Software Specification: Neoclassical Economics Tracker

Executive Summary

The Neoclassical Economics Tracker is a web-based intelligence platform that automatically identifies and profiles economists, pundits, and media figures who comment on economic policy, with particular focus on their positions regarding MMT vs. neoclassical economics. Similar to DeSmogBlog's approach to climate denial tracking, this tool maintains a comprehensive database of economic commentators, their institutional affiliations, funding sources, and historical statements on key economic issues. The platform combines content analysis, speaker identification, and automated research to help MMT advocates quickly assess the credibility and potential biases of voices in economic debates. Users can upload articles, videos, or transcripts, and the system will automatically identify speakers, distinguish them from merely mentioned names, and provide detailed MMT-informed background profiles including funding sources, prediction track records, and historical positions on economic theory.

Tool Overview

Type: Web application with content analysis engine

Description: A comprehensive profiling platform that automatically identifies speakers in economic content and provides detailed background analysis from an MMT perspective, including funding sources, institutional affiliations, and historical positions on economic theory.

Primary Users: MMT advocates, progressive economists, policy researchers, activists consuming economic media, and content creators preparing rebuttals to neoclassical arguments

Problem Solved: Eliminates the time-consuming manual research needed to understand the credibility and potential biases of economic commentators in media, allowing MMT advocates to quickly assess who they're dealing with in debates and media coverage

Key Features

[MVP] Content Upload and Analysis

Upload articles, video files, audio files, or paste URLs/transcripts for automatic speaker identification and content analysis

User Story: As an MMT advocate, I want to upload a video of an economic debate so that the system can identify who is speaking and provide their profiles

Complexity: medium

[MVP] Speaker vs. Mention Detection

AI-powered analysis to distinguish between people who are speaking/being interviewed vs. people merely mentioned in passing

User Story: As a researcher, I want the system to identify only the primary speakers in content so that I don't get irrelevant profiles for every name mentioned

Complexity: complex

[MVP] Voice Recognition Integration

HuggingFace-based speaker diarization and identification to extract speaker names from audio/video content

User Story: As a user analyzing a YouTube video, I want the system to automatically identify who is speaking so that I don't have to manually enter names

Complexity: complex

[MVP] Economist Profile Database

Searchable database of economist and pundit profiles with institutional affiliations, funding sources, MMT positions, and credibility scores

User Story: As an MMT advocate, I want to quickly lookup any economist's background and MMT stance so that I can understand their potential biases

Complexity: medium

[POST-MVP] Automated Profile Research

Web scraping and API integration to automatically gather background information from academic databases, think tank websites, and media archives

User Story: As a platform administrator, I want profiles to be automatically updated with new information so that the database stays current without manual effort

Complexity: complex

[MVP] MMT Position Tracking

Historical tracking of each person's statements and positions on MMT, deficit spending, inflation, and other key economic issues

User Story: As a researcher, I want to see how someone's position on MMT has evolved over time so that I can understand their ideological consistency

Complexity: medium

[POST-MVP] Prediction Accuracy Scoring

Track and score the accuracy of economic predictions made by profiled individuals over time

User Story: As an MMT advocate, I want to see who correctly predicted major economic events so that I can assess their credibility

Complexity: complex

[MVP] Credibility Dashboard

Visual dashboard showing credibility indicators, bias scores, and funding source analysis for each profiled individual

User Story: As a user consuming economic media, I want a quick visual summary of a speaker's credibility and potential conflicts of interest

Complexity: simple

[FUTURE] Real-time Content Analysis API

API endpoint that can analyze live streams or real-time content for speaker identification and instant profile lookup

User Story: As a power user, I want to analyze live economic debates in real-time so that I can get immediate context on speakers

Complexity: complex

User Workflows

Video Analysis Flow

Steps:

1. User uploads video file or provides YouTube URL
2. System extracts audio and runs speaker diarization
3. AI identifies distinct speakers and attempts name recognition
4. User confirms or corrects speaker identifications
5. System displays profile cards for each identified speaker
6. User can drill down into detailed profiles and credibility analysis

Screens: Upload Dashboard, Speaker Identification, Profile Results, Detailed Profile View

Article Analysis Flow

Steps:

1. User pastes article text or URL
2. NLP engine extracts all person names
3. AI categorizes names as speakers/authors vs. mentioned individuals
4. System looks up existing profiles or flags for research
5. User sees analysis results with profile summaries
6. User can contribute additional information or corrections

Screens: Content Input, Name Extraction Results, Profile Summary, Edit Profile

Manual Lookup Flow

Steps:

1. User searches for specific economist or pundit name
2. System displays profile if exists, or offers to create new profile
3. User views comprehensive background information and MMT stance analysis
4. User can export profile data or share analysis with others

Screens: Search Interface, Profile Database, Detailed Profile, Export Options

Data Requirements

Person Profile

Fields: name, title, institutions, funding_sources, academic_credentials, mmt_stance_score, credibility_rating, bio, photo_url, social_media_handles

Relationships: Connected to Statements, Predictions, Media Appearances, and Funding Sources

Storage Notes: Full-text search enabled on bio and statements fields

Statement

Fields: person_id, content, source_url, date, topic_tags, mmt_relevance_score, context

Relationships: Belongs to Person Profile, tagged with Economic Topics

Storage Notes: Indexed for semantic search and topic categorization

Media Content

Fields: title, url, content_type, upload_date, transcript, identified_speakers, analysis_status

Relationships: Links to Person Profiles through speaker identification

Storage Notes: Store original files in object storage, metadata in database

Institution

Fields: name, type, funding_sources, political_alignment, economic_school, website

Relationships: Connected to Person Profiles through affiliations

Storage Notes: Hierarchical structure for parent/subsidiary organizations

Economic Prediction

Fields: person_id, prediction_text, date_made, target_date, outcome, accuracy_score, topic

Relationships: Belongs to Person Profile, used for credibility scoring

Storage Notes: Time-series data for tracking prediction accuracy over time

Integrations with MMT Ecosystem

MMT Lens

Type: API

Share profile data and credibility scores to enhance MMT Lens fact-checking capabilities

Data Exchanged: Person profiles, credibility scores, and institutional affiliations

Real Progressives Knowledge Base

Type: Shared database

Cross-reference economist profiles with MMT educational content and contributor information

Data Exchanged: Economist classifications, MMT stance data, and educational content tags

Technical Considerations

Suggested Stack: Python/Django backend, React frontend, PostgreSQL database, Redis caching, HuggingFace Transformers for NLP, FFmpeg for media processing

Hosting: Cloud-based (AWS/GCP) with CDN for media files and horizontal scaling capability

Authentication: OAuth integration with social logins, role-based access for contributors vs. viewers

Key Challenges:

• Speaker identification accuracy in multi-speaker audio
• Avoiding false positives in name extraction
• Keeping profile information current and accurate
• Managing potential legal concerns around public figure profiling

MVP Scope

Included in MVP:

• Content Upload and Analysis
• Speaker vs. Mention Detection
• Basic Voice Recognition
• Economist Profile Database
• MMT Position Tracking
• Credibility Dashboard

Excluded from MVP (Future):

• Automated Profile Research
• Prediction Accuracy Scoring
• Real-time Analysis API
• Advanced speaker identification

Success Criteria: Users can successfully upload economic content, get accurate speaker identification 80%+ of the time, and receive useful MMT-informed background profiles for identified individuals

Development Phases

Phase 1: Core Infrastructure

Deliverables:

• Database schema and models
• File upload system
• Basic NLP pipeline for name extraction
• Simple profile CRUD interface

Phase 2: Content Analysis Engine

Deliverables:

• Speaker vs. mention detection algorithm
• Basic voice recognition integration
• Profile matching system

Phase 3: MMT Intelligence Layer

Deliverables:

• MMT stance tracking system
• Credibility scoring algorithm
• Visual dashboard interface

Next Steps

1. Set up development environment with Python/Django and React
2. Design database schema for person profiles and content analysis
3. Research and test HuggingFace models for speaker diarization
4. Create wireframes for core user interfaces
5. Establish data collection procedures for initial profile database