Intro.deck.0-3

Customization in PrimeWill Shown 31 March 2014

There are 2 distinct problems:

• Which entities can a stripe represent?

• How do users edit their stripes?

Which entities can a stripe represent?

A stripe can represent entire domains (L0)

A stripe can represent primary categories of a domain (L1)

A stripe can represent structured data (L1+)

By extension, a stripe could represent:

• Subtopics of L1 categories

• Generated categories (sources, topics, etc.)

• Dynamic content (filtered/customized structrued data, individual live entities)

A stripe probably shouldn’t represent:

• Static content (a single article, slideshow, or video)

• Multiple distinct entities simultaneously

What’s left? We’ve covered every type of entity in Prime.

Now we have a taxonomy of possible stripes.

domain entities

group entities

NOT static entities

dynamic entities

NOT multiple entities

But we have a problem:

News Entertainment

?

We have 3 options:

Represent news topics as categories of News only

Represent news topics relevant to a domain as categories of that domain only

Represent overlapping topics as independent entities in each applicable domain

How could we represent overlaps as members of both categories? I don’t think we could do so consistently.

Which mental model is more intuitive to you?

• I think that stories concerned primarily with a more specific domain than News should be considered only a member of that domain.

• In cognitive psychology, ideal categories involve a collection of objects which are very self-similar when compared to each other and very dissimilar when compared to nonmember objects. [J. D. Smith]

• We may see it as ‘entertainment news article’, but I predict more users will identify a story about ‘Joaquin Phoenix’ more closely to movie times than to political stories.

• In other words, an entertainment story’s subject is arguably more salient than the fact that it’s a story.

Once we choose a model, we must apply it consistently everywhere.

All the time.

Always.

How do users edit their stripes?

This problem has sub-problems:

• How do we build a menu to provide access to all the potential stripe-worthy entities?

• How separate is the task of editing stripes from consuming stripes?

How do we build a menu to provide access to all the potential stripe-worthy entities?

The menu should offer a consistent mental model, regardless of the items represented.

We can accomplish that by:

pinning the L0 and search-autocomplete to one column, and

providing a consistent list interface in the second column.

⌕

News

Sports

Finance

Entertain-ment

Weather

Autos

Food & Drink

Health & Fitness

‹ News

World News

Custom topic

✓ Africa ⌉

South Asia ⌉

East Asia ⌉

Oceania ⌉

Europe ⌉

South America ⌉

North America ⌉

A compact, rectangular menu like that can be used in many different ways, and is terrifically responsive.

How separate is the task of editing stripes from the task of consuming stripes?

It's as separate as we want… but:

• The more features we provide for editing, the more separate editing will need to be to keep the relevant interaction model in working memory.

• The less separate editing is, the more contextual editing can be

• If we have different modes of editing, we risk overcomplicating the mental model.

Thanks!