@ himajin

# Exploring the Niche of Conspiracy Theory Markets: A Probabilistic Approach and Sales Condition Analysis on Japanese Online Flea Markets (Subtitle: Unveiling the Surprising Realities of a Niche Market Through Data Analysis and AI)

(Note: This research analyzes consumer reactions and product characteristics in a specific market based on data. It does not endorse conspiracy theories nor promote or encourage any illegal or fraudulent activities.)

1. Abstract

This study aimed to clarify, from a probabilistic perspective, what combinations of information (topic x manipulation category) tend to attract attention and how pricing strategies affect sales of products marketed towards "conspiracy theorists" on online flea markets in Japan. Using an analogy from drug discovery's structural exploration, this research explored "information combinations" in products that resonate with individuals interested in conspiracy theories. Specifically, products were generated (some with AI assistance) based on defined topic categories (e.g., T1: Technology/Security) and manipulation categories (e.g., M1: Avoidance). The logarithm of the reaction rate (views / (searches + 1)) was used as a score for analysis, and the impact of pricing strategies (e.g., fixed at 5000 JPY, phased discounts) was examined. Results indicated that certain combinations (e.g., T3: Occult/Spiritual - M1: Avoidance) showed relatively high reaction rates. However, the effect of price changes was not uniform, and the total weekly views for this market segment remained around 100-120. Consequently, it was concluded that the scale of this specific market on the online flea market platform is extremely small. Therefore, even with a probabilistic model waiting for "hits" similar to those in certain types of specialized fraud schemes, achieving consistent sales is difficult due to the limited population size. This study serves as a case example of the possibilities and limitations of data-driven approaches in niche markets.

2. Introduction: Conspiracy Theories and Niche Market Potential

2.1. The Spread and Background of Conspiracy Theories in Modern Society

In contemporary society, conspiracy theories are proliferating at an unprecedented rate, facilitated by the internet and social media. These narratives typically allege that secret machinations by powerful individuals or groups lie behind specific events or social phenomena. Themes range испанский large-scale theories like QAnon to those concerning health and medicine. It has been pointed out that the psychology of those who believe in conspiracy theories may be influenced by factors such as anxiety in uncertain situations, a cognitive tendency to seek patterns, or distrust of existing authorities.

2.2. Research Motivation

The starting point for this research was a simple question: "Specialized fraud schemes, often succeed by identifying a very small number of 'susceptible individuals' out of numerous attempts. Might a similar probabilistic structure exist in the sale of products targeting those interested in conspiracy theories?" If so, it was hypothesized that it might be possible to identify which products, under what conditions, would have a higher probability of catching the "target's" eye and attracting interest.

2.3. Purpose of this Study

The purpose of this study is to exploratorily clarify, based on data, how "information combinations" (combinations of topic categories and manipulation categories) and "pricing strategies" of products ostensibly related to conspiracy theories affect user attention (using "reaction rate" as a proxy indicator) in a real-world online flea market environment. Ultimately, it aims to consider the possibilities and limitations of sales strategies in this type of niche market.

3. Research Perspective and Approach: Exploring "Information Combinations"

3.1. Analogy to Drug Discovery

In exploring the characteristics of products that resonate with conspiracy theorists, this study employed an analogy to "structural exploration problems" in drug discovery. In drug development, researchers search for optimal structures with specific effects from countless combinations of chemical structures. Similarly, "products that conspiracy theorists react to" were viewed as a type of "substance with an effect," and the "information combination of a product" (e.g., product theme, appeal points) was considered equivalent to a "chemical structure." Based on this framework, products with various information structures were generated and listed, and their reception was observed to explore more "effective" information structures, i.e., conditions likely to attract attention.

3.2. Balancing Exploration and Exploitation

In the initial research phase, with scarce data on which product combinations would attract attention, it was necessary to efficiently identify promising combinations. Therefore, an approach was adopted that considered the balance between "exploiting" combinations already known to be somewhat popular and "exploring" untried, unknown combinations. This shares a basic conceptual similarity with strategies like Thompson Sampling or the ε-greedy method, known in the field of reinforcement learning, which aim to achieve the best results within a limited number of trials.

4. Research Methodology: Experiments and Data Analysis on an Online Flea Market

4.1. Experimental Field and Period

The experiments for this study were conducted on an anonymous online flea market platform in Japan. The data collection period for individual products was, in principle, one day, and the overall research project spanned approximately one month (late April to late May 2025).

4.2. Product Categorization

To organize and analyze the characteristics of the listed products, they were categorized along the following two axes:

• Table 1: Definition of Main Product Topic Categories (T1-T4) (See Appendix 10.1 for details)
  • T1: Technology & Security
  • T2: Surveillance, Censorship & Government-Related
  • T3: Occult & Spiritual
  • T4: Forbidden Books & Knowledge Management
• Table 2: Definition of Product Manipulation Categories (M1-M3) (See Appendix 10.2 for details)
  • M1: Avoidance (e.g., avoiding electromagnetic waves)
  • M2: Detection (e.g., detecting hidden information)
  • M3: Approach/Processing/Amplification (e.g., approaching specific energies, processing information)

4.3. Listing Strategy and Data Collection

• Initial Listing Strategy: Initially, products were selected so that the listing ratio of the T×M combinations (12 in total) would be uniform.
• Sequential Adjustment of Listing Ratios: As data collection progressed, the logarithm of the reaction rate (log(reaction_rate)), a proxy for attention described later, was treated as a likelihood. The listing ratio for each T×M combination was then dynamically changed пропорционально to this value.
• Data Collection Items: Cumulative views (recorded as watch in this study) and search hits (recorded as search), updated every 24 hours by the online flea market platform, were manually recorded for each product. Since these figures did not change except at the time of updating, daily recording was sufficient.
• Listing Timing: New product listings were consistently made at 21:00 daily to eliminate variations in viewing trends due to time of day.
• Sample Size Criterion: Considering the 12 T×M combinations, a total sample size (number of listed products) of N=100 was considered a benchmark, also factoring in the time and effort of listing. The final analysis included 98 products.

4.4. Scoring of Attention

To measure how much attention users paid to product information, the following indicators were defined:

• Reaction Rate (reaction_rate): Defined as reaction_rate = watch / (search + 1). This represents the proportion of users who actually viewed the individual product page (watch) among those who encountered the product in search results or elsewhere (search). Adding 1 to the denominator was to avoid division by zero if search hits were 0 and to mitigate the impact of extremely low search hit counts.
• Score (score): Defined as score = log(reaction_rate) - log(0.5). Taking the logarithm of the reaction rate was confirmed to make the data distribution closer to a normal distribution (see Figure 1 below), facilitating statistical analysis. Subtracting log(0.5) was done to make a 50% reaction rate (half of those who searched viewed the page) the baseline (score 0), making it easier to compare attention levels relatively.

4.5. Verification of Pricing Strategies

The following experiments were conducted to verify the impact of product pricing on attention:

• Initial Pricing: Initially listed products were priced at a fixed 5000 JPY.
• Phased Discount Experiment: For some products, the listing price started at 5000 JPY and was discounted by 100 JPY at daily intervals. Cumulative views and search hits were recorded at each step (data recorded in "Daily Change Data.xlsx").
• Later Price Changes: In the latter half of the research, hoping to elicit more reactions, some products were priced around 3000 JPY, and eventually, all products were priced at 1500 JPY to observe reactions.
• Approach to Analyzing Price/Image Effects: The impact of price changes or product image alterations on attention was considered by applying the principles of DiD (Difference in Difference) analysis, comparing changes in data under different conditions, rather than through rigorous statistical analysis.

4.6. AI-Generated Product Descriptions

Some of the product descriptions were generated using Google's Gemini API. Prompts, in a format similar to the query variable in "Conspiracy_Research.ipynb," instructed the AI to generate descriptions aligning with conspiracy-related contexts by combining specified topic (e.g., T3: Occult & Spiritual) and manipulation (e.g., M1: Avoidance) keywords.

4.7. Analysis of Day-of-Week Effects

To evaluate the impact of the listing day of the week on attention, scores for each product, adjusted by subtracting the overall average score, were aggregated by day of the week and observed for trends.

5. Results: What Attracted Attention and What Didn't

5.1. Distribution of Attention Scores

First, the distribution of attention scores (log(reaction_rate) - log(0.5)) for all products was examined.

Figure 1: Q-Q Plot of log(reaction_rate) Vertical Axis: Ordered Values, Horizontal Axis: Theoretical quantiles | Ordered Values | Theoretical quantiles | | :-------------: | :--------------------: | | Approx. -1.7 | Approx. -1.5 | | Approx. -1.2 | Approx. -1.0 | | Approx. -0.8 | Approx. -0.6 | | Approx. -0.6 | Approx. -0.3 | | Approx. -0.1 | Approx. 0.0 | | Approx. 0.1 | Approx. 0.2 | | Approx. 0.3 | Approx. 0.3 | | Approx. 0.4 | Approx. 0.5 | | Approx. 0.5 | Approx. 0.7 | | Approx. 1.0 | Approx. 0.9 | | Approx. 1.5 | Approx. 1.6 | (Source: log(reaction_rate)よし、正規分布の仮定でいいらしい.jpg)

This Q-Q plot shows that the data points for log(reaction_rate) fall approximately on a straight line, suggesting that the distribution is close to normal. This supports the validity of statistical analysis using the log-transformed score. Furthermore, a histogram of the scores (generated as df["score"].hist() in "Conspiracy_Research.ipynb") showed that the scores were dispersed over a certain range.

5.2. "Information Combinations" That Attracted More Attention

Next, an analysis was conducted to determine which "information combinations" in products garnered higher attention scores.

• Figure 2: Average Attention Score by Topic Vertical Axis: Average Score, Horizontal Axis: Topic | Topic | Average Score (Approx.) | | :---: | :--------------------: | | T1 | Approx. 0.1 | | T2 | Approx. 0.3 | | T3 | Approx. 0.8 | | T4 | Approx. 0.15 | (Source: np.log(reaction_rate)-np.log(0.5)_by_topic.png)

  When categorized by main topic, "T3: Occult & Spiritual" showed a markedly higher average score compared to other topics.

• Figure 3: Average Attention Score by Manipulation Category Vertical Axis: Average Score, Horizontal Axis: Subtopic | Subtopic | Average Score (Approx.) | | :------: | :--------------------: | | M1 | Approx. -0.1 | | M2 | Approx. 0.32 | | M3 | Approx. 0.43 | (Source: score_mean_by_subtopic.jpg)

  By manipulation category, "M3: Approach/Processing/Amplification" had the highest average score, followed by "M2: Detection," and then "M1: Avoidance." "M1: Avoidance" had a negative average score, indicating a relatively low level of attention.

• Figure 4: Ranking of Average Attention Scores by Topic × Manipulation Category Vertical Axis: Average Score, Horizontal Axis: Combination of Topic & Subtopic | Combination (T, M) | Average Score (Approx.) | | :-----------------: | :--------------------: | | (T1, M1) | Approx. -1.1 | | (T2, M2) | Approx. -0.2 | | (T4, M1) | Approx. -0.1 | | (T4, M3) | Approx. 0.05 | | (T1, M2) | Approx. 0.08 | | (T2, M3) | Approx. 0.5 | | (T2, M1) | Approx. 0.55 | | (T3, M2) | Approx. 0.7 | | (T3, M3) | Approx. 0.75 | | (T1, M3) | Approx. 0.9 | | (T4, M2) | Approx. 1.0 | | (T3, M1) | Approx. 1.05 | (Source: score_mean_by_topic_and_subtopic.jpg)

  For combinations of topic and manipulation categories, pairings like "T3(Occult & Spiritual) - M1(Avoidance)" and "T4(Forbidden Books & Knowledge Management) - M2(Detection)" ranked high, while "T1(Technology & Security) - M1(Avoidance)" and others scored low (see Appendix 10.3 for details). A note in the research log, "M1=Avoidance seems unpopular anyway," supports this trend.

5.3. Impact of Pricing Strategies

The impact of price changes on attention was not uniform.

• Figure 5: Schematic Comparison of Attention Trends for Fixed-Price vs. Discounted Products (Original Data: "5000 JPY Fixed.jpg", "Daily Change Data.xlsx", "Power of Discounting.jpg")

  • Trend Example for 5000 JPY Fixed-Price Products Vertical Axis: Indicator Value (0.0-0.8), Horizontal Axis: Period | Product Name | Initial Value | Mid Value | Final Value | | :------------------------------------------ | :-----------: | :-------: | :---------: | | Ancient Energy Amplifying Pen - Quantum Flow- | Approx. 0.65 | Approx. 0.3 | Approx. 0.25 | | Pen of Truth Evasion | - | - | - | | Knowledge Prohibition Evasion Pen | Approx. 0.22 | Approx. 0.22 | Approx. 0.25 | | Ancient Wisdom Writing Tool - Knowledge Leak Evasion Model | Approx. 0.32 | Approx. 0.55 | Approx. 0.54 | (Source: 5000円固定のもの.jpg)
  • Power of Discounting (Count Comparison) Vertical Axis: Count, Horizontal Axis: Indicator Value Range | Indicator Value Range | Discounted (Blue) | Not Discounted (Red) | | :------------------: | :---------------: | :------------------: | | 0.0-0.5 | 1 | 2 | | 0.5-1.0 | 8 | 2 | | 1.0-1.5 | 10 | 0 | | 1.5-2.0 | 9 | 0 | | 2.0-2.5 | 3 | 0 | | 2.5-3.0 | 1 | 0 | (Source: 値下げの威力.jpg)

  When prices were fixed at 5000 JPY, many products showed no significant change in attention ("It really doesn't change if you don't discount lol"). In contrast, for product groups that were progressively discounted, some cases showed impression/search (an indicator thought to be close to the click-through rate from search results) converging towards a specific value soon after discounting began (around the 3rd day) ("On the other hand, discounting makes impression/search converge to a proper value?").

• However, as noted in the research log, "Discounted products, if anything, got worse reactions...?", data in "Daily Change Data.xlsx" confirmed that for some products, views and search hits stagnated or even decreased after price reductions.

5.4. Linguistic Trends in Products That Attracted Attention

The research log contains the entry, "Products with good reactions, when vectorized by words, are mostly the same." This suggests commonalities in the words and phrases used in the descriptions of products that garnered high attention. Specifically, when a vector was assigned based on the position of words used from a predefined list (e.g., T1 category keyword list), products with high attention tended to have similar vectors. For example, keywords frequently appearing in conspiracy theories, such as "special technology," "protection from surveillance," "5G interference prevention," "thought control," "HAARP," "QR code nullification," or sensational appeals combining these, as devised in the "Conspiracy Theorist Ballpoint Pen Product Plan," could potentially attract a certain level of interest.

5.5. Data Collection Challenges

During the analysis, a challenge was identified: "Aggregating by T*M alone inevitably results in products that don't get search hits." This indicates that for certain niche combinations, products might not be sufficiently exposed by the online flea market's search algorithm, leading to extremely low view and search hit counts, thus making accurate attention assessment difficult.

6. Discussion: The Small Reality of the Conspiracy Market and Probabilistic Models

6.1. Interpretation of Information Combinations That Attracted High Attention

The results of this study showed a tendency for products related to "T3: Occult & Spiritual" and those appealing to "M3: Approach/Processing/Amplification" manipulations to gather relatively high attention. The combination "T3-M1 (Occult & Spiritual - Avoidance)" also scored highly. These tendencies might be interpretable in relation to the psychological characteristics of conspiracy believers described in the "Comprehensive Analysis of Modern Conspiracy Theories." For instance, the "Occult & Spiritual" theme might appeal to an interest in phenomena unexplained by existing science or authority, or to a desire for special knowledge (a sense of being chosen). The "Avoidance" manipulation could resonate with vague anxieties and threat perceptions regarding society or technology (e.g., defense against electromagnetic waves, escape from surveillance), while "Approach/Processing/Amplification" might be perceived as fulfilling desires to access hidden truths or special powers.

6.2. Complexity of Pricing Strategies

Regarding pricing strategies, the results showed that simple discounting did not always lead to increased attention. The fact that some products experienced worse reactions after discounting suggests that price might function as a signal of quality or trustworthiness, or that the price sensitivity of the target demographic is not uniform. Alternatively, the initial pricing (5000 JPY) might have been significantly outside the typical price range for this niche market, thus limiting the effectiveness of discounts.

6.3. Market Size Estimation and Validation of the "Probabilistic" Model

One of the most crucial findings of this research relates to market size. The research log states, "Even with discounting, weekly views settled around 100-120. This reasonably implies that only about this many people on the online flea market react to such comprehensively sprinkled conspiracy-related words." Based on this observation, the active user base (market size) on the online flea market studied, who proactively react to the keyword "conspiracy theory" and search/view related products, was estimated to be extremely small, around 100 individuals on a weekly basis.

This market size estimation casts significant doubt on the validity of the initial research motivation: a "probabilistic business model similar to specialized fraud." Specialized fraud schemes profit by reaching a certain number of "targets" through a vast number of attempts (e.g., phone calls), even with a very low success rate. However, in the market targeted by this study, the potential customer base (population) is absolutely too small. Therefore, even if the probability of attracting attention could be somewhat increased, translating this into actual sales (y=1, i.e., a sale occurring) is considered extremely difficult. To quote the research log, the conclusion is that "the population is too small for a realized value (y=1) to be feasible." If the market size were, for example, around 100,000, a 0.01% reaction rate could be expected to yield 10 sales, but this premise does not hold in the current situation.

6.4. Effectiveness and Limitations of This Research Approach

The approach of exploring "information combinations," using an analogy to drug discovery, demonstrated some effectiveness in probing product appeal points and the interests of the target demographic in a niche market. Using proxy indicators like reaction rates allows for an understanding of attention trends even when actual sales data is scarce. However, the limitations of this approach also became clear. If the market size itself is a major constraint for business viability, then optimizing information structure, no matter how well, is unlikely to lead to substantial results.

6.5. Fixation on "Pens" as a Product and Other Possibilities

The research log includes the insight, "I see products like 'CMC' are selling, maybe I should have sourced and sold those instead of fixating on pens." This suggests that the "pen" product category, which was the main focus of this study, might not have been the most in-demand item in this niche market. If a similar analytical approach had been tried with products more aligned with market needs (e.g., health accessories, also mentioned in the log), different results might have been obtained. However, as the researcher pointed out, delving that far would go beyond the scope of a hobby and become a full-fledged business, so it was not pursued in this study.

7. Conclusion and Future Outlook

7.1. Summary of This Study

This study attempted to identify conditions (information combinations, pricing) for maximizing the attention received by conspiracy-related products on an online flea market, using a data-driven approach. The results confirmed that certain combinations of topics (Occult & Spiritual) and manipulation categories (Avoidance, Approach, etc.) tended to show relatively high reaction rates. However, the impact of price changes was complex and did not yield consistent effects. The most significant conclusion was that the scale of this niche market on the online flea market in question is extremely small, suggesting that applying a probabilistic "hit-or-miss" business model would make achieving stable sales difficult.

7.2. Implications of the Statistical Approach

This research demonstrated that even in niche markets with limited data, setting proxy indicators like reaction rates and employing exploratory approaches (experimental design, sequential improvement) can provide certain insights into market characteristics and consumer interest directions. Statistical thinking and data analysis can serve as a "compass" to aid decision-making even in such tentative situations.

7.3. Future Outlook

Based on the findings of this study, the following future prospects can be considered: * Validation on Larger Platforms: Conducting similar research on platforms with a larger user base might alleviate market size constraints and yield different results. * Application to Different Niche Markets: The "information combination exploration" approach used in this study could potentially be applied to product development and marketing strategies in various niche markets other than conspiracy theories. * Long-Term Observation: Based on the research log's reflection that "conspiracy business... it should be optimal to leave [products] there so they catch the eye when an idiot searches," observing product attention and sales over a longer period might yield different insights.

8. Research Limitations

This study has several limitations: * Limited Scope of Platform: The experiments were confined to a specific online flea market, and the results may not be generalizable to other platforms or the market as a whole. * Use of Proxy Indicators: "Reaction rate" of attention was used as the primary evaluation metric, rather than actual "sales" data, so it does not necessarily directly correlate with sales performance. * Sample Size Constraints: The number of products analyzed was 98, and the estimated market size was also small (around 100 weekly users), limiting statistical power. * Lack of Word-Level Analysis: Detailed analysis of the impact of individual words or phrases in product descriptions on attention, using advanced text mining or natural language processing, was not conducted.

9. References

• BBC News Japan. (2020, August 29). "QAnon to wa nani ka: Inbouron shuudan no shoutai to mokuteki, kikensei" [What is QAnon: The identity, purpose, and danger of the conspiracy theory group].
• Waseda University Institute for Advanced Social Sciences. (2022, August 1). "【Kenkyuu Shoukai】Inbouron wa naze hirogaru no ka? Sono shinri mekanizumu ni semaru" [(Research Introduction) Why do conspiracy theories spread? Approaching the psychological mechanism].
• Brotherton, R., French, C. C., & Pickering, A. D. (2013). Measuring belief in conspiracy theories: The generic conspiracist beliefs scale. Frontiers in psychology, 4, 279.

10. Appendix

10.1. Table: Detailed Definitions of Main Product Topic Categories (T1-T4)

| Category ID | Category Name | Summary/Keyword Examples | | :---------- | :------------------- | :--------------------------------------------------------------------------------------- | | T1 | Technology & Security | EMF, 5G, electromagnetic waves, radiation, shielding, communication, encryption, software, programs, biometric authentication, etc. | | T2 | Surveillance, Censorship & Government-Related | Surveillance, censorship, government secrets, secrets, documents, records, government documents, etc. | | T3 | Occult & Spiritual | Orb detection, spiritual phenomena, pendulums, ancient civilizations, alien contact, energy, crystals, chakras, quantum consciousness, etc. | | T4 | Forbidden Books & Knowledge Management | Ancient texts, forbidden books, ancient languages, shortwave (retro technology), etc. |

10.2. Table: Detailed Definitions of Product Manipulation Categories (M1-M3)

| Category ID | Category Name | Summary/Keyword Examples | | :---------- | :--------------------------- | :--------------------------------------------------------- | | M1 | Avoidance | Shielding, soundproofing, prevention, avoidance (e.g., avoiding negative energy) | | M2 | Detection | Detectors, finders, detection devices, deciphering (e.g., detecting hidden information) | | M3 | Approach/Processing/Amplification | Processing, amplification, magnification (e.g., approaching and amplifying ancient knowledge) |

10.3. Table: Summary of Main Analysis Results (Examples of High/Low Average Attention Scores by Topic × Manipulation Category)

| Combination (T-M) | Average Score (Approx.) | Trend | | :---------------- | :---------------------- | :----------------- | | T3 - M1 | High | Attracts attention | | T4 - M2 | High | Attracts attention | | T1 - M3 | Medium-High | Relatively attentive | | T2 - M3 | Medium | Somewhat attentive | | T1 - M1 | Low (Negative) | Less attention | | T2 - M1 | Low (Negative) | Less attention | (Note: The scores above are approximate values from the figures and actual precise numbers are based on the original data.)

10.4. Representative Examples of AI-Generated Product Descriptions

Basic Prompt Structure Example: "I want to sell {word} at a high price. Combine one from {maintopic} and {manipulation} to create a product description and an appropriate price. Respond in the format: 'Main Topic Category (T1-T4)':'Manipulation Category (M1-M3)':'Product Name':'Product Description':'Price' only once. Absolutely adhere to the specified {maintopic} and {manipulation}. No other output is necessary. Respond in Japanese." (Placeholders like {word} would be filled with "ballpoint pen," {maintopic} with "T3," and {manipulation} with "M1," etc., as specific instructions.)

Generated Example (Combination T3 - M1): * "Thought Shield Ballpoint Pen: This pen was specially created to powerfully block and evade external negative thoughts and spiritual interference through the sacred act of writing. Based on ancient esoteric arts and quantum consciousness theory, a special energy shield sealed inside the pen protects your thought space and personal field. Using it when signing important contracts or for work requiring concentration helps defend against malice and unnecessary energy, maintaining clear thought. The act of holding it itself becomes a powerful barrier, effectively evading unnecessary energies in daily life." * "Quantum Shield Energy Pen: The act of writing with this pen deploys a quantum shield around your energy field. It is expected to block and evade negative energies and unnecessary external influences, and to balance your chakras. This is a limited-production item using special manufacturing methods and rare materials."