Skip to main content

Any Tracker v2: Harm Reduction Through Intelligent Substance Tracking

· 14 min read
Max Kaido
Max Kaido
Architect

Any Tracker represents a pragmatic approach to substance use in the real world - acknowledging that people make their own choices and providing tools to make those choices safer and more informed.

Core Philosophy: Pragmatic Harm Reduction​

Any Tracker is built on three fundamental principles:

  1. Reality Acceptance: We acknowledge that substance use exists across the spectrum from supplements to controlled substances. Rather than moralizing, we focus on making that reality safer.

  2. User Autonomy: We respect users as responsible individuals with their own preferences and risk tolerances - similar to how financial systems respect different investor profiles.

  3. Optimization Within Preferences: The system aims to optimize safety and effectiveness within user-defined parameters rather than imposing external values.

Technical Architecture: Privacy by Design​

The system architecture is specifically designed to eliminate data responsibility risks:

  1. Open-Source Client: A transparent Telegram Mini App built with TypeScript/Node.js that users can verify and trust.

  2. Anonymized Engine: A separate Python/TypeScript backend with an open API that processes only anonymized data, eliminating deanonymization risks for both markets and users.

  3. Self-Hosted AI: Leveraging state-of-the-art open models via Ollama for local inference, ensuring data never leaves the user's trusted environment.

  4. Graph-Based Analysis: Using Neo4j to analyze consumption patterns and identify optimal schedules without compromising privacy.

Value Proposition: Beyond Simple Tracking​

Any Tracker is not just a logging tool but an intelligent assistant that:

  1. Minimizes User Input: Natural language processing allows users to simply type "took 500mg taurine" rather than navigating complex interfaces.

  2. Maximizes Execution Probability: By analyzing user behavior patterns, the system suggests the most likely next steps at optimal times.

  3. Provides Intelligent Warnings: The system analyzes substance interactions and warns about tolerance buildup or potentially harmful combinations.

  4. Protects From Scams: By integrating with legitimate marketplaces, users gain protection from the prevalent fake sites in the darknet ecosystem.

The Way of Change: Eastern Philosophy Meets Modern Psychology​

Any Tracker incorporates a profound philosophical approach to habit formation and change:

  1. The Quantum State of Habits: We recognize that all potential habits exist simultaneously in a state of possibility - like Schrödinger's cat, both present and not present until observed. When a user tracks something even once, we acknowledge this as evidence of a habit already in existence, merely at a frequency approaching zero.

  2. The Middle Way of Behavioral Change: Drawing from Buddhist philosophy, we avoid the extremes of forcing change or accepting stagnation. Instead, we craft a middle path that respects the user's current state while gently guiding them toward their desired outcome.

  3. The Path of Least Resistance: Rather than triggering the brain's natural resistance to change (what psychologists call "reactance"), we design interventions so subtle they fall below the threshold of conscious effort. This aligns with Lao Tzu's concept of "wei wu wei" (doing without doing) - accomplishing goals without forced action.

  4. Kairos Over Chronos: Ancient Greeks distinguished between chronos (sequential time) and kairos (the right moment). Any Tracker embodies this distinction by identifying the perfect moment for intervention - when the user is most receptive to change - rather than relying on rigid schedules.

  5. The Hermes Principle: Named for the messenger god who moved between realms, we serve as the intermediary between the user's current and desired states, guiding transitions without judgment or force. Like Hermes, we facilitate passages between different states of being.

This approach represents a synthesis of Eastern philosophical traditions (particularly Taoism and Buddhism), modern behavioral psychology (especially B.J. Fogg's Tiny Habits methodology and Richard Thaler's Nudge Theory), and ancient Greek concepts of time and transformation. By acknowledging that change is a path rather than a destination, we create a system that works with human nature rather than against it.

User Experience: Respecting Individual Preferences​

The system treats each user as having unique preferences, represented as a cost function used in all calculations:

  1. Personalized Risk Profiles: Like traders have risk/reward preferences, substance users have their own tolerance for various effects and side effects.

  2. Preference Learning: The system gradually learns user preferences through minimal interactions rather than extensive questionnaires.

  3. Non-Intrusive Suggestions: Recommendations appear at optimal moments based on learned patterns rather than arbitrary schedules.

Business Model: Sustainable Development​

The project's sustainability relies on multiple revenue streams:

  1. Self-Hosted Solutions: Marketplaces can pay for custom engine setups that maintain their security requirements.

  2. SaaS API Access: Premium features available through API access with advanced AI recommendations.

  3. Marketplace Integration: Darknet markets can use Any Tracker as a tool for user retention and fraud prevention.

  4. Anonymized Analytics: Aggregated insights into substance usage trends without compromising individual privacy.

Implementation Roadmap​

  1. Prototype Development: Create a basic Telegram Mini App with core tracking functionality.

  2. AI Pattern Analysis: Test machine learning models on anonymized datasets to identify optimal usage patterns.

  3. Architecture Separation: Fully separate the client and engine components with clear API boundaries.

  4. Marketplace Partnerships: Identify initial partners for testing and refine requirements based on their feedback.

  5. Premium Features: Define and implement value-added capabilities that justify subscription or one-time payments.

Conclusion: A New Paradigm for Substance Tracking​

Any Tracker represents a paradigm shift in how we approach substance use tracking:

  • Empowerment Over Judgment: Providing users with tools to make their own choices safer
  • Privacy as Foundation: Creating a technical architecture that inherently protects anonymity
  • Intelligence Beyond Logging: Using AI to predict optimal timing and combinations
  • Harm Reduction in Practice: Acknowledging reality and working to improve outcomes within it

By focusing on optimizing within user preferences rather than dictating behavior, Any Tracker creates a system that's more likely to be used and therefore more likely to have a positive impact on user well-being.

Questions for Further Exploration​

If you're analyzing this concept, consider exploring these questions:

1. Philosophical and Ethical Dimensions​

How might the "Quantum State of Habits" approach be extended beyond substance tracking to other domains of behavior change? What philosophical traditions beyond Buddhism and Taoism might offer complementary frameworks for understanding the nature of habit formation and the ethics of subtle influence?

Consider the ethical implications of a system that operates in legally gray areas while simultaneously reducing harm. How does this approach challenge conventional binary thinking about ethical systems? What historical precedents exist for technologies that operated in similar ethical territory but ultimately proved beneficial?

2. Technical Implementation Challenges​

The system proposes using reinforcement learning to optimize user schedules based on individual preferences. What specific RL algorithms would be most appropriate for this use case, considering the sparse reward signals and the need to balance exploration (trying new approaches) with exploitation (using known effective approaches)?

How might the graph database structure be optimized to capture not just substance interactions but also contextual factors (time of day, preceding activities, social environment) that influence effectiveness and user experience? What query patterns would be most valuable for extracting actionable insights?

3. Business and Product Evolution​

Beyond darknet marketplaces, what other potential integration partners might benefit from this technology? Consider both conventional and unconventional applications that might not be immediately obvious.

How might this system evolve if it gained significant adoption? What network effects could emerge from aggregated anonymized data, and how might these create additional value propositions beyond the initial concept? What defensive moats could protect the business from potential competitors?

GPT-4.5 Response​

Great! Given your comprehensive context for Any Tracker v2, I'll dive directly into the provided exploration questions with insights tailored specifically to your project:

1. Philosophical and Ethical Dimensions​

Extending the "Quantum State of Habits" Beyond Substance Tracking​

  • The idea of habits existing in a quantum state—both present and absent until consciously observed—can be effectively applied to domains like productivity, health, and mental well-being.

    • Example:
      • Productivity: Tasks and routines could be considered latent habits. Each initial action (tracking even once) shifts these from potential into reality, activating motivational and cognitive triggers.
      • Mental Health: Subtle interventions around mental wellness—like gratitude journaling—could similarly use quantum habit recognition to reinforce beneficial practices without pressure.

  • Philosophical traditions beyond Buddhism and Taoism might enrich this framework:

    • Stoicism: Particularly its focus on daily reflective practices (Marcus Aurelius’s meditations), emphasizing acknowledgment without judgment.
    • Existentialism (Jean-Paul Sartre, Viktor Frankl): The concept of authenticity aligns well—users could track behaviors as expressions of self-actualization rather than imposed obligations.

Ethical Implications and Historical Parallels​

  • Navigating Ethical Gray Areas: Operating in legally ambiguous spaces challenges conventional binary ethics. Your system implicitly questions the moral absolutism of traditional harm-reduction debates, suggesting a spectrum of benefit vs. harm rather than a black-and-white dichotomy.

  • Historical Precedents: Technologies that operated ethically ambiguously yet ultimately proved beneficial:

    • Cryptography (PGP/GnuPG): Initially viewed as ethically and legally problematic, cryptography eventually became fundamental to privacy and security.
    • Needle exchange programs: Initially controversial, later recognized as essential public health interventions.

2. Technical Implementation Challenges​

Reinforcement Learning (RL) Algorithms for Sparse Reward Environments​

Given the sparse reward signals and the need for a balance between exploration and exploitation:

  • Deep Q-Networks (DQN): Suitable for handling discrete actions and states with user behaviors, enabling incremental improvements based on user feedback.

  • Proximal Policy Optimization (PPO): Effective for continuous and nuanced behavioral optimization, adjusting interventions subtly and responsively based on minimal user signals.

  • Contextual Bandit Algorithms: A simpler alternative suitable for learning from minimal interactions—excellent for personalized recommendations without extensive user input.

Optimizing Graph Databases (Neo4j) for Rich Context Capture​

  • Structure Optimization: Nodes can represent substances, activities, timestamps, environments, emotional states, and social contexts. Edges can represent relationships or sequences of events.

    • Key Relationships to Index:
      • Sequential patterns (time-based)
      • Contextual triggers (social or emotional environments)
      • Compound interactions (combinations and effects)

  • Query Patterns:

    • Queries analyzing frequent contextual pairings or sequences.
    • Graph traversals highlighting potential risks (e.g., tolerance increases after repeated usage).
    • Temporal analysis for identifying effective timing (kairos-based recommendations).

3. Business and Product Evolution​

Potential Integration Partners Beyond Darknet Marketplaces​

  • Healthcare and Wellness Platforms: Applications tracking prescription medication adherence, dietary supplements, or mental health interventions.

  • Corporate Productivity Tools: Leveraging habit formation tracking to boost employee productivity or well-being.

  • Legal but Sensitive Domains: Cannabis markets or other regulated substances where harm reduction and consumer safety are highly valued.

Potential Network Effects and Additional Value Propositions​

  • Aggregated Anonymized Insights: Could create comprehensive industry benchmarks, inform research studies, or drive policy advocacy.

  • Cross-Platform Recommendations: If users track across various aspects (health, productivity, substance use), cross-domain insights could provide uniquely personalized, holistic recommendations.

  • Community-Driven Content and Recommendations: User-generated insights and feedback loops could further enrich AI recommendations, building trust and engagement.

Defensive Moats for Long-Term Competitive Advantage​

  • Trust and Privacy by Design: Emphasizing transparency and user-owned data protection makes replicating trust challenging for competitors.

  • Data-Driven Personalization: Accumulated behavioral insights create a robust barrier—competitors would struggle to replicate nuanced personalization quickly.

  • Strong Marketplace Integrations: Established partnerships become critical dependencies for ecosystem actors, strengthening long-term defensibility.

ChatGPT Project custom instructions:

########

I'm developing Any Tracker, an AI-powered Telegram mini-app dedicated to intelligent substance tracking, pragmatic harm reduction, and user empowerment. The core philosophy respects user autonomy, acknowledges real-world substance use across the full spectrum without moralizing, and focuses on maximizing safety within user-defined risk preferences. Technically, it features an open-source, transparent client, a privacy-preserving anonymized backend, local inference with state-of-the-art open models via Ollama, and Neo4j-powered graph analysis to identify safe, optimized patterns. Philosophically, it integrates Eastern concepts (Taoism's "wei wu wei", Buddhism's Middle Way), quantum habit theory (habits exist in potential states until observed), and modern behavioral psychology (Tiny Habits, Nudge Theory), facilitating subtle, effective behavioral change. I'm currently exploring broad technical, philosophical, ethical, and business model possibilities.

#########

DeepThink (R1) Response

1. Philosophical and Ethical Dimensions​

Extension of "Quantum State of Habits": The concept could apply to behavioral domains like mental health (e.g., tracking sporadic meditation sessions as "proto-habits"), financial habits (e.g., irregular savings treated as latent fiscal responsibility), or environmental behaviors (e.g., occasional recycling as a habit in superposition). It reframes any single action as evidence of a habit’s potential, encouraging incremental growth.

Complementary Philosophical Traditions:

  • Stoicism: Focuses on aligning actions with controllable outcomes, resonating with the system’s emphasis on optimizing within user-defined parameters.
  • Existentialism: Highlights individual autonomy, reinforcing the "User Autonomy" principle by treating choices as self-defined commitments.
  • Ubuntu Philosophy: Emphasizes communal well-being, which could ethically ground harm reduction as a collective benefit.

Ethical Implications: The system challenges binary ethics (legal vs. illegal) by prioritizing pragmatic harm reduction, akin to historical precedents like needle exchanges or Portugal’s decriminalization model. These initiatives faced ethical scrutiny but demonstrated that non-judgmental approaches reduce societal harm. The tension lies in balancing legality with compassion—a "middle way" that acknowledges human complexity without condoning risky behavior.

2. Technical Implementation Challenges​

Reinforcement Learning (RL) Algorithms:

  • Inverse RL: Infers user preferences from sparse interactions, ideal for learning personalized cost functions without explicit feedback.
  • Contextual Bandits: Balances exploration (testing new substance schedules) and exploitation (using known safe patterns) by incorporating contextual data (time, mood).
  • Meta-Learning: Adapts quickly to individual users by leveraging anonymized aggregate data, addressing sparse rewards through shared patterns.

Graph Database Optimization:

  • Node Structure: Represent substances, users (anonymized), contexts (e.g., "social," "morning"), and effects as interconnected nodes.
  • Temporal Edges: Use relationships like TAKEN_AT(time: "10:00") or OCCURS_WITH(activity: "exercise") to capture context.
  • Query Patterns:
    • Path Analysis: Identify common sequences leading to adverse effects (e.g., Substance A → Substance B → Interaction).
    • Community Detection: Cluster users with similar risk profiles to predict optimal schedules.
    • Temporal Clustering: Find usage peaks correlated with specific times or activities.

3. Business and Product Evolution​

Unconventional Integration Partners:

  • Telehealth Platforms: Monitor medication adherence while flagging interactions with supplements or recreational substances.
  • Fitness Apps: Track supplement stacks and pre-workout combinations, offering synergy/tolerance insights.
  • Clinical Research: Anonymized data could reveal real-world substance use patterns, aiding studies on microdosing or polypharmacy.

Network Effects & Defensive Moats:

  • Data Network Effects: Aggregated trends enhance AI’s predictive power (e.g., identifying emerging counterfeit substances).
  • Trust as a Moat: Open-source clients and privacy-first design foster user loyalty, while darknet marketplace integrations create high switching costs.
  • Ecological Moats: Partnerships with Telegram (Mini Apps) and self-hosted AI (Ollama) embed the tool into existing ecosystems, deterring competitors reliant on centralized infrastructure.

Evolution Trajectory: At scale, the system could become a "Google Maps for behavior change"—not dictating destinations but optimizing routes based on personal thresholds. Anonymized analytics might uncover public health trends (e.g., regional opioid analogs), positioning Any Tracker as a bridge between individual autonomy and collective well-being.

Grok 3.5 response:

#######

Below are detailed responses to the three questions posed in the query, based on the context of "Any Tracker v2: Harm Reduction Through Intelligent Substance Tracking." Each response explores the philosophical, technical, and business dimensions of the system, drawing from its core principles and proposed architecture.

1. Philosophical and Ethical Dimensions​

Question: How might the "Quantum State of Habits" approach be extended beyond substance tracking to other domains of behavior change? What philosophical traditions beyond Buddhism and Taoism might offer complementary frameworks for understanding the nature of habit formation and the ethics of subtle influence? Consider the ethical implications of a system that operates in legally gray areas while simultaneously reducing harm. How does this approach challenge conventional binary thinking about ethical systems? What historical precedents exist for technologies that operated in similar ethical territory but ultimately proved beneficial?

Answer:

Extending the "Quantum State of Habits"​

The "Quantum State of Habits" posits that habits exist in a state of potentiality—simultaneously present and absent until observed or acted upon—much like quantum superposition. Beyond substance tracking, this concept can apply to various behavior change domains:

  • Diet: A user might have the potential to adopt a plant-based diet. Tracking a single meatless meal could be seen as evidence of this habit existing at a low frequency, which Any Tracker could nudge toward consistency by suggesting optimal meal times or recipes.
  • Exercise: The potential to become a runner exists in someone who occasionally jogs. The system could identify "kairos" moments (opportune times) to suggest runs, gradually increasing frequency without triggering resistance.
  • Mental Health: Practices like meditation or journaling could be in a nascent state. Tracking one session could prompt subtle recommendations to build a routine, respecting the user’s current state.

This approach aligns with Any Tracker’s philosophy of subtle influence and optimization within preferences, making it versatile for any habit formation scenario.

Complementary Philosophical Traditions​

Beyond Buddhism (the Middle Way) and Taoism (wei wu wei), other traditions can enrich this framework:

  • Stoicism: Emphasizes accepting what cannot be changed while focusing effort on what can. This complements Any Tracker’s reality acceptance and harm reduction focus—acknowledging substance use (or other behaviors) as a given and providing tools to manage it safely.
  • Existentialism: Highlights individual freedom and responsibility. This resonates with user autonomy, as Any Tracker respects users’ choices, offering information rather than mandates, akin to an existentialist’s emphasis on self-defined meaning.
  • Pragmatism: Focuses on practical outcomes over ideological purity. Any Tracker’s emphasis on optimizing safety and effectiveness within user preferences reflects a pragmatic approach to complex, real-world challenges.

These traditions enhance the ethics of subtle influence by grounding it in acceptance, autonomy, and practicality, ensuring interventions align with human nature.

Ethical Implications and Challenges to Binary Thinking​

Operating in legally gray areas—such as tracking controlled substances—raises ethical questions but also challenges binary notions of right and wrong. Any Tracker reduces harm (e.g., warning of dangerous interactions) while existing outside traditional legal frameworks. This suggests that ethical value can emerge from ambiguity, prioritizing outcomes (safety, well-being) over strict adherence to rules.

  • Challenge to Binary Ethics: Conventional ethics often frames actions as legal/illegal or moral/immoral. Any Tracker’s harm reduction approach shows that gray areas can yield positive results, shifting the focus from prohibition to pragmatic improvement.

Historical Precedents​

Technologies operating in ethically ambiguous spaces have proven beneficial:

  • Needle Exchange Programs: Initially controversial and legally murky, these programs reduced HIV transmission among drug users by providing clean needles, demonstrating harm reduction’s efficacy.
  • Safe Injection Sites: Operating in some regions despite legal challenges, these facilities lower overdose deaths and connect users to treatment, showing that pragmatic interventions can outweigh strict legal conformity. These examples validate Any Tracker’s approach: working within reality’s constraints can improve outcomes, even if it defies traditional ethical or legal norms.

2. Technical Implementation Challenges​

Question: The system proposes using reinforcement learning to optimize user schedules based on individual preferences. What specific RL algorithms would be most appropriate for this use case, considering the sparse reward signals and the need to balance exploration (trying new approaches) with exploitation (using known effective approaches)? How might the graph database structure be optimized to capture not just substance interactions but also contextual factors (time of day, preceding activities, social environment) that influence effectiveness and user experience? What query patterns would be most valuable for extracting actionable insights?

Answer:

Suitable Reinforcement Learning Algorithms​

Reinforcement learning (RL) is ideal for optimizing user schedules based on preferences, especially with sparse reward signals (e.g., infrequent feedback like “I felt good” or “I didn’t like that”). Key algorithms include:

  • Multi-Armed Bandits (MAB): These balance exploration and exploitation efficiently in simple scenarios. Each “arm” could represent a schedule (e.g., “500mg taurine at 8 AM” vs. “300mg at noon”). MAB learns which schedules maximize user satisfaction with minimal trials, ideal for sparse rewards.
  • Contextual Bandits: An extension of MAB, these incorporate context (e.g., time of day, mood, prior activities) to refine suggestions. For example, suggesting caffeine in the morning but not late at night based on user feedback and context improves personalization.
  • Deep Q-Networks (DQN): For complex state spaces (e.g., multiple substances, contexts, and long-term effects), DQNs use neural networks to approximate optimal actions. They could model intricate user behavior patterns, though they require more data and computational resources.

Given Any Tracker’s focus on minimal input and privacy, contextual bandits strike a practical balance: they adapt to sparse rewards and contextual nuances without excessive complexity.

Optimizing the Graph Database Structure​

A graph database (e.g., Neo4j) can capture substance interactions and contextual factors:

  • Nodes:
    • Substances: e.g., taurine, caffeine, with properties like dosage and effects.
    • Users: Anonymized profiles with learned preferences (e.g., risk tolerance).
    • Contexts: e.g., time of day (morning), preceding activities (workout), social environment (alone).
  • Edges:
    • User-Substance: Tracks usage (e.g., “User A took 500mg taurine”).
    • Substance-Substance: Represents interactions (e.g., “caffeine + taurine increases alertness”).
    • Context-Substance: Links context to outcomes (e.g., “taurine at night reduces sleep quality”).
  • Properties: Edges and nodes store data like frequency, user feedback (e.g., “felt energized”), and timestamps.

This structure enables dynamic modeling of how contexts influence substance effectiveness, supporting Any Tracker’s “kairos over chronos” principle.

Valuable Query Patterns​

Actionable insights can be extracted via these queries:

  • Optimal Schedules: “Find the best times and dosages for User A’s substances based on past feedback and context.” (e.g., “Suggest taurine post-workout, not bedtime.”)
  • Harmful Interactions: “Identify combinations or contexts linked to negative feedback.” (e.g., “Flag caffeine + alcohol in the evening.”)
  • Personalized Recommendations: “Recommend new substances or schedules based on similar users’ patterns.” (e.g., “Users like A found 200mg magnesium relaxing at night.”)

These queries leverage the graph’s relational power to deliver precise, privacy-respecting insights.

3. Business and Product Evolution​

Question: Beyond darknet marketplaces, what other potential integration partners might benefit from this technology? Consider both conventional and unconventional applications that might not be immediately obvious. How might this system evolve if it gained significant adoption? What network effects could emerge from aggregated anonymized data, and how might these create additional value propositions beyond the initial concept? What defensive moats could protect the business from potential competitors?

Answer:

Potential Integration Partners​

Beyond darknet marketplaces, Any Tracker’s technology can benefit diverse partners:

  • Healthcare Providers: Manage complex medication schedules (e.g., for chronic illnesses) or track substance use in harm reduction programs, integrating warnings and optimal timing.
  • Wellness and Fitness Apps: Optimize supplement use (e.g., protein, creatine) for performance goals, aligning dosages with workouts or sleep cycles.
  • Research Institutions: Study anonymized substance use patterns to inform public health policies or clinical trials, leveraging aggregated data without privacy risks.
  • Mental Health Platforms: Track mood-enhancing supplements or medications, suggesting kairos moments for intake based on user context (e.g., stress levels).

Unconventional partners might include employers (tracking caffeine for productivity) or educational platforms (optimizing study aid supplements), broadening the system’s scope.

Evolution with Significant Adoption​

With widespread use, Any Tracker could evolve into:

  • A Predictive Ecosystem: Enhanced AI models predict user needs with greater accuracy, suggesting interventions before users request them.
  • A Community Hub: Anonymized insights foster peer knowledge-sharing (e.g., “Most users find X effective at Y time”), akin to crowd-sourced reviews.
  • A Marketplace Facilitator: Deeper integration with legitimate vendors improves transaction safety and efficiency.

Network Effects from Aggregated Data​

Aggregated anonymized data could yield:

  • Improved Recommendations: More data refines RL models, tailoring suggestions to nuanced preferences (e.g., “Morning people prefer caffeine at 7 AM”).
  • Community Insights: Trends like “80% of users report better sleep with magnesium” enhance decision-making.
  • Marketplace Efficiency: Data-driven vendor ratings reduce scams, boosting trust and retention.

These create value propositions like precision personalization and community-driven credibility, extending beyond basic tracking.

Defensive Moats​

To protect against competitors:

  • Data Network Effects: The more users contribute data, the smarter the system becomes, making it hard for new entrants to match its accuracy.
  • Brand and Trust: A reputation for privacy (via self-hosted AI) and effectiveness builds loyalty, especially in sensitive domains like substance use.
  • Integration Partnerships: Deep ties with marketplaces, healthcare, or wellness apps create switching costs, locking in users and partners.

These moats ensure Any Tracker remains a leader in intelligent, privacy-focused tracking.

These answers explore the philosophical depth, technical feasibility, and business potential of Any Tracker, aligning with its mission of pragmatic harm reduction and user empowerment.