Five Hat Racks in the AI Era
Richard Saul Wurman’s Framework Still Organizes Everything
In 1989, Richard Saul Wurman proposed that all information can only be organized in five ways: Location, Alphabet, Time, Category, and Hierarchy (LATCH). Thirty-five years and an AI revolution later, the framework still holds - and becomes more useful, not less.
The Five Ways
Location: Organize by where things are. Maps, floor plans, geographic distributions. When position matters, location organizes.
Alphabet: Arbitrary but universal order. Dictionaries, indexes, directories. When you know the name, alphabet finds it.
Time: Chronological organization. Timelines, histories, schedules. When sequence matters, time organizes.
Category: Group by shared attributes. Taxonomies, genres, types. When similarity matters, category organizes.
Hierarchy: Arrange by magnitude. Rankings, org charts, size comparisons. When relative importance matters, hierarchy organizes.
That’s it. Every organization scheme you’ve ever encountered is one of these five - or a combination.
Why It Still Matters
The LATCH framework endures because it describes fundamental cognitive patterns, not technology-specific solutions.
AI systems still need to present information. Users still need to navigate it. The question isn’t whether to organize information - it’s which of these five approaches fits the use case.
A search results page uses Time (recency), Category (type facets), and Hierarchy (relevance ranking). A knowledge graph uses Category (entity types) and Hierarchy (relationship strengths). An AI-generated report uses Time (narrative flow) and Category (topic sections).
The vocabulary is useful precisely because it’s technology-agnostic.
Choosing the Right Rack
Different questions call for different racks:
“Where is it?” → Location “What’s it called?” → Alphabet “When did it happen?” → Time “What kind is it?” → Category “How important is it?” → Hierarchy
When AI generates responses, the organizational choice shapes understanding. A list of recommendations organized by category (type) communicates differently than the same list organized by hierarchy (confidence) or time (recency).
AI Application
AI systems benefit from explicit LATCH thinking:
Retrieval: Which rack should search use? Time-based (recent first)? Hierarchy-based (most relevant)? Category-based (faceted)?
Generation: How should output be organized? Categorical groupings? Chronological narrative? Hierarchical prioritization?
Interface design: How should users navigate? Location-based exploration? Alphabetic lookup? Temporal browsing?
Making these choices explicit improves both system design and user experience.
The Combination Principle
Real interfaces combine multiple racks. A file system uses Location (folders) and Alphabet (sorted names). An e-commerce site uses Category (product types), Hierarchy (ratings), and Time (new arrivals).
The skill is knowing which combinations serve which purposes. LATCH provides the vocabulary for that design conversation.
Timeless Architecture
Information architecture frameworks come and go. LATCH persists because it maps to how humans cognitively organize the world.
AI doesn’t change this. AI generates, retrieves, and processes information - but humans still consume it through the same five organizational lenses they always have.
The hat racks are still there. AI just gives us more to hang on them.
Which of the five organization methods does your primary AI interface use? Is that the right choice for your users’ primary questions?
Related: 02-molecule—five-hat-racks, 02-molecule—taxonomy-design, 02-atom—maps-as-patterns