Dominic Cummings: Ideas Beyond the Blog

Loathe him, or like him, Dominic Cummings' has had a significant influence on the UK. The EU referendum and the Boris wave washed over Britain — and there may be more to come as Reform rises and ideas like ARIA begin to spread.

This short essay though isn't about that. Instead, it argues that Cummings' dense, ambitious writing style means his ideas are often best understood through a simulacrum. Dialogue with an LLM allows the casual reader to understand more than they might glean from navigating Dom's sprawling blog posts alone.

Dominic Cummings' writing features remarkable (even exhilarating) intellectual leaps. His blog posts integrate concepts from seemingly disparate fields, including political strategy, cognitive science, physics, history, management theory, military doctrine, and complexity science. Consider this characteristic example:

"Explore the frontiers of the science of prediction across different fields from physics to weather forecasting to finance and epidemiology. For example, ideas from physics about early warning systems in physical systems have application in many fields, including questions like: to what extent is it possible to predict which news will persist over different timescales, or predict wars from news and social media? There is interesting work combining game theory, machine learning, and Red Teams to predict security threats and improve penetration testing (physical and cyber). The Tetlock/IARPA project showed dramatic performance improvements in political forecasting are possible, contra what people such as Kahneman had thought possible. A recent Nature article by Duncan Watts explained fundamental problems with the way normal social science treats prediction and suggested new approaches — which have been almost entirely ignored by mainstream economists/social scientists. There is vast scope for applying ideas and tools from the physical sciences and data science/AI — largely ignored by mainstream social science, political parties, government bureaucracies and media — to social/political/government problems (as Vote Leave showed in the referendum, though this has been almost totally obscured by all the fake news: clue — it was not 'microtargeting')."

This is typical Cummings. He traverses disciplines, urging exploration of predictive science across physics, meteorology, finance, and epidemiology. He connects physical early warning systems to news persistence and conflict forecasting, then pivots to game theory, machine learning, and Red Teams for security threat prediction. His breadth of knowledge is undeniable, yet this density creates a whirlwind that can obscure his central insights beneath layers of interconnected thoughts.

In contrast, a RAG-powered chatbot provides a superior way to understand Dom's thoughts. First, structure and coherence are improved; a chatbot can arrange ideas drawn across various texts into a coherent, streamlined argument. Second, the connections between disparate concepts become clearer and more explicit; the chatbot makes explicit connections that Cummings frequently implies or rapidly skips past. Third, filtering capabilities allow the chatbot to remove tangential details and repetitive content, focusing purely on core insights. Lastly, accessibility is enhanced by presenting complex, dense concepts in a digestible, clearly articulated format.

This synthesized output contrasts with the often overwhelming experience of reading Cummings' unfiltered prose, which has a "firehose" quality — rich in insight but difficult to parse without sustained effort. Yet, wading through this density can yield significant rewards. Personally I'm grateful to him for surfacing stuff like Boyd and Steve Hsu’s podcast. Even my current job working in AI research can trace its lineage through Dom to Ziv’s blog.

This model of engaging with a complex thinker through an AI simulacrum holds broader potential. Imagine applying similar RAG-powered chatbots to the dense works of philosophers, economists, or historical figures whose original texts are notoriously difficult to penetrate. Such tools could democratise access to challenging ideas, allowing a wider audience to grasp core concepts and engage critically with influential, albeit difficult, authors, effectively creating interactive simulacra of their thought.

A RAG-powered chatbot trained on Dominic Cummings' extensive writings provides a more insightful, structured, and accessible presentation of his ideas than his original texts alone typically allow. Such AI tools represent a promising shift in how we engage with complex thinkers and extensive textual bodies, potentially transforming our capacity to assimilate and critically evaluate sophisticated intellectual material. Or maybe I just wanted to provoke real Dom into a debate with his facsimile.