Annotations (8)
“You don't necessarily need all of that user base to do this, either, or rather, it doesn't need to be your user base, because you might not need to build your own Mechanical Turk. If this kind of knowledge generalises enough, it might just be an API call from a world model. You can rent the cold start.”
Strategy & Decision Making · Technology & Engineering · Business & Entrepreneurship
DUR_ENDURING
API eliminates network effect moat via knowledge transfer
“The limitation of this, though, is that none of these systems really know why you watched those, or bought those, or looked at those, and they don't really know what those things are. Amazon has close to a billion SKUs, but all it knows about them is the metadata typed in by humans and some level of purchase correlation.”
Technology & Engineering · Philosophy & Reasoning · Psychology & Behavior
DUR_ENDURING
Correlation without causation: systems see patterns, not meanings
“But an LLM is, at a minimum, a step change in automated understanding of both what and why. The model can look at those words, images, videos and products, and all that metadata, and connect them to patterns that have some kind of understanding, or at any rate, some vastly broader kind of correlation. Today, Amazon will know that if you buy packing tape, you might want bubble wrap.”
Technology & Engineering · Strategy & Decision Making · Business & Entrepreneurship
DUR_CONTEXTUAL
LLMs infer context chains correlation can't see
“Any consumer internet system with critical mass becomes, in part, a Mechanical Turk. It looks at what the users do and draws conclusions from that. Amazon knows that if you bought X, you might buy Y, because it's seen what lots of people buy and saw that if they buy X, they're likely to buy Y. Google's dominance of search is based in part on seeing what people search for, and what they click on and what they search for next.”
Technology & Engineering · Strategy & Decision Making · Economics & Markets
DUR_ENDURING
Scale creates intelligence via user behavior aggregation
“The other half of the cold start problem, though, is that Amazon or Tiktok need to see what you yourself do so that they can work out what graphs might fit. The new user flow needs to throw enough ideas at you, and lead through enough choices, as painlessly as possible, that it can start to make those matches.”
Technology & Engineering · Business & Entrepreneurship · Psychology & Behavior
DUR_CONTEXTUAL
Minimize onboarding friction while maximizing signal capture
“Of course, this is a network effect, and comes with a cold start problem that's a barrier to entry. How do you provide those recommendations, suggestions and connections before you have users, and how do you get users before you can do that? (This is also a question for all machine learning startups.)”
Strategy & Decision Making · Business & Entrepreneurship · Technology & Engineering
DUR_ENDURING
Chicken-and-egg: value requires users, users require value
“Google, Amazon, Meta, TikTok, Tinder and a bunch of other companies (Uber, Doordash, Venmo) all know something about you. But they each have a very partial view, like the story of the blind men feeling an elephant. Your phone has a much wider view, in theory, although Apple and Google are very cautious in what they do with that, but it's limited in different ways: your phone could know what you bought on Amazon or what images you saw in Instagram, but it wouldn't see the graph that led Meta to s...”
Technology & Engineering · Strategy & Decision Making · Philosophy & Reasoning
DUR_ENDURING
Each platform holds fragment; no one sees whole
“The internet removed all of the old filters, curation and editing, so that now we have effectively infinite product, infinite media and infinite retail, and no way to or find or see what we don't know. The filters we had from the internet were very imperfect, and now we have a radically new and different kind of filter. That seems like a bigger question than replacing Google.”
Technology & Engineering · Culture & Society · History & Geopolitics
DUR_ENDURING
Abundance creates discovery problem requiring new filters
Frameworks (1)
Cold Start Problem Solution Ladder
Strategies for building network effects from zero users
Network-effect businesses face a circular dependency: users create value, but you need value to attract users. This framework outlines progressive strategies for breaking the cycle, from manual curation to API-based knowledge transfer to friction-minimized onboarding. Each approach trades different resources (time, capital, user experience) to accumulate the critical mass needed for self-sustaining network effects.
Components
- Manual Curation Phase
- Knowledge Transfer via API
- Friction-Minimized Onboarding
Prerequisites
- Clear understanding of target user behavior patterns
- Access to external data sources or APIs
- Product design capabilities for low-friction onboarding
Success Indicators
- Decreasing time-to-first-recommendation for new users
- Increasing recommendation relevance scores
- Declining reliance on manual curation or external APIs
Failure Modes
- Getting stuck in manual phase, never achieving algorithmic scale
- Over-indexing on API solutions, failing to build proprietary advantage
- Privacy backlash from aggressive data collection
Mental Models (5)
Network Effects
Systems ThinkingA network effect exists when each additional user increases the value of the pro
In Practice: Article explores how consumer internet platforms create value through aggregate
Demonstrated by Leg-be-001
Chicken-and-Egg Problem
Strategic ThinkingA circular dependency where you need A to get B, but you need B to get A. Common in two-sided market
In Practice: Explicit discussion of cold start problem in recommendation systems
Demonstrated by Leg-be-001
Correlation vs. Causation
Probability & StatisticsTwo variables may be correlated without one causing the other.
In Practice: Discussion of how traditional recommendation systems identify patterns without understanding causation
Demonstrated by Leg-be-001
Abstraction Levels
Decision MakingDifferent levels of abstraction reveal different insights. Operating at too low a level obscures patterns; operating at too high a level loses actionable detail.
In Practice: Article moves between concrete examples and abstract principles
Demonstrated by Leg-be-001
Build vs. Buy vs. Rent
Strategic ThinkingThree approaches to acquiring capabilities: build internally (control, customization, long-term cost
In Practice: Discussion of renting AI capabilities via API instead of building proprietary recommendation systems
Demonstrated by Leg-be-001
Connective Tissue (1)
Blind Men and the Elephant (Buddhist/Hindu parable)
The ancient parable of blind men touching different parts of an elephant, each reaching contradictory conclusions about what an elephant is, serves as a precise parallel for modern data fragmentation. Just as each blind man had accurate but incomplete information (the trunk feels like a snake, the leg like a tree, the tail like a rope), each digital platform possesses accurate but partial knowledge about users. Google knows search intent, Amazon knows purchasing behavior, Meta knows social connections. None sees the complete picture. The parable illustrates why synthesis across fragmented knowledge sources creates disproportionate value: the person who can integrate the trunk, leg, and tail descriptions understands elephants better than any individual observer. Similarly, an AI assistant with access to multiple platform graphs could develop holistic user understanding impossible for any single platform.
Author uses parable explicitly to explain fragmentation of user knowledge across digital platforms
Glossary (1)
Mechanical Turk
LITERARY_ALLUSIONAn 18th-century chess-playing automaton that appeared mechanical but secretly concealed a human operator inside
“Any consumer internet system with critical mass becomes, in part, a Mechanical Turk.”
Concepts (3)
network effect
CL_ECONOMICSA phenomenon where each additional user increases the value of a product or service for all other users
cold start problem
CL_STRATEGYThe chicken-and-egg challenge where a system needs users to create value but needs value to attract users
LLM (Large Language Model)
CL_TECHNICALAn AI system trained on vast text corpora that can understand and generate human-like text
Synthesis
Synthesis
Migrated from Scholia