The Problem Today
Current System Rewards
- Citations — popularity, not validity
- Journal prestige — brand, not rigor
- Novelty — interesting, not replicated
- Positive results — 95% of negatives never published
What Actually Matters
- Methodology quality — how was it done?
- Replication status — does it hold up?
- Statistical power — could it detect the effect?
- Conflicts of interest — who benefits?
Science runs on attention signals. The next infrastructure layer must rank claims by evidence strength — replications, contradictions, methodology — not popularity.
Follow The Money
The $2.9 Trillion Research Machine
Follow the money from global R&D spending through institutions to papers — and see where the signal breaks down.
The Attention Problem
60% of papers are never cited — invisible to science. And citations measure popularity, not evidence quality. This is what Socratic fixes.
$2.9T
Global R&D
3M
Papers/Year
60%
Never Cited
~50%
Don't Replicate
But where does the money actually go?
The Institutional Tax on Science
Before experiments run, institutions take their cut. Universities lose 34% to overhead. Of what's left, 80% goes to salaries.
12%
Gov Labs
15%
Corporations
25%
Nonprofits
34%
Universities
For Every $1M University Grant
~13%
reaches the bench
The $2.9T Research Machine
Where the money flows — and the signal breaks
Global R&D ($2.9T)
Institutions
~3M Papers/Year
Citations measure attention, not evidence. 60% of papers are invisible. ~50% don't replicate.
But where does money actually go?
The Institutional Tax
$506B burned before research starts
Of "research" money:
For a $1M university grant:
$340K
overhead
$528K
salaries
$132K
science
The 11 Phases of Science
Science has distinct phases from question to practice. Each can be reimagined with the right infrastructure. Click any phase to explore how it transforms.
Question
Socratic maintains live map of ALL claims, evidence strength, gaps, contradictions. AI generates testable hypotheses: "If A and B, then C should follow — but C untested."
Pre-Reg
Hypothesis submitted to blockchain — timestamped, immutable, encrypted until complete. Prediction market: others bet on P(true), creating prior from crowd wisdom.
Funding
Score = Evidence quality (30%) + Novelty (25%) + Researcher FICO (20%) + Feasibility (15%) + Impact (10%). No human review for initial filter.
Design
AI generates optimal sample size, controls, randomization, blinding. Pre-analysis plan auto-generated. 10,000 simulations run before real experiment.
Data
Robotic data collection, direct instrument → database, no human touch. Every data point timestamped on blockchain, cannot be modified.
Analysis
Pre-registered analysis runs automatically: data locked → script runs → results committed. Multiverse analysis shows ALL reasonable variants.
Review
Immediate deposit. AI Review (instant): GRIM, p-curve, methods. Expert Review: top 5 auto-invited, paid $500-2000, named, citable.
Publication
No journals, no gatekeepers. All work immediately public. "Journals" become curators. Ranked by evidence grade, not journal name.
Replication
Replication value = Impact × Uncertainty × Feasibility. Funders post bounties, labs compete, paid regardless of outcome.
Integration
Every claim auto-extracted, normalized, linked. SUPPORTS / CONTRADICTS / EXTENDS mapped. Real-time updates when new papers publish.
Translation
Evidence threshold triggers: Grade A → alert clinicians. AI drafts guidelines, humans review. Days, not years.
The 5-Layer Infrastructure Stack
Science 2.0 requires five interconnected infrastructure layers. Socratic is Layer 1 — the evidence foundation everything else builds on.
Translation
Incentives
Execution
Intelligence
Evidence
← Socratic TodayWhat Would Change
Idea to Publication
Faster iteration, less bureaucracy
Replication Rate
Pre-registration + incentives
Discovery to Practice
Real-time evidence translation
Funding Waste
Evidence-based allocation
Retraction Detection
Automated integrity checks
Negative Results Published
Mandatory registration
The goal: Transform science from a prestige economy to a truth economy.
Roadmap
Building Science 2.0 is a multi-year journey. Here's how we get there.
Phase 1: Evidence Foundation
Now → 6 monthsComplete claim extraction, grading, and relationship mapping
Phase 2: Intelligence Layer
6-18 monthsAI Scientist, SLR generator, AI pre-review
Phase 3: Execution Layer
18-36 monthsPre-registration integration, data repository, lab partnerships
Phase 4: Incentives Layer
36-60 monthsPrediction markets, replication bounties, Researcher FICO
Phase 5: Translation Layer
60+ monthsEHR integration, clinical decision support, feedback loops
This is what we're building toward.
Socratic is the foundation — the evidence layer that makes everything else possible. Every claim extracted, every contradiction mapped, every evidence grade assigned brings us closer.