid truths in Instagram's renarrativization of shadowbanning make it compelling and have provoked second-guessing and confusion among influencers.
“In fact, it is the carefully laid truths in Instagram’s renarrativization of shadowbanning that makes it compelling and which has provoked second-guessing and confusion among influencers, which I will now discuss.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable).
Interpretationreasoning§ DiscussionDifferent vantage points afford different insights.
“Different vantage points afford different insights.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Premisecitation§ DiscussionInstagram's user data allows for various macro-level insights, as guided by a core belief in platforms as neutral actors.
“Instagram’s user data allows for various macro-level insights, as guided by a core belief in platforms as neutral actors (Gillespie, Citation2010).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Gillespie, Citation2010
Not checked in the free run.
May combine two claims.
Interpretationreasoning§ DiscussionBlack box gaslighting is a means of hardening perceptions of platforms' epistemic authority on their algorithms, and, as a result, undermining the credibility of outside critics.
“It is a means of hardening perceptions of platforms’ epistemic authority on their algorithms, and, as a result, undermining the credibility of outside critics.”
Quote in the paperSubstantial claimMay combine two claims.
Interpretationreasoning§ DiscussionConsequently, Instagram has maintained the principal authority to confirm or deny knowledge claims about algorithmic moderation on the platform and, so, destabilize criticism.
“Consequently, Instagram has maintained the principal authority to confirm or deny knowledge claims about algorithmic moderation on the platform and, so, destabilize criticism.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Interpretationcitation§ MethodOver time, Instagram has come closer to affirming shadowbanning.
“Over time, Instagram has come closer to affirming shadowbanning.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.
Interpretationnone given§ MethodAs influencers look to Instagram to verify information, they reinforce the positioning of Instagram as the principal arbiter of truth about its algorithms and rely less on their own experiences and insights.
“As influencers like Jessica look to Instagram to verify information, they reinforce the positioning of Instagram as the principal arbiter of truth about its algorithms and rely less on their own experiences and insights.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.
Interpretationnone given§ DiscussionPublic outcry is integral in effecting change, but only when perceived as credible criticism.
“This kind of public outcry is integral in effecting change, but only when perceived as credible criticism.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Interpretationreasoning§ MethodAt face value, Instagram's statements refute specific ideas about shadowbanning, which narrows the scope of the issue.
“At face value, Instagram’s statements refute specific ideas about shadowbanning (e.g., shadowbanning results from using too many hashtags and/or constitutes content removals), which narrows the scope of the issue.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable).
Interpretationreasoning§ DiscussionThe findings suggest an imperative to protect users' capacity to draw conclusions about algorithms that part with 'official truths' certified by platforms.
“In short, the findings suggest an imperative to protect users’ capacity to draw conclusions about algorithms that part with ‘official truths’ certified by platforms.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable).
Interpretationreasoning§ DiscussionThis knowledge offers tools for identifying and critically reflecting on the platform's attempts (whether deliberate or not) to re-write what they knew.
“This knowledge offers tools for identifying and critically reflecting on the platform’s attempts (whether deliberate or not) to re-write what they knew.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S2 (checkable).
Interpretationreasoning§ DiscussionRelying uncritically on platforms' statements as indices for evaluating the credibility of users' claims risks reinforcing perceptions of the exclusive epistemic authority of platforms.
“Relying uncritically on platforms’ statements as indices for evaluating the credibility of users’ claims risks reinforcing perceptions of the exclusive epistemic authority of platforms.”
Quote in the paperNot a substantial claim
The paper relies on this without support.
Credible criticism by users is necessary for efforts to mitigate vulnerabilities in the platform economy and ensure algorithms operate in the public’s best interest.
“I argue that black box gaslighting interferes with efforts to mitigate vulnerabilities in the platform economy and, more broadly, to ensure algorithms operate in the public’s best interest.”
The paper relies on this without support.
The black box nature of algorithms results from corporate secrecy and technical complexity.
“Black box gaslighting rests on the black box nature of algorithms, which results from corporate secrecy and technical complexity.”
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Content moderation challenges are compounded by opaque communication, resulting in a crisis of trust.
“These challenges are compounded by opaque communication surrounding content moderation (Cotter, 2023), resulting in a crisis of trust (Zuckerman & Rajendra-Nicolucci, 2023).”
From Stuck in the middleware with you: the challenges of capitalizing a market-oriented approach to platform governance (2025) · cites this paper as Cotter, 2023
Cites
The Black Box Society: The Secret Algorithms That Control Money and Information
Pasquale, F. (2015) · doi:10.4159/harvard.9780674736061 · Cited by 3,796 · see citing works
How the machine “thinks”: Understanding opacity in machine learning algorithms
Burrell, J. (2016) · Big Data & Society · doi:10.1177/2053951715622512 · Cited by 3,196 · cited 7× · see citing works
“Keeping tight-lipped helps platforms protect algorithms as their ‘ secret sauce ’ and prevent people from ‘ gaming the system ’ (Burrell, 2016; Pasquale, 2015; Ziewitz, 2019).”
BackgroundMethodInfluentialBig other: Surveillance Capitalism and the Prospects of an Information Civilization
Zuboff, S. (2015) · Journal of Information Technology · doi:10.1057/jit.2015.5 · Cited by 2,978 · cited 1× · see citing works
“Indeed, platforms have gained substantial market power primarily via ‘ ownership of the means of behavioral modi fi cation ’ (Zubo ff , 2015, p. 82).”
Backgroundvan Dijck, J., Poell, T., & de Waal, M. (2018) · doi:10.1093/oso/9780190889760.001.0001 · Cited by 2,951 · see citing works
Algorithms at Work: The New Contested Terrain of Control
Kellogg, K. C., Valentine, M. A., & Christin, A. (2020) · Academy of Management Annals · doi:10.5465/annals.2018.0174 · Cited by 2,732 · see citing works
Gillespie, T. (2010) · New Media & Society · doi:10.1177/1461444809342738 · Cited by 2,522 · cited 2× · see citing works
“Instagram ’ s user data allows for various macro-level insights, as guided by a core belief in platforms as neutral actors (Gillespie, 2010).”
BackgroundAnanny, M., & Crawford, K. (2018) · New Media & Society · doi:10.1177/1461444816676645 · Cited by 1,836 · see citing works
If...Then: Algorithmic Power and Politics
Bucher, T. (2018) · doi:10.1093/oso/9780190493028.001.0001 · Cited by 1,223 · see citing works
Thinking critically about and researching algorithms
Kitchin, R. (2016) · Information Communication & Society · doi:10.1080/1369118x.2016.1154087 · Cited by 1,189 · see citing works
Playing the visibility game: How digital influencers and algorithms negotiate influence on Instagram
Cotter, K. (2019) · New Media & Society · doi:10.1177/1461444818815684 · Cited by 1,025 · see citing works
Self-citationGillespie, T. (2014) · Media Technologies · doi:10.7551/mitpress/9042.003.0013 · Cited by 1,016 · see citing works
(Not) Getting Paid to Do What You Love: Gender, Social Media, and Aspirational Work
Duffy, B. E. (2017) · Yale University Press · doi:10.12987/yale/9780300218176.001.0001 · Cited by 931 · cited 3× · see citing works
“Digital in fl uencers are a type of microcelebrity who have accrued a large number of followers and make their living by creating content, building relationships with and communities among users, and promoting products and services (Abidin, 2016; Du ff y, 2017).”
BackgroundAbidin, C. (2016) · Social Media + Society · doi:10.1177/2056305116641342 · Cited by 682 · see citing works
Managing visibility on YouTube through algorithmic gossip
Bishop, S. (2019) · New Media & Society · doi:10.1177/1461444819854731 · Cited by 535 · cited 4× · see citing works
“However, users have valuable insights to share about algorithms, because algorithmic outcomes arise from highly contextual data inputs that become visible through located, embodied experiences (Bishop, 2019; Cotter, 2020).”
BackgroundInfluentialWest, S. M. (2018) · New Media & Society · doi:10.1177/1461444818773059 · Cited by 445 · see citing works
TURNING UP THE LIGHTS ON GASLIGHTING
Abramson, K. (2014) · Philosophical Perspectives · doi:10.1111/phpe.12046 · Cited by 422 · see citing works
Sweet, P. L. (2019) · American Sociological Review · doi:10.1177/0003122419874843 · Cited by 411 · see citing works
Caplan, R., & Gillespie, T. (2020) · Social Media + Society · doi:10.1177/2056305120936636 · Cited by 283 · see citing works
Transparent to whom? No algorithmic accountability without a critical audience
Kemper, J., & Kolkman, D. (2019) · Information, Communication & Society · doi:10.1080/1369118x.2018.1477967 · Cited by 260 · cited 3× · see citing works
“At the same time, it must be said that discerning the consequences and a ff ordances of algorithms, even for platform owners, often requires more than mere access to source code or design speci fi cations (Kemper & Kolkman, 2019; Kroll, 2018).”
BackgroundAnxiety, panic and self-optimization: Inequalities and the YouTube algorithm
Bishop, S. (2018) · Convergence: The International Journal of Research into New Media Technologies · doi:10.1177/1354856517736978 · Cited by 253 · see citing works
“Gaming the System”: Platform Paternalism and the Politics of Algorithmic Visibility
Petre, C., Duffy, B. E., & Hund, E. (2019) · Social Media + Society · doi:10.1177/2056305119879995 · Cited by 225 · see citing works
Algorithmic precarity in cultural work
Duffy, B. E. (2020) · Communication and the Public · doi:10.1177/2057047320959855 · Cited by 141 · see citing works
Kroll, J. A. (2018) · Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences · doi:10.1098/rsta.2018.0084 · Cited by 132 · see citing works
Creator Governance in Social Media Entertainment
(2019) · doi:10.1177/2056305119883428 · Cited by 112 · see citing works
Rethinking gaming: The ethical work of optimization in web search engines
Ziewitz, M. (2019) · Social Studies of Science · doi:10.1177/0306312719865607 · Cited by 102 · see citing works
Epistemic dimensions of gaslighting: peer-disagreement, self-trust, and epistemic injustice
Spear, A. D. (2023) · Inquiry · doi:10.1080/0020174x.2019.1610051 · Cited by 97 · see citing works
The Drama of Metrics: Status, Spectacle, and Resistance Among YouTube Drama Creators
Christin, A., & Lewis, R. (2021) · Social Media + Society · doi:10.1177/2056305121999660 · Cited by 63 · see citing works
Flagged first, then by citations
Research works printed but not linked: 13
- Are C., 2020, Instagram quietly admitted algorithm bias
- Charmaz K., 2014, Constructing grounded theory
- Corbin J., 2014, Basics of qualitative research
- Cotter, K. (2020). Critical algorithmic literacy: Power, epistemology, and platforms [Doctoral dissertation]. Michigan State University.
- Craig D., 2019, Social media entertainment
- Dorpat T. L., 1996, Gaslighthing
- Duffy B. E., 2019, Social Media + Society, 1
- Flyverbom M., 2016, International Journal of Communication, 10, 110
- Gillespie T., 2018, Custodians of the internet
- Mosseri A., 2021, Shedding more light on how Instagram works
- O’Meara V., 2019, Social Media+ Society, 5, 1
- Rosenblat A., 2016, International Journal of Communication, 10, 3758
- Stuart F., 2020, Ballad of the bullet
Cited by
Sorted by citations
Strategic Invisibility: How Creators Manage the Risks and Constraints of Online Hyper(In)Visibility
Stegeman, H. M., Are, C., & Poell, T. (2024) · Social Media + Society · doi:10.1177/20563051241244674 · Cited by 45 · see citing works
Hallinan, B., Rothenstein, O., & John, N. A. (2025) · Communication and Change · doi:10.1007/s44382-025-00018-9
“These challenges are compounded by opaque communication surrounding content moderation (Cotter, 2023), resulting in a crisis of trust (Zuckerman & Rajendra-Nicolucci, 2023).”
“Creator burnout is real”: Risk, responsibility, and un/speakability in the creator economy
Nguyen, R., & Duffy, B. E. (2026) · New Media & Society · doi:10.1177/14614448261459667 · Cited by 1 · see citing works
- Authors
- Cotter, K.
- Affiliations
- Pennsylvania State University (US)
- Corresponding author
- Cotter, Kelley
- ORCID
- Cotter, Kelley
- Venue
- Information, Communication & Society
- Publisher
- Informa UK Limited
- Published
- 2021-10-28
- Volume
- 26
- Issue
- 6
- Pages
- 1226-1243
- ISSN
- 1369-118X (ISSN-L)
- 1468-4462
- OpenAlex
- W3210971126
- Semantic Scholar
- 8dc366fa13f51f6285243a9ef3dca6e7ff15301e
- Cited by
- 149
- References
- 27
- Citations by year
- 2026: 45 · 2025: 30 · 2024: 34 · 2023: 22 · 2022: 17 · 2021: 1
- FWCI
- 23.16
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- Information Communication & Society (publishedVersion, cc-by-nc-nd)
- Funders
- NSF (SES-1946678)
- Topics
- Social Sciences › Safety Research › Ethics and Social Impacts of AI
- Social Sciences › Sociology and Political Science › Misinformation and Its Impacts
- Social Sciences › Law › Freedom of Expression and Defamation
- Keywords
- algorithmic governance, algorithmic opacity, shadowbanning, Instagram influencers, platform accountability, epistemic authority, corporate secrecy, social networking sites
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-06.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-06.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
