“Draw a ‘word you should never search’ from a gacha, then get sent to the search results a few seconds later.”
That sounds completely modern: search engines, random draws, forced navigation. A child in 1900 would need an explanation of the word “search” before anything else.
But strip away the technology and the structure suddenly looks old.
You are told not to look. You do not know what will appear. It is a little scary. You want to tell a friend. You want to try one more time.
That sounds surprisingly close to school ghost stories, secret games, collecting cards, and chain letters.
The intuition that “humans have not changed that much” is useful, with one correction: the same content does not survive for a century; some of the reasons it feels compelling survive, while the outer delivery system gets replaced.
The short answer: screens changed more than human motives did
Children in 1900 and children today do not live in the same social world. Education, family life, advertising, cities, media, and friendship networks have changed enormously.
Still, historical sources and modern research repeatedly reveal familiar patterns:
- wanting to peek at the unknown;
- enjoying fear when it remains playful and controllable;
- sharing secrets that mark an in-group;
- collecting, completing, and owning;
- trying what peers appear to be doing;
- copying rituals and passing them along;
- reacting to boundaries such as “forbidden,” “rare,” or “almost complete.”
A useful model is therefore:
relatively stable motives × peer-to-peer transmission × the media tools of the era.
Paper, playgrounds, oral tradition, and postal mail became social media, random-reveal interfaces, short video, and generative AI.
The console changed. Not every game mechanic did.
1. From 1890s card albums to modern gacha, the family resemblance is real
By the late nineteenth century, trade cards and cigarette cards were already collected and pasted into albums. The Smithsonian preserves an album dated 1893.[1] Baseball cards around 1909 were also deeply connected to collecting and trading cultures that included children.[2]
Then came postal chains. A U.S. “ancient prayer” chain from 1906 instructed recipients to copy the prayer and send it to nine people, warning of misfortune if they did not. Luck and money chains expanded in the 1920s and 1930s.[3]
In the 1950s and 1960s, folklorists Iona and Peter Opie collected games, rhymes, stories, beliefs, and habits from about 20,000 British children. UCL’s later “Playing the Archive” project brought that material into conversation with play among children today, explicitly examining both continuity and change.[4]
Japan offers another example. Research on school ghost stories has followed how elementary-school children read, tell, reshape, and use those stories in school life.[5][6]
Today the surfaces are different: “do not search” lists, quizzes, gacha, rankings, short-form video, generative AI.
Yet uncertainty, secrets, fear, collection, peers, and repetition keep reappearing.
2. Pattern one: “I am scared, so I want to look” can become play
Fear does not simply produce avoidance.
Children also choose tag games, dark spaces, ghost stories, scary media, and thrill rides where fear is part of the fun.
A 2025 study surveying caregivers of 1,600 children aged 1–17 found that 93% were reported to enjoy at least one form of recreational fear. The preferred activities changed with age.[7]
The important point is not “more fear is always better.”
The enjoyable zone often includes an exit: you can stop, laugh with friends, or know that the danger is not literally happening to you.
A school ghost story works partly because the ghost is a story or game. If the ghost actually occupied the school, it would stop being folklore and become an emergency.
3. Pattern two: a secret can be social glue, not just information
“Do not tell anyone” changes a relationship.
Across five studies involving 452 children aged 3–12, children around age six began to use secret sharing as a strong cue that two people were friends. They treated sharing a secret as more socially informative than merely sharing an object or an ordinary fact.[8]
That makes many childhood cultures easier to understand.
Knowing the seven mysteries of a school, a hidden trick, a rare card, or “that weird site” is not only about the information. It is also about being inside the circle that knows.
Modern limited links and insider memes can activate the same social function.
4. Pattern three: collect it, complete it, make it mine
The durability of card culture is less mysterious once collecting itself is treated as a motive.
Research with preschoolers found that some children already showed meaningful collecting motivation. Some were even willing to incur a social “cost” by sharing in order to obtain a collectible toy they wanted.[9]
Ownership concepts also appear early. Studies with two- and three-year-olds show that ownership affects how children take and allocate resources.[10]
That makes “100 to collect,” “one left,” “my encyclopedia,” and “my result” powerful structures.
Paper cards can become digital achievements, yet the pull of an incomplete set and the feeling of ownership can travel with them.
5. Pattern four: children’s culture spreads when imitation itself becomes play
Humans do not copy only efficient actions.
Experiments on cultural transmission among preschoolers show that even causally unnecessary actions can persist when they are embedded in play and passed between children.[11]
That looks very familiar online.
Follow the same steps. Show the result. Pass it on. Change it slightly.
Chain letters, playground rituals, memes, challenges, and template quizzes are not identical phenomena, but they are relatives in one important respect: copyable play formats travel well.
Research on urban legends adds another piece. Stories producing stronger disgust were judged more worth passing on, and legends containing more disgust motifs were more widely distributed on urban-legend websites.[12]
Enjoyment and transmissibility are not the same variable, but surprise, fear, and disgust can sometimes push both.
6. Why a “do-not-search gacha” is structurally strong
Break the idea into components and it is doing several jobs at once.
| Element | What it creates |
|---|---|
| “Do not search this” | A forbidden information gap |
| Gacha | Incomplete control over the outcome |
| A short countdown | Time to imagine what is coming |
| Possible horror | A peek at stimulation from relative safety |
| Different results | A reason to repeat |
| Shareable outcomes | A natural “What did you get?” conversation |
So “it is scary” is not enough as an explanation.
It combines uncertainty × prohibition × randomness × anticipation × social sharing.
A study of 310 Dutch participants aged 7–17 found that restrictive age labels and violent-content labels increased the attractiveness of fictional video games.[13]
That does not mean writing “forbidden” on anything guarantees popularity. Responses depend on the reason for the restriction, perceived autonomy, content, culture, and age.
7. The most useful framework is three layers: motive, transmission, interface
If you want to reinterpret older culture today, “turn an old ghost story into AI” is too shallow.
Copy the structure, not only the surface.
Deep motive: why the person wants to engage
Curiosity, playful fear, collecting, ownership, mastery, or peeking past a boundary.
Transmission: why it reaches friends
Secrets, comparable results, repeatable procedures, visible peer behavior, surprise, and humor.
A 2026 meta-analysis covering 79 independent reports and more than 9,500 participants found a meaningful overall effect of peer influence in adolescence. Peer norms—information that peers actually engage in a behavior—were more influential than mere peer presence.[14]
On social media, experiments with participants aged 13–21 also found that photos displayed with many Likes were more likely to receive Likes from participants themselves.[15]
Interface: the amplifier available in that era
- 1900s: cards, mail, oral transmission;
- 1950s: playgrounds, magazines, television;
- 1990s–2000s: forums, personal websites, search engines;
- now: gacha, rankings, short video, share buttons, generative AI.
We do not need to invent a new human motive every generation. We do keep inventing new amplifiers.
8. “Virality” should not be a single score
A more useful model separates four outcomes.
Trigger
Does someone want to try it once?
Uncertainty, prohibition, surprise, playful fear, and a clear promise can help.
Repeat
Is there a reason to do it again?
Randomness, incomplete sets, collection, mastery, and changing outcomes can help.
Transmit
Does someone want to send or show it to a friend?
Secrets, comparison, surprise, humor, disgust, and easily explainable results can help.
Identity
Does the result become “my result” or “my character”?
Ownership, quizzes, customization, belonging, and personal records can help.
Something can be fun but fail to spread because Trigger is strong while Transmit is weak.
A chain letter can be mediocre content while remaining highly transmissible because forwarding is built into the format itself.
9. But “humans never change” is not a universal spell
This is where the counterevidence matters.
Scarcity preferences show cultural variation. A study of 366 children aged 4–11 in Israel and Taiwan found that a preference for scarce objects emerged around age seven among Israeli children but did not emerge in the Taiwanese sample.[16]
So “rare = every child everywhere wants it” is too simple.
Adolescent reward sensitivity is also more complicated than “teenagers are reward machines.” Reviews find that results differ by incentive type, magnitude, uncertainty, and social context.[17]
A better model is:
candidate universal factors + age weights + cultural weights + the media tools of the period.
Humans may share recurring tendencies. The parameter values are not identical for everyone.
10. With generative AI, tune the reaction function instead of asking for “a viral idea”
This is where historical analysis becomes practical.
If you ask an AI only for “a viral 2026 website,” it is likely to remix visible current trends.
Instead, specify the underlying dimensions:
uncertainty 0.9
playful fear 0.7
secret 0.8
collection 0.2
peer comparison 0.9
random reveal 1.0
forbiddenness 0.7
identity 0.4
repeatability 0.8
That combination could produce new forms without copying a particular ghost story:
- generate a different fictional school mystery each time;
- let friends compare outcomes;
- occasionally reveal a rare alternate presentation;
- gradually complete a harmless collection.
A different vector—high collection, ownership, scarcity, and exchange but no fear—points from nineteenth-century card culture toward modern digital collecting.
The goal is not to recreate “what was popular in 1900.”
The goal is to extract response patterns that survived across 1900, 1950, 2000, and 2026, then replace only the contemporary interface.
11. A model built only from hits will break immediately
If the dataset contains only successful cards, ghost stories, chain letters, and gacha, every success can be explained after the fact.
You also need failures:
- toys that did not catch on;
- magazine gimmicks that disappeared quickly;
- quizzes nobody shared;
- web toys people used only once;
- games with similar mechanics that produced different regional outcomes.
Only success-plus-failure data lets you estimate which factors actually matter.
The target should not be a museum of old hits. It should be a dataset for estimating human response functions.
12. Especially with children, “effective” and “appropriate to deploy” are separate questions
A mechanic can work and still be inappropriate.
Paid randomized rewards should be treated separately from free random information reveals. A 2026 systematic review synthesized 26 empirical studies from 2017–2025 on adolescent loot-box engagement and problematic outcomes.[18]
In a Japanese sample of 1,615 fourteen-year-old gamers, loot-box purchasers were more likely to meet a measure of problem online gaming. Because the study was observational, it does not prove that loot boxes caused the problem.[19]
Likewise, genuinely forcing users toward dangerous sites, removing exit options, monetizing scarcity pressure, or turning real-world danger into a challenge should not be confused with harmless experimentation on curiosity and play.
You can create surprise without turning the interface into a trap.
Conclusion: if you study 100 years, collect the reason things were passed on
Cards moved from paper to screens.
Chain letters moved from postal mail to social media.
Ghost stories moved from playgrounds to search and video.
Random surprise moved from lots and packs to gacha interfaces.
Underneath, familiar motives keep returning:
I want to know. I want to be safely scared. I want a secret. I want to collect. I want to share with my group. I want the result to feel like mine.
So a century-scale dataset should not ask only “What became popular?”
It should ask:
Why did someone touch it? Why did they repeat it? Why did they pass it to a friend? Why did they make it part of themselves?
Label those across eras, and the result stops being nostalgia. It becomes a parameter set that modern generative systems can actually use.
Humans may not have changed as much as the interfaces have.
Sources
- Smithsonian National Museum of American History, “Cigarette Card Album” (album dated 1893) americanhistory.si.edu
- Smithsonian Magazine / Smithsonian Books, “How the Tobacco Industry Lured Customers with Baseball Cards” (2024), with historical evidence on children collecting and trading cards around the turn of the twentieth century smithsonianmag.com
- Smithsonian Magazine, “Before Chain Letters Swept the Internet, They Raised Funds for Orphans and Sent Messages From God” (2020), including the 1906 U.S. prayer chain and the expansion of luck/money chains in the 1920s–1930s smithsonianmag.com
- UCL Institute of Education, “Playing the Archive: From the Opies to the digital playground” (2025), based on the Opies’ collection from about 20,000 UK children in the 1950s–1960s and comparison with children today ucl.ac.uk
- 吉岡一志, 『学校の怪談』はいかに読まれているか――小学生へのアンケートをもとに, 子ども社会研究 14 (2008) jstage.jst.go.jp
- 吉岡一志, 子ども文化論再考――「学校の怪談」をめぐる教師と子どもたちの編み合わせ, 子ども社会研究 27 (2021) jstage.jst.go.jp
- Recreational Fear Across Childhood: A Cross-Sectional Study of Scary Activities that Children Enjoy (2025), PubMed PMID 40522538 pubmed.ncbi.nlm.nih.gov
- Secret to friendship: Children make inferences about friendship based on secret sharing (2018), total N=452, PubMed PMID 30284884 pubmed.ncbi.nlm.nih.gov
- Collectible toys and decisions to share: I will gift you one to expand my set (2011), PubMed PMID 21288252 pubmed.ncbi.nlm.nih.gov
- Give and take: Ownership affects how 2- and 3-year-olds allocate resources (2019), PubMed PMID 31097201 pubmed.ncbi.nlm.nih.gov
- When the transmission of culture is child’s play (2012), PubMed PMID 22479524 pubmed.ncbi.nlm.nih.gov
- Emotional selection in memes: the case of urban legends (2001), PubMed PMID 11761305 pubmed.ncbi.nlm.nih.gov
- Age and violent-content labels make video games forbidden fruits for youth (2009), N=310, PubMed PMID 19255016 pubmed.ncbi.nlm.nih.gov
- A Meta-Analysis of Susceptibility to Peer Influence Effects in Adolescence: The Role of Method of Influence and Type of Behavior (2026), 79 independent reports, >9,500 participants, PubMed PMID 42383466 pubmed.ncbi.nlm.nih.gov
- The Power of the Like in Adolescence: Effects of Peer Influence on Neural and Behavioral Responses to Social Media (2016), PubMed PMID 27247125 pubmed.ncbi.nlm.nih.gov
- Cultural differences in the development of a preference for scarce objects (2018), N=366, Israel and Taiwan, PubMed PMID 30431290 pubmed.ncbi.nlm.nih.gov
- The Influence of Different Kinds of Incentives on Decision-Making and Cognitive Control in Adolescent Development: A Review of Behavioral and Neuroscientific Studies (2018), PubMed PMID 29875720 pubmed.ncbi.nlm.nih.gov
- Adolescents and loot boxes: a systematic review of behavioral mechanisms and problematic outcomes (2026), 26 empirical studies, PubMed PMID 42202489 pubmed.ncbi.nlm.nih.gov
- Adolescent Problem Gaming and Loot Box Purchasing in Video Games: Cross-sectional Observational Study Using Population-Based Cohort Data (2021), Japanese sample N=1,615 at age 14, PubMed PMID 33560241 pubmed.ncbi.nlm.nih.gov
