Talkie vs WorldMonitor

Talkie is a 13-billion parameter language model trained exclusively on English-language texts published before 1931 — the largest vintage language model built to date. Created by researchers Nick Levine, David Duvenaud (University of Toronto), and Alec Radford (of GPT and DALL-E fame), it represents a novel approach to understanding what training data really does to a model. The research insight is elegant: modern LLMs are so thoroughly contaminated by modern internet data (directly or through distillation) that it's nearly impossible to isolate what the model "knows" from what it absorbed during training. Talkie solves this by hard-cutting the training corpus at 1931 — predating digital computers entirely. This lets the team run controlled experiments impossible with contemporary models, such as teaching the model to write Python from examples alone and measuring how quickly it generalizes. Talkie was trained on ~260 billion tokens of historical text and fine-tuned using direct preference optimization with Claude as judge on structured historical documents (etiquette manuals, letter-writing guides). It's openly available on Hugging Face for research use. It also happens to produce wonderfully formal, slightly anachronistic prose.

WorldMonitor is an ambitious solo-built open-source project that aggregates 500+ news and data feeds across 15 categories — geopolitical events, financial markets, military movements, infrastructure alerts, disease outbreaks, space events, and more — into a single real-time dashboard with a 3D interactive globe at its center. Each country gets a dynamic risk score. Events are geolocated and pinned to the globe. You can drill into any region for a synthesized AI briefing. The AI analysis layer runs entirely on Ollama — no API key, no external cloud calls. The system connects to your local Ollama instance and uses whichever model you prefer to generate briefings, summaries, and threat assessments from the aggregated feeds. The globe itself renders 45 switchable data layers including conflict zones, trade routes, weather systems, submarine cable infrastructure, and satellite coverage maps. The project launched on GitHub four days ago and already has over 51,000 stars — one of the fastest-growing repos this week. It's AGPL-3.0 for personal use (commercial license required for business deployment). The real story is what it reveals about the appetite for serious geopolitical and global risk tooling outside the expensive Bloomberg/Palantir tier — and the fact that a small team built something this polished as an open-source first release.