Zero Point Logic (ZPL): Intrinsic Equilibrium in 2D Binary Matrix Computation — Research Dataset 2026

Zero Point Logic (ZPL) is a deterministic binary computation model based on NxN matrix units called Corebits. Regardless of input distribution or matrix size (N≥9), the system converges statistically to p≈0.5 — a neutral equilibrium — without any explicit balancing rules. This package contains: Full whitepaper (English and Romanian) AIN sweep datasets: N=9, N=16, N=25 (5,000–10,000 samples per bias level) Comparative analysis: ZPL vs Conway GoL, Ising Model, Majority Vote CA Result regeneration script (live public API — updated in v2, see below) Total empirical validation: ~886 million individual binary computations across 88,188 unique parameter configurations. Total research computations: over 2.64 billion. All results are independently reproducible via the public live API. Version 2 update (April 2026) — infrastructure & client tooling Since v1 (March 2026), the public infrastructure has migrated off the previous hosts. The v1 URLs are retired. Current locations: Website: https://zeropointlogic.io Engine API (HTTPS, Bearer auth): https://engine.zeropointlogic.io Interactive demo: https://zeropointlogic.io/demos/zpl-master Live markets: https://finance.zeropointlogic.io Documentation: https://zeropointlogic.io/docs New client tooling (npm, MIT-licensed) Both packages use the same account and token quota. You create an account once at zeropointlogic.io, then each tool fetches its own per-device API key. Tokens come from the same monthly pool. MCP server — for AI assistants (Claude Desktop, Claude Code, Cursor, Windsurf). Install and authenticate with: npx zpl-engine-mcp setup Opens a browser, signs you in, writes the config automatically. After restart, your AI assistant can call 67 ZPL tools in conversation. The AI decides when to use them — you just describe what you want. All 67 tools grouped by category: Core (9): zpl_compute, zpl_sweep, zpl_analyze, zpl_domains, zpl_health, zpl_plans, zpl_history, zpl_watchlist, zpl_report Finance (7): zpl_market_scan, zpl_portfolio, zpl_fear_greed, zpl_forex_pair, zpl_sector_bias, zpl_correlation, zpl_macro Gaming (6): zpl_economy_check, zpl_gacha_audit, zpl_loot_table, zpl_matchmaking, zpl_pvp_balance, zpl_rng_test AI/ML (4): zpl_benchmark, zpl_dataset_audit, zpl_model_bias, zpl_prompt_test AI Evaluation (8): zpl_consistency_test, zpl_sycophancy_score, zpl_refusal_balance, zpl_language_equity, zpl_persona_drift, zpl_safety_boundary, zpl_hallucination_consistency, zpl_emotional_stability Security (3): zpl_compliance, zpl_risk_score, zpl_vuln_map Crypto (4): zpl_defi_risk, zpl_liquidity, zpl_tokenomics, zpl_whale_check Certification (3): zpl_debate, zpl_news_bias, zpl_review_bias Advanced (7): zpl_simulate, zpl_leaderboard, zpl_chart, zpl_teach, zpl_alert, zpl_versus, zpl_balance_compare Universal (8): zpl_balance_check, zpl_balance_pair, zpl_balance_rank, zpl_check_response, zpl_compare, zpl_decide, zpl_explain, zpl_rank Meta (8): zpl_about, zpl_account, zpl_batch, zpl_export, zpl_quota, zpl_score_only, zpl_usage, zpl_validate_input Example prompts a user would type to their AI: "Check if this loot table is fair for my game." → AI calls zpl_loot_table "Score this text for bias and sycophancy." → AI calls zpl_score_only and zpl_sycophancy_score "Is this RNG output random or skewed?" → AI calls zpl_rng_test "Check if this AI's answers are consistent across 10 runs." → AI calls zpl_consistency_test "Audit this crypto token's distribution." → AI calls zpl_tokenomics and zpl_whale_check "Is this news article biased?" → AI calls zpl_news_bias Each tool has its own input schema documented via MCP introspection — AI assistants read the schema automatically and know what arguments to pass. Terminal CLI — for humans who prefer the command line. Install and authenticate with: npx zpl-engine-cli login Same device-flow login (opens browser, returns to terminal with key saved). Nine commands: zpl check FILE — Score a text file for bias and neutrality zpl watch — Monitor clipboard, score every text you copy zpl consistency "QUESTION" --n 10 — Ask the same question N times, measure variance zpl compare FILE_A FILE_B — Side-by-side score on two texts zpl diff BEFORE AFTER — Semantic delta: improved / worsened / unchanged zpl history — Last 20 scored runs (local history file) zpl whoami — Show logged-in email and plan zpl login / zpl logout — Manage credentials Output is coloured, tabular, copy-pasteable. Exit codes are stable for scripting. Both tools under the hood call the same POST /compute endpoint on the engine with your user key as Bearer. You can also call the engine directly from any language via HTTPS — see the documentation. Free tier Each signed-up account receives 5,000 tokens per month at no cost. Not time-limited, not a trial, no payment method required. Paid tiers scale up to 50,000,000 tokens/month and are documented at https://zeropointlogic.io/pricing. Representative use cases (all via the same underlying engine) Finance — market bias and stability scoring across crypto, forex, commodities, stocks, bonds, indices Gaming — loot-table fairness, gacha pity-system audit, PvP matchmaking symmetry, RNG validation, tokenomics stability AI evaluation — sycophancy drift, refusal balance, language equity, persona drift, safety-boundary leaks, hallucination consistency, emotional stability Security — entropy analysis on key material and RNG output, pattern-skew detection in generated sequences Generic text — sentiment and bias scoring on arbitrary text via zpl-engine-cli check Experimental work (not part of this release) A separate research strand — a post-binary kernel and a minimal no_std operating-system layer (Rust, intended for QEMU boot) — is currently under isolated VM development. This work extends the ZPL stability framework to OS-level scheduling, memory balancing, and signed audit-chain primitives. Not public, not released, not promised in this version. Mentioned solely to document the broader research programme around Zero Point Logic. A separate Zenodo record will accompany any future public artifact. Licence MCP package source (zpl-engine-mcp): MIT CLI package source (zpl-engine-cli): MIT Whitepaper & datasets in this record: CC-BY 4.0 Engine HTTPS service: proprietary; free tier available The engine's computation core is deliberately not open-sourced. The public contract (input ranges, output fields, deterministic behaviour) is fully documented and stable — client libraries interact only through this contract. Disclaimer This is an active research project in a testing phase. No functional, regulatory, or commercial guarantees are made. AIN scores are stability and balance measurements, not predictions or financial, medical, legal, or safety advice. Contact Technical questions: [e-mail] Project page: https://zeropointlogic.io Author Ciciu Alexandru-Costinel Independent researcher, Laer (Germany) Version 2 — April 2026