Taste Cabinet
An archive of judgment you can actually practice with.
Project snapshot
Design, UI, UX, Product, Engineering, Strategy
Growing
JavaScript, HTML, CSS, Cloudflare Workers
A live, growing cabinet with pinboard and constellation views, interactive visual specimens, authored connections, practical critique exercises, responsive browsing, and shareable deep links.
Turn design principles into real practice without collapsing into a quote archive, a generic course library, or a dashboard full of content.
A self-initiated working v1 that turns design principles into interactive specimens and paired critique exercises. The growing cabinet is built to move people from collecting references to naming and applying judgment.
How the system works
Start deliberately small
The first release prioritizes a readable point of view and a dense connection graph over the appearance of a large library.
Make the idea operable
Each artifact translates an abstract principle into a visual object that can be manipulated, inspected, or tested.
Hold two ideas at once
Authored pairings reveal agreement and productive tension, then turn the relationship into a critique lens.
Separate prototype proof from source proof
The working product can be shown honestly while publication-grade source and rights verification remain a separate gate.
The problem
Most inspiration libraries teach collection, not judgment. They are good at helping designers save references and weak at helping them explain why something works, where a principle breaks, or what to do differently in their own work.
Taste is often treated as instinct or personal preference. That makes it difficult to teach, critique, or apply deliberately.
The approach
Build a learning object that feels like a collected cabinet rather than a course dashboard. Each principle gets a memorable visual specimen, Nikki’s curatorial framing, and connections to other ideas in the collection.
The museum metaphor is functional. Dark cabinet surfaces, paper as an object, gold framing, and orange actions make each idea feel selected and placed instead of generated into a content grid.
The teaching loop
The product moves through four steps: browse the Atlas, operate a specimen in the Vitrine, hold two ideas together in a Connection, then use that tension on something you made.
The bet is that judgment becomes more usable through repeated comparisons between an idea, visible evidence, and a real design decision.
Browse the cabinet
Encounter ideas as a spatial collection instead of an endless feed.
Operate one specimen
Use the interaction and curatorial note to make the principle concrete.
Compare two lenses
Feel where two principles agree, sharpen, or pull against each other.
Critique your own work
Use the pairing to name what survives, what should change, and the next move.
What I designed and built
I designed the product model, visual system, artifact schema, interaction patterns, specimen behaviors, connection graph, and the framework-free web implementation. The launch collection includes authored pairings, two desktop cabinet modes, responsive mobile browsing, and direct links to artifact and exercise states.
The prototype has no production content service or application framework. That kept the first release inspectable and made the content, interactions, and visual language the actual product.
The scope decision
The project originally held a larger demo set and a separate Second Right Answer exercise route. I retired the old demo records and folded the useful practice idea into the Connection surface.
Starting with a focused launch set is deliberate. The collection can grow, but the next meaningful proof is whether one complete cabinet → artifact → connection session helps someone make a more specific design decision.
Where it is now
Taste Cabinet is a live working v1 and a self-initiated product, interaction, content-system, and creative-development case study. The launch curation is live, and the collection structure is ready to grow.
It is not yet a validated learning product or a publication-grade scholarly archive. Source and rights verification remain a separate editorial gate, and the next product step is a small observed learning test.