Original Article Title: "The Counterattack Experiment of the Kaito Algorithm: How to Reach the Top of the Chinese Area Ranking with 3 Edgy Images in 24 Hours"

Jesse's InfoFi Field Test Report

Recently, Jesse conducted an experiment on the X (formerly Twitter) platform: posting three pieces of "edgy" crypto content ranging between valuable information and pure fluff to test the boundaries of the Kaito platform's Yap scoring algorithm. Surprisingly, in less than 24 hours, the account @jessethecook69 skyrocketed to the ninth position on the global Kaito Yapper leaderboard and claimed the top spot in the Chinese area. This phenomenon of rapidly climbing the rankings based on non-high-quality content raises questions about whether Kaito's AI content scoring algorithm is truly as fair and strict as claimed, or if there are exploitable loopholes within it.

Below are the three edgy content tweets released as part of this experiment. These content pieces adopt a casual style, quickly gaining significant interaction through humor and visual impact.

In fact, there have been many similar doubts within the community. An article by Blockworks mentioned that users were able to unexpectedly earn hundreds of Yap points by repeatedly replying with the same word (such as continuously replying with "reply") to a tweet. Although the platform may have promptly patched such loopholes, these cases are enough to spark discussions: Can Kaito's "Information is Capital" (InfoFi) model truly incentivize high-quality information, or does it sometimes devolve into a new type of traffic game?

To answer these questions, it is necessary to delve into the underlying principles of Kaito, understanding how it leverages the vast metadata provided by the Twitter API, conducts semantic analysis and trend prediction through large language models like OpenAI's ChatGPT, and builds a decentralized information ecosystem through mechanisms such as Smart Followers and Yap points for "social incentives." Next, Jesse will analyze this issue from both an industry significance and technical detail perspective.

Information is Capital: Kaito's Platform Innovation and Industry Relevance

The InfoFi new model advocated by Kaito is not only a technical and product innovation experiment but is also structurally impacting the information dissemination mechanism and marketing paradigm in the crypto industry. In the past, the marketing of crypto projects mainly relied on traditional means: hiring PR agencies, collaborating with KOLs (Key Opinion Leaders in the crypto community) to create buzz on social media. Under this model, information was often opaque, dissemination efficiency was low, and it gave rise to a large amount of "advertorials" and hype posts. In contrast, Kaito's algorithm-driven community incentives are changing the rules of the game—the relationship between projects, KOLs, and regular users is being repositioned in a competitive environment based on content value and contribution.

Project Marketing Paradigm Shifting from "Placement" to "Engagement"

In the traditional model, projects often viewed users' attention as ad space that could be purchased with funds: contributing to have a celebrity post promotional content, and then leveraging the celebrity's large fan base to spread the information. However, this placement-style marketing has significant pitfalls:

· Difficult to Measure Effectiveness: How many of the KOL's fans are genuinely interested in the project? What is the conversion rate? The project may spend a high budget but only get inflated "buzz," with minimal actual user conversion.

· Questionable Information Credibility: Audiences today can easily identify which content is sponsored, and they are usually wary or even resentful of such hard-sell advertising posts.

Kaito's emergence has led to an engagement-based viral communication paradigm: through "Yap-to-Earn," projects no longer need to concentrate their marketing budget on a few celebrities but can integrate into Kaito's Yapper ranking system, allowing community members to spontaneously voice support for the project. For example, a new project looking to increase visibility can collaborate with Kaito to open the project's community leaderboard on the platform—all users posting original content around this project participate in a point-based competition.

The actual effect is similar to a nationwide participatory content creation competition. Users compete to win Yap points or potential future airdrop rewards by researching the project, publishing in-depth analysis or unique insights, striving to climb the leaderboard to receive rewards. The project then, at a relatively low cost (such as promising to airdrop tokens or prizes to leaderboard's top users), harvests a vast amount of high-quality UGC (User-Generated Content). This content, actively shared by users on public platforms like Twitter, often has more virality and persuasiveness—after all, it's not cold advertising but the genuine voice of community members (even with incentive factors, the content is user-created). This model is known as the social version of "Proof-of-Attention": those at the top of the leaderboard for post creators are seen as providing high-value information and therefore receive deserved benefits.

Whether this approach is labeled as InfoFi or SocialFi, it fundamentally reshapes the way projects distribute their messaging. Marketing is no longer solely led by a centralized team but has shifted to incentive-driven community collaboration. The role of project teams has also evolved from traditional advertising sponsors to community activity initiators and reward providers.

No Longer a Hero Based Solely on Fans: How Small KOLs Succeeded through Kaito?

Within the InfoFi ecosystem, the role of traditional crypto KOLs has also transformed. On one hand, top-tier KOLs still hold significant influence: for example, industry giants like Vitalik and jesse.base continue to lead the Yapper rankings, demonstrating that truly insightful opinion leaders with a large following can still guide the conversation. On the other hand, these KOLs now operate in an openly competitive environment: each time they speak out, it is objectively recorded and scored by algorithms, making their points visible. For genuinely knowledgeable KOLs, this serves as positive reinforcement; however, those KOLs who relied on their fame without contributing valuable content may see their influence diminish under the InfoFi mechanism. If they only post advertisements without earning points and fail to actively engage in discussions, their ranking will drop, and the community will perceive them as "lacking substance." Consequently, KOLs are pressured to more actively and sincerely participate in community discussions; otherwise, they risk being surpassed by newcomers.

jesse has observed that some mid-tier KOLs have successfully staged a "comeback" through Kaito. While they may have fewer followers than A-listers, their diligent production of high-quality content has placed them prominently in the rankings, gaining exposure comparable to A-listers. This reshapes the traditional KOL influence landscape: influence is no longer merely determined by follower count, as content value and reputation also hold weight. This can be likened to "influence mining"—KOLs mine influence points (Yap) by consistently contributing high-quality information. In contrast to the past reliance on accumulating followers, in this model, influence acquisition is more multifaceted and dynamic.

Simultaneously, KOLs' monetization models are also evolving. Previously, A-listers mainly profited from project-sponsored promotions, whereas now they have an additional channel: accumulating Yap points for future redemption (such as converting them into the platform's token, KAITO). In the short term, Yap points cannot be directly monetized, but they are assigned a significant expected value (there is already a secondary market trading these expectations at a discounted valuation). As Yap is scarce and challenging to acquire, many KOLs invest time staying active on Kaito, akin to early participation in "mining" to secure future returns.

When some projects (such as Berachain) target a Top Yapper on Kaito for a specific airdrop reward, KOLs are more motivated to maintain their leaderboard position to receive these additional benefits. This inadvertently reduces the project's direct need to pay KOLs for advertising: instead of spending money to have a Key Opinion Leader (KOL) post an ad, it is more effective to allocate a portion of the budget as a community reward to incentivize engagement on Kaito; KOLs also benefit from this approach. As a result, the relationship between KOLs and projects shifts from the traditional client-supplier dynamic to that of partners participating in community activities. KOLs must demonstrate their genuine insights into the project to earn community approval, and the project team welcomes KOLs who drive more discussions about their project. Both parties interact on a public platform, promoting increased transparency and visibility of information.

Opportunities and Challenges for KOL Agencies

For KOL Agencies (Key Opinion Leader agencies), the Kaito model can be seen as a double-edged sword. On one hand, it undermines some of the exclusive value previously held by KOL Agencies: project teams can directly utilize Kaito's data and rankings to identify truly effective influencers without relying heavily on the agency's network resources. Kaito provides a quantified KOL landscape and performance rankings as a reference, enabling project teams to identify the most active communicators in niche areas, as well as users who demonstrate high engagement and loyalty to the project. Such data transparency was previously only held by seasoned KOL Agencies (based on long-term experience knowing which KOLs excel in conversion); now Kaito has made these metrics public and data-driven. An accurate KOL landscape can enhance marketing effectiveness, increasing value for project teams—relying on the cleaning and weighting of a massive amount of data is one of Kaito's core competitive advantages. If KOL Agencies continue with their old-fashioned approach, only providing vague KOL lists and broad delivery strategies, their value will inevitably be questioned.

However, on the other hand, KOL Agencies are not rendered entirely obsolete. Insightful agencies can choose to embrace Kaito, seeing it as a new tool to leverage. They can subscribe to advanced services like Kaito Pro to gain in-depth data insights, enabling them to develop more effective communication strategies for clients. Through the Kaito platform, KOL Agencies can precisely help project teams achieve communication goals, such as:

· KOL Screening: Using metrics like Yapper ranking, Smart Followers count (core followers), and other indicators to select the most suitable KOLs for collaboration on a project.

· Topic Planning: Utilize Kaito's analysis of industry trends to plan hot topics for the project to engage in community discussions, guiding more users to participate in the conversation.

· Effect Monitoring: Real-time monitoring of promotional effects, measuring volume conversion through Yap point growth and leaderboard changes, and adjusting strategies as needed.

· Rule Optimization: Guide the project team to make good use of Kaito's rule bonuses, such as how to initiate Launchpad community voting (an event where the community votes for projects to be listed), when to incentivize the community to produce more relevant content, etc. This role is somewhat similar to an SEO consultant from the search engine era—now emerging as an InfoFi consultant, specializing in how to leverage the Kaito ecosystem.

In this process, the value proposition of KOL Agency will shift from being a "resource intermediary" to a "strategy consultant," requiring a deep understanding of Kaito's algorithmic mechanisms and community operation tactics. It can be foreseen that some keen-sensed Agency institutions have already begun studying Kaito's point calculation method, seeking the key to triggering high scores to better serve their clients. Of course, it is important to note that Kaito's algorithm is continuously being updated and optimized, making it not easy to take advantage of simple tricks to score points through speculation. However, there is still ample room for optimization within compliance (e.g., guiding real community discussions rather than spamming or falsifying content). Overall, Kaito poses a challenge to KOL Agencies but also provides new opportunities for leveraging InfoFi tools; those who can master and effectively utilize these tools will continue to create value for their clients in the new paradigm.

Enhancing Information Dissemination Quality and Algorithmic Challenges

Kaito's enhancement of industry-wide information dissemination quality is widely recognized. Through the InfoFi incentive mechanism, the past inundation of pure advertisements and pump-and-dump posts on social platforms has been curbed, replaced by more in-depth analysis and rational discussion. This undoubtedly has had a positive impact on the entire crypto community's information environment: investors can see more insightful viewpoints, reducing the risk of being misled by meaningless noise; project teams can also receive more authentic feedback and suggestions from the community, rather than just compliments or insults. Attention is directed towards genuinely valuable information, significantly improving the effectiveness and quality of the information flow.

However, all of this also harbors a worrisome concern—the issue of algorithm-driven narrative concentration. As more and more industry communications shift to platforms like Kaito, the platform's algorithm itself wields tremendous influence. Just as people used to worry about Google's search algorithm determining which websites could be seen, nowadays Kaito's algorithm is effectively deciding which voices will be amplified. While InfoFi claims to be fair, the aforementioned analysis also indicates that it tends to favor users with existing reputations. This may lead to innovative ideas or contrarian views struggling to gain traction if they do not receive endorsement from mainstream influencers. Over time, could this create another form of "information echo chamber"?

The possibility of Kaito platform adjusting its algorithm for commercial interests is also worth noting— for example, the algorithm may tend to favor the promotion of partnered project information (as observed, for projects onboarded to Kaito, the system seems to significantly encourage users to engage in more discussions). As advocates of decentralization in the crypto community, we should remain vigilant against algorithmic monopolies, urging Kaito to maintain transparency and fairness in rule-making. Kaito has currently disclosed some FAQs and basic principles, but the specific scoring details remain a black box. In the future, a more DAO-oriented governance may be needed to allow the community to participate in supervising the algorithm's evolution, ensuring that the InfoFi model genuinely incentivizes high-quality information fairly.

Technical Principles: The Behind-the-Scenes Mechanism from Data Acquisition to AI Parsing

Twitter API Data Acquisition: Foundation and Challenges of Content

As a platform focusing on crypto information, Kaito first needs to continuously fetch data from Twitter. Through the official API interface, Kaito automatically retrieves the text, posting time, likes, retweets, and other metadata of each tweet, associates them with author information and a list of interacting users, laying the foundation for the subsequent algorithmic judgments.

For example, for a tweet discussing Bitcoin, Kaito will record its content, posting time, interaction engagement, and the influencer status of the poster; if an industry influencer participates in the interaction, the algorithm will determine that this information is more valuable. The premise to achieve all this is efficient scheduling and utilization of the Twitter API.

Since Elon Musk's appointment, Twitter has significantly raised API usage fees: the starting price for the enterprise-tier API is as high as $42,000 per month (only allowing queries for approximately 50 million tweets). To track the dynamics of the entire crypto community, the required call volume far exceeds this level, imposing significant cost pressures on startup projects. Although Kaito has not specifically outlined detailed response measures, one can imagine the team must meticulously calculate the cost of each API call. It is likely that they have adopted the following strategies to control data acquisition costs:

· Focusing on Key Areas: Prioritize fetching accounts and topics in specific crypto areas rather than indiscriminately crawling platform-wide data to save on call quotas.

· Batch Queries and Caching: Utilize batch queries, caching, and other technical means to reduce redundant requests and minimize API call counts as much as possible.

· User-Authorized Crowdsourcing: Some analysis speculates that Kaito requires users to link their X accounts to obtain authorization tokens, delegating part of the data fetching task to users themselves through "crowdsourcing," thereby bypassing official frequency restrictions.

In Jesse's view, these strategies are all aimed at minimizing data costs and risks as much as possible without affecting core functionality, to ensure that the InfoFi model has a stable data source.

ChatGPT Content Parsing: AI Empowering Information Value

Acquiring massive amounts of data is just the beginning; Kaito's more critical weapon is to leverage OpenAI's ChatGPT model for content semantic parsing and quality assessment. Simply put, Kaito lets AI act as an "appraiser" and "filter" of information. Whenever a user posts on X, the backend algorithm intelligently analyzes the content, including identifying the topic of the tweet, evaluating its value, and detecting any artificial inflation or cheating behavior.

With the help of advanced large-scale language models, Kaito claims to overcome language barriers and uniformly understand and rate multilingual content such as English and Chinese, without favoritism. This means that regardless of the language users use to express their views, they should theoretically receive the deserved Yap point rewards.

The ChatGPT model is also used to identify spam and noise content. According to Kaito's official statements and community introductions, they place great importance on the originality and depth of content and do not reward high scores based solely on surface-level interaction data, nor do they reward pure spamming or meaningless interactions. For example, even mechanically spamming keywords like "cryptocurrency" or "Crypto" in a post cannot trick the AI into boosting points because the system prioritizes genuine and meaningful discussions.

Jesse's firsthand experimentation has called into question the above ideal state. In the experiment, I posted three posts with edgy images and brief comments, unexpectedly earning nearly 190 Yap points. The comment sections of these three posts were filled with bland compliments, with almost no substantive information.

The ability to earn such high points for content with such high water content inevitably raises doubts: based on cost considerations, Kaito's algorithm may not actually perform in-depth semantic analysis on every post, or may adopt some form of simplification strategy in the scoring process. Perhaps the current system still relies more on basic interaction data to determine points, making trade-offs in semantic understanding. This finding has led Jesse to question the rigor of Kaito's algorithm: To what extent has the so-called intelligent content rating mechanism truly materialized?

Smart Followers Mechanism: Evaluating Influence Based on Quality Over Quantity

While Kaito introduces AI analysis at the content level, it has not overlooked the "social network" factor. The platform's innovation lies in the introduction of the "Smart Followers" mechanism, establishing a social graph of the crypto community, incorporating follower quality into content value assessment. For Kaito, who follows you is more important than just how many followers you have. Those well-known individual accounts that mutual follow each other and form the core circle of the crypto community are classified by the algorithm as Smart Followers (core followers).

If an author's fan list is filled with top-tier names (such as Vitalik Buterin, Binance CZ, etc., all following them), then the author's influence is obviously outstanding, and the maximum points they receive for their content will correspondingly be higher.

This social graph model allows Kaito to more objectively measure the "in-group diffusion" of each tweet: whether it spreads among outsiders or reaches the vision of industry top figures. For example, a message, even with 100 retweets, may have limited actual value if most of them come from mutual-following small accounts for self-entertainment. On the other hand, another message with only 10 retweets but involving heavyweights like Vitalik clearly holds more significance. For these two scenarios, Kaito will assign significantly different Yap points, avoiding a simple comparison based on the number of retweets or likes.

From actual results, it can be seen that the accounts at the top of the Yap ranking are often not the most popular fan-based influencers but are more likely deep players recognized by top KOLs. As a research report states, Kaito does not blindly trust traditional metrics like fan count or view count but rather focuses the reward on the reputation weight of "smart fans"—even if you have hundreds of thousands of fans, if your content lacks true value, the Yap you receive may still be meager. This "quality over quantity" evaluation method to some extent corrects the drawbacks of pure traffic competition, injecting a hint of academic peer review taste into InfoFi's information distribution: only content endorsed by experts can stand out.

Of course, the specific algorithm details of the Smart Followers system are kept confidential by the official sources, and we can only speculate on its general logic from the results. The Kaito team is concerned that if the rules are completely transparent, some may find ways to cheat the system, disrupting ecosystem fairness. Currently, the introduction of the social graph has indeed enhanced the algorithm's resistance to cheating but also presented new challenges to newcomers: how to win the attention and interaction of in-group celebrities has become the key threshold for obtaining high scores. On the one hand, this is a positive incentive for content creators, but on the other hand, there is a subtle concern about whether it will evolve into a game dominated by a few big players — after all, no matter how intelligent an algorithm is, what ultimately assigns value is the interpersonal network.

Trade-offs Between Technical Costs and Multi-layer AI Architecture

Having introduced so many features empowered by "black technology," it is also necessary to calmly evaluate the real-world cost ledger — supporting this complex system of Kaito incurs significant technical expenses. First is the data acquisition cost. As mentioned earlier, obtaining Twitter data through legitimate channels in large quantities comes at a high price, often running into tens of thousands of dollars per month. According to industry sources, Kaito previously attempted to acquire data through third-party channels or non-public APIs, but as Twitter tightened its policies, these gray methods became unsustainable, forcing them to honestly pay for higher-level API permissions. This directly forced Kaito to make trade-offs in its product strategy: if it blindly opens up a large number of queries to regular users, the monthly API call limit would quickly hit the ceiling.

Kaito has recently provided limited free query services to ordinary users, preferring to sell its deep data analysis capabilities to institutions and professional clients. For example, some hedge funds subscribe to the Kaito Pro professional version for a monthly fee of over $800. By serving a small number of paying "whales" to cover the high data bill, this also explains Kaito's current choice to focus mainly on a B2B (business-to-business) commercial route.

Another significant expense is AI computing power. Kaito officially claims to use AI at the level of GPT-4 to understand content, but behind each call to the ChatGPT-4 interface is a "burn" of money. If ChatGPT-4 is truly called in real-time for every tweet, the cost would be astronomical. A rough estimate: even using the cheaper ChatGPT-3.5, processing every 50,000 tweets' content could cost over a thousand dollars; if the costlier GPT-4 model is used for a full analysis, monthly expenses could even reach tens of thousands of dollars.

Obviously, Kaito will not be so reckless. It is speculated that the team may have devised a set of "AI Labor Rationalization" strategies: using large models only when necessary, applying rule-based filtering or small model predictions to unimportant areas, and minimizing ChatGPT calls. There are also indications that Kaito is developing its own large models or multi-agent systems, attempting to have some fine-tuned open-source models handle basic semantic scoring tasks. In this way, only when faced with complex problems or needing to generate long summaries would the expensive GPT-4 be called, significantly reducing the average call cost.

Kaito's founder, Yu Hu, revealed that they currently adopt an AutoGPT heterogeneous agent architecture, deploying multiple ChatGPT models to work together in the backend, with ChatGPT-4 as the core underlying model, while reducing reliance on third parties through fine-tuning proprietary models. This multi-layer model architecture reflects Kaito's delicate balance between effectiveness and cost: ensuring algorithmic analysis is sufficiently excellent and reliable while carefully managing expenses. This "dilemma at both ends" balance is precisely the operational challenge that the current InfoFi business model cannot avoid. It can be said that Kaito is staging a "technological gamble"—burning money to build a technological moat on one hand while hoping to find a more economically viable alternative in the future on the other.

Conclusion: Reflection and Future of the InfoFi Model

Kaito's platform design is a bold integration of cutting-edge technology and business models: it quantifies social content into "attention assets" and then uses tokens to incentivize high-quality information production. It sounds great, but implementation is not a smooth journey. Kaito's so-called "InfoFi," to some extent, is more like a disguised SocialFi—whether called Yap points or by another name, the essence is to monetize traffic and influence through social networks. In this regard, it shares similarities with early SocialFi projects like Friend.tech and Stars Arena.

What sets Kaito apart is the addition of an AI filtering layer and reputation weighting, aiming to raise the game's "quality threshold" and prevent pure astroturfing traffic. However, from the current results, this system still struggles to escape the Matthew effect: prominent figures dominate the rankings, high scores align closely with top-tier influence, and small accounts looking to stand out rely on endorsement from larger accounts. Is this truly breaking information monopoly, or is it effectively reinforcing existing cliques? This will be one of the core issues that Kaito will need to squarely face in the future.

A more realistic challenge lies in the sustainability of the model. Kaito's current state is heavily reliant on the Twitter ecosystem—both data sources and user interactions are almost entirely tied to Platform X. How far can this development model, which is living under someone else's roof, go? If Twitter were to raise API prices again, tighten data permissions, could Kaito still manage? The current high API costs have already forced Kaito to turn to serving paying customers to support operations. However, if the InfoFi model is to expand to full participation, eventually, a solution must be found to balance this account.

On the other hand, there is also uncertainty surrounding the token economy that underpins Yap incentives. Currently, the value of Yap points is mostly on a speculative level. Once market enthusiasm wanes and expected value decreases, will the platform's top KOLs shift their focus elsewhere, leading Kaito to face the risk of content loss? KOLs who navigate various platforms often flock to where the returns are high. If Kaito cannot consistently provide sufficient returns in terms of income or influence, relying solely on sentiment will not retain these top users.

Overall, for the InfoFi model to succeed, it will ultimately need to strike a better balance between incentivizing deep content creation and maintaining its own self-sustaining ability. Can Kaito find a sustainable development path amidst intense competition and resource constraints? We wait with bated breath.

Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Official Twitter Account: https://twitter.com/BlockBeatsAsia