J. Roinevirta jj@basedapp.com	|	J. Lehtonen jimi@basedapp.com

​

Introduction

Public opinion is still measured with instruments from the broadcast era: polls, surveys, and panels. They are slow, expensive, and biased. In a networked environment where attitudes shift rapidly and expression is performative, these tools fail to capture the dynamics that matter. A decade of social-media sentiment analysis was supposed to fix this. Platforms and funds treated the social feed as a continuous panel and trained NLP models to infer sentiment from text and engagement. The approach fails for a structural reason: the feed is not a measuring instrument. As synthetic content now outpaces human output, its observables are contaminated, turning the task from estimation into adversarial inference that scale cannot fix. This is both an epistemic problem and an economic one. Capital allocation, brand strategy, policy, and risk management depend on reading collective perception. A useful response has emerged under the banner of information finance: use incentive design to extract knowledge from distributed human judgment. Prediction and betting markets are successful examples. They encode what people expect to happen when an event has a clear resolution condition. Based operationalizes this logic for domains where resolution does not occur. It extends information-finance mechanisms from forecasting to collective judgment. Most questions of interest do not have such a condition. They concern alignment, approval, and values rather than outcomes. Here the relevant distinction is simple:

• Prediction markets price the is: what participants think will happen.
• Based prices the ought: what participants think should prevail.

Normative judgements shape the environment in which decisions are made: which side of a geopolitical conflict currently holds public sympathy; the level and trajectory of consumer confidence; which brands or figures face boycott pressure, and how strong that pressure is. These judgments do not resolve at a fixed time, yet institutions need a continuous, tamper-resistant read on them. Based introduces a market structure for these judgments. Participants trade according to their model of the public mind. They profit when their view aligns with the consensus revealed by the market. Each position is an economically weighted statement about what the participant believes others believe. The result is a continuous index of collective belief. Where prediction markets forecast resolvable events, Based measures consensus on questions without resolution. In information-finance terms, it extends market mechanisms from probability to perception and supplies an instrument suitable for real-time decision making in domains where surveys and NLP no longer suffice. This paper formalizes that instrument and outlines the system architecture that implements it. Further explorations into the financial-economic system designs, data classification models, and statistical properties of the resulting signal will be discussed in the white paper.

​

Based’s Contribution to InfoFi: Hermeneutic Quantization

Based, like any other InfoFi concept, uses a distributed, economically incentivized network of human users to operate as “Data Providers”. Based leverages the collective interpretation of humans to value any subjective topic through a non-resolving, bidirectional, derivative-like smart contract. It incentivizes users to express their best estimate of the current or future social sentiment regarding any subject of public discourse, rewarding alignment with consensus and penalizing deviation. By utilizing machine-led classification, clustering, and aggregation of the content, we combine the economic data to produce insights about topics in a process labelled “Hermeneutic Quantization” (HQ):

• Hermeneutics: The process of understanding and interpretation, performed by the distributed human network (Data Providers).
• Quantization: The process of combining, weighting, and mapping this human-interpreted data into a structured value or data stream, performed by machines (AI/ML).

​

System Overview

The Based System is built to extract, interpret, and deliver quantifiable data on subjective topics. The system’s core is the Based Protocol, which manages the on-chain markets. Data flows into the system via two primary interfaces:

• Based: A gamified tap-to-trade social media app where users interact with content and, as a by-product, act as Data Providers.
• Based Pro: A prosumer & professional trading-focused platform offering access to aggregated Sentiment Indices.

The underlying markets on both interfaces are managed by the Based Protocol, which implements a novel economic structure designed to continuously elicit information about subjective (and therefore evolving) topics. The non-economic data (images, videos, comments, descriptions) is indexed and classified by the off-chain Hermeneutic Quantization (HQ) module. This module combines the machine classification with the financial signals from the Based Protocol to produce a strong, sybil-resistant, low-noise signal that is served to Data Consumers via the Hermes platform. Figure 1: Overview of the Based System for Hermeneutic Quantization. These system components are described briefly below.

​

Data Providers

The system elicits data through two distinct avenues:

1. Based: This is the granular, low-level market structure. Every post is an individual Market. Users’ interactions, which are similar to a traditional social media feed, generate economic signals (trades) and off-chain content signals (comments, media). The economic design incentivizes users to express their view based on what they believe the aggregate view of others will be.
2. Based Pro: This platform is designed for power users. The Index Manager aggregates multiple low-level Markets into broader Sentiment Indices (or “baskets”). These indices are more liquid and allow traders to take positions on higher-level Research Topics without needing to engage with the granular data.

In short, Based collects precise but weaker signals from a large number of diverse users, while Based Pro collects stronger but inexact signals from a smaller number of market participants. Together, they form the Data Providers.

​

Onchain: Based Protocol & Index Manager

Every post made to Based automatically becomes a Market, created and managed by the Based Protocol. A Market’s trading data is a sybil-resistant signal regarding the post’s content. The Protocol rewards users who correctly anticipate the collective view of other users in that Market, not unlike how prediction markets reward accurate forecasting of outcomes. The Index Manager, guided by machine classification, combines multiple Markets into a Sentiment Index. This index is designed to track a specific Research Topic and is highly liquid vis-a-vis individual markets. The price of an individual Market is influenced by both its own trading activity and the performance of any Sentiment Indices it belongs to. The Index Price, in turn, is based on the net exposure of its constituents, weighted by the HQ module. This interplay creates a dynamic and elastic relationship between the granular Markets and the aggregated Indices, ensuring more accurate and nuanced insights. Figure 2: the relationship between Market and Sentiment Index pricing is both dynamic and elastic (simplified)

​

Offchain: Hermeneutic Quantization

HQ is the engine that merges disparate data streams into a single, high-quality metric. It combines:

• Off-chain data: Comments, media, descriptions, media interpretations, etc.
• On-chain economic data: Price, volume, open interest, which represent the validated human interpretation.

​

The Core Inversion: Humans Interpret, Machines Classify

HQ’s design is driven by our core principle: we use AI where it excels and avoid it where it falls short of human capability.

1. AI/ML for Classification (Where Machines Excel): HQ uses NLP and Convolutional Neural Networks (CNN) solely to classify the content (e.g., to label a post as being about “Tesla” and “Battery Technology”). This is an objective structural task for which AI is excellent.
2. Humans for Interpretation (Where Machines Fail): HQ does not assume AI/ML methods can correctly infer sentiment. Instead, it leaves the subjective act of interpretation to the distributed human network of Data Providers. The economically-backed trading data – financial positions with real value at risk – is used as the validated, low-noise sentiment measure.

This is the fundamental difference from existing systems, which explicitly delegate the critical, subjective task of sentiment inference to an algorithm. By securing the data with financial positions, we arrive at a minimally gameable, low-noise data set that correctly captures the nuances of the public mind.

​

Data Consumers

Based’s data is intended for anyone needing objective, quantifiable information on subjective topics. Our initial focus is on financial use cases via the Hermes platform, with future expansion into broader business analytics and enterprise solutions. Hermes provides visualizations and access to data structured around Research Topics – well-defined subjects with clear boundaries. HQ supplies Hermes with the forward-looking, consensus-based data to answer questions such as:

• “What is the overall public perception of company X’s new product line?”
• “Was the recent regulatory change Y received positively by the market segment Z?”
• “Which sports equipment category within Z’s manufacturing base is most likely to see the strongest consumer uptake next year?”

Critically, Hermes does not provide backward-looking or even real-time data. It provides forward-looking, consensus-based data, as inferred and expressed by the system’s economically incentivized Data Providers.

​

Roadmap

Based is built on a modular design, allowing us to deploy and iterate on our core components systematically. Our roadmap focuses on expanding utility and decentralization:

• Phase 1/ Closed Beta: Testing of the core systems and Based social media application for intuitiveness and usability. This phase validates the functionality of the consumer facing application.
• Phase 2/ Open Beta: Full deployment of the Based Protocol on a mainnet, bootstrapping the initial network of Data Providers via the Based social media application. This phase validates the core economic mechanisms and sybil-resistance.
• Phase 3/ Full Data Provider Launch: Full deployment of the Index Manager on mainnet, bootstrapping the initial network of prosumer Data Providers. This phase validates the economic design and functionality of the Based Pro prosumer-focused index platform.
• Phase 4/ Hermes Platform Launch: Introduction of the Hermes data platform, offering structured data access to our first tranche of Data Consumers. This confirms the commercial viability of the HQ signal.
• Phase 5/ Decentralization & Open Data Access: Transitioning key governance aspects of the Protocol to the community. Expanding the HQ module to integrate with third-party data consumers via public APIs and licensing models, enabling business analytics and enterprise use cases beyond finance.

​

Conclusion: Quantifying the Public Mind

Based re-architects how collective belief is measured. We replace the flawed assumption of machine interpretation with a distributed, economically-incentivized network of human judgment. By putting value at risk, Hermeneutic Quantization (HQ) delivers the clearest, most robust signal on subjective reality ever created. We are building the definitive oracle for beliefs, perceptions, and consensus – a necessary infrastructure layer for any market or enterprise dependent on understanding the true public mind.

​

Team

Based is developed by a founding team with backgrounds in investment banking, derivatives, and on-chain data infrastructure. Jimi Lehtonen brings expertise in market structure and the delivery of financial signals from his work in investment banking.
Juuso Roinevirta contributes to smart-contract architecture and incentive design, drawing on his experience in derivatives and financial engineering.
Heikki Vänttinen focuses on data transformation and go-to-market strategy, informed by his previous work in oracle networks. The founders are supported by engineers, product designers, and marketers with experience across decentralized finance, data analytics, and consumer-application development.

​

Engage with Based

If you are an engineer that would like to work on solving hard data problems, please see our careers -section. If you find Based’s data useful in your work, please get in touch. And if you’d like to contribute to one of the most ambitious restructurings in data collection by using the gamified Based app, visit basedapp.com.