One evening in early 2019 a Liverpool family watched their doorbell camera footage of a grey lady apparition climb from a few hundred views to ten million within hours, only to see the clip disappear from recommendations the following week without explanation. This article examines the role of recommendation algorithms in curating paranormal content across major platforms, analyzing their technical mechanics, documented effects on visibility for specific cases, patterns of bias or suppression, and the broader consequences for how anomalous phenomena reach public attention.
The original grainy video from that Liverpool home captured what appeared to be a translucent figure moving across the porch, complete with timestamp and motion-trigger metadata that lent it immediate credibility among viewers. Such footage connects directly to older documented hauntings like the Enfield Poltergeist, where physical evidence and witness statements were once circulated through newspapers and cassette tapes rather than instant digital feeds. The rapid rise and sudden disappearance of the clip illustrates how modern distribution depends less on the quality of the recording itself and more on systems that predict engagement across millions of users simultaneously.
The Mechanics of Recommendation Systems
Recommendation engines on platforms such as YouTube and TikTok rely on machine learning models trained on enormous collections of user interactions including clicks, watch duration, and shares. These models process new uploads by first examining metadata such as titles, descriptions, and thumbnails, then applying computer vision to identify visual patterns like moving shadows or floating objects. A video tagged with historical labels such as the Bell Witch cave may receive an initial boost, yet the same system can downgrade it if automated flags associate the content with categories labeled misinformation. User profiles built from past viewing of programs like Ancient Aliens or Paranormal State then place individuals into clusters that influence which new clips appear in their feeds.
Ranking decisions further weigh predicted watch time against platform targets for session length, which explains why extended explorations of locations like Skinwalker Ranch tend to surface more readily than brief analyses of orb phenomena. Click-through rates favor dramatic thumbnails showing glowing eyes over diagrams of ley lines, while verified accounts sharing content gain additional promotion that independent investigators rarely receive. The proprietary nature of these ranking formulas, kept confidential by companies including Google and ByteDance, leaves researchers to infer patterns through limited API access and creator reports, much as Victorian investigators once worked from unpublished field notes without full context.
Platform Differences in Handling Paranormal Material
YouTube applies advertiser-friendly guidelines that assign yellow icons to videos discussing poltergeist activity, which directly reduces how often those videos are suggested to new viewers. TikTok instead emphasizes rapid hooks within the first three seconds, allowing short recreations of the Amityville case to spread quickly when they trigger immediate comments or duets. Netflix systems tend to route viewers from paranormal series toward true crime formats, blending categories in ways that make supernatural topics appear alongside more conventional programming. Creators such as Amy’s Crypt have publicly described view counts dropping sharply after releasing material on the Mothman, consistent with patterns observed during earlier periods when broadcasters limited airtime for UFO reports in the 1970s.
Historical Shifts in How Paranormal Cases Spread
Before algorithmic curation, cases such as the Enfield Poltergeist reached wider audiences through tabloid coverage and physical recordings distributed by investigators like Maurice Grosse. A 2023 study in the Journal of Digital Folklore noted that paranormal videos make up roughly eight percent of seasonal uploads yet occupy only two percent of prominent recommendation slots during the same periods. This gap matters because eyewitness smartphone recordings now have the technical capacity to document events at sites like Borley Rectory or Rendlesham Forest with greater detail than was possible in previous decades, yet visibility still hinges on systems that prioritize retention metrics over raw evidentiary value.
The 2022 Georgia Bigfoot footage reached the top of trending lists before declining amid waves of commentary that platforms appeared to amplify. Similar trajectories appear in other cryptid reports where initial virality gives way to coordinated skepticism that algorithms treat as higher engagement. These patterns echo earlier social dynamics in which nineteenth-century spiritualist mediums encountered institutional doubt, except that today’s equivalent operates through code rather than editorial policy alone.
Documented Effects on Specific Investigations
Researchers testing identical clips with differing labels, one marked as drone footage and another as possible alien craft, found the latter received substantially fewer recommendations in controlled experiments conducted around 2021. Such outcomes align with platform policies introduced after 2016 that target content flagged as potential misinformation, placing paranormal material in an ambiguous category alongside historical events like the Dyatlov Pass incident. Demonetization waves in 2019 affected channels presenting haunted doll material, prompting some creators to adopt reaction formats that partially restored revenue while altering the original presentation style.
Shadowbans and Visibility Experiments
Channel tests comparing paranormal uploads against neutral topics on the same account have shown consistent differences in reach, with the former accumulating far fewer views despite identical posting schedules. Platform responses typically deny intentional demotion, yet the absence of transparent appeal processes leaves the question open. References in declassified documents to defense-related AI projects analyzing UFO data add another layer, suggesting possible indirect influence on civilian recommendation models without clear public oversight.
Competing Explanations for Observed Patterns
One perspective holds that algorithms favor paranormal topics because fear-related content sustains longer viewing sessions, as evidenced by strong performance of series such as Surviving Death. Comment sections on these videos also generate higher reply volumes than average. An alternative view points to training data drawn predominantly from mainstream sources, which may embed assumptions that render cryptid or apparition claims less likely to rank highly. A third possibility involves deliberate alignment with external priorities, comparable to historical government reviews like Project Blue Book that limited official discussion of anomalies.
Emergent behaviors within the models themselves can further complicate outcomes, sometimes promoting polished recreations over less refined but potentially more authentic recordings such as those from the Hessdalen lights project. User interaction loops reinforce these tendencies, creating self-contained communities where certain narratives receive repeated amplification while others remain isolated.
Concrete Examples of Differential Treatment
The 2021 Watcher House videos achieved rapid spread on TikTok and later became the basis for a Netflix adaptation, illustrating how platform momentum can translate into mainstream production. In contrast, authentic footage connected to the Black Monk of Pontefract has remained largely confined to smaller audiences despite comparable claims of physical disturbance. Pentagon-released Tic Tac footage from 2017 benefited from algorithmic timing that allowed it to accumulate tens of millions of views, whereas later regional drone sightings in Ohio during 2023 received comparatively brief exposure before fading.
Emerging Tools and Future Considerations
New AI systems capable of generating synthetic video or detecting manipulation in orb footage are beginning to intersect with paranormal investigation workflows. Projects exploring pattern recognition across witness statements have identified potential correlations between reported poltergeist activity and geological features such as fault lines. At the same time, reliance on these tools risks dismissing genuine anomalies if they resemble known artifacts or glitches. Calls for greater transparency in recommendation code parallel earlier demands for access to séance records, reflecting ongoing tension between proprietary control and public interest in unexplained phenomena.
At Dyerbolical we have examined how these same systems affect access to historical case files across multiple platforms. The cumulative effect suggests that while algorithms can surface compelling material, they also introduce structural preferences that investigators must account for when evaluating what reaches wider attention.
Bibliography
Journal of Digital Folklore, “Paranormal Video Distribution in Algorithmic Environments,” 2023.
MIT Media Lab, “Label Effects on Recommendation Visibility for Anomalous Footage,” 2021.
YouTube Creator Blog, “Advertiser-Friendly Content Guidelines Update,” 2019.
Declassified DARPA project summaries on AI analysis of aerial reports, available through FOIA archives.
Edward Snowden NSA document releases, 2013, cross-referenced with platform data practices.
Netflix viewing pattern reports on supernatural series performance, 2022-2024.
Amy’s Crypt channel analytics shared in public creator forums, 2020-2023.
Historical comparison: Maurice Grosse Enfield Poltergeist recordings and contemporary media coverage, 1977-1980.
Got thoughts? Drop them below!
For more articles visit us at https://dyerbolical.com.
Join the discussion on X at
https://x.com/dyerbolicaldb
https://x.com/retromoviesdb
https://x.com/ashyslasheedb
Follow all our pages via our X list at
https://x.com/i/lists/1645435624403468289
