Skip to content

Layout Playground

Open the playground from the site navigation to compare layouts and analyze their typing behavior. Select a layout to see its keyboard, layers, and statistics. Use the top search field to find another layout, or return to the comparison table through navigation.

The Browse table starts with Layout, Author, SFB, SFS, Alt, Roll, Redir, and Weak columns. These metric columns use the same query definitions as the Metrics view. SFB and SFS use same-finger, different-key pairs. Redir includes every redirect category, while Weak includes redirects with neither an index finger nor a thumb.

Open the Metrics menu to add a metric column or reset the six default metric columns. A column’s header menu can edit its query and optional label. Editing a saved metric column creates an inline column definition without changing the saved metric. Drag column headers to reorder them, and resize columns by their header edges. The Layout column stays pinned during horizontal scrolling. Column definitions, visibility, order, and widths are saved with the workspace.

Cell backgrounds show population percentiles. Numeric text remains white. A percentile describes a layout’s position in that statistic’s comparison population, not an overall quality score.

Open a statistic to see its comparison histogram, layout ranking, and available contributor sequences. The histogram initially shows the central 90% of the population; switch to the full range to inspect the tails. This changes the chart’s visible range, not the underlying statistics.

Contributor keycaps show the actual keys and chord membership. Hold rails show duration: a rail runs from a held key or chord to its release, while an open rail continues beyond the visible excerpt. A held trigger appears once rather than being repeated beside every event.

Select a contributor sequence to browse its matching words and occurrence graphs.

A contributor is a keystroke n-gram that matches a statistic. Its count includes each matching sequence once, even when that sequence involves several fingers or distance buckets. Matching sequences load as you browse the contributor pages.

Outputs shows the final text typed by the selected keystroke sequence. One sequence can produce different text in different contexts. Each output’s percentage divides its weighted occurrence count by the selected contributor’s total occurrence count, so these percentages describe that contributor rather than the whole corpus.

When a typing path replaces text it produced earlier, the graph shows that text as transient. Outputs lists the final text after the replacement.

Word n-grams rank the shortest continuous word span associated with each match. A match that reaches into two words contributes their combined span, including the space between them. Repeated occurrences of the same span are combined.

Counts can be fractional when the analysis retains multiple equally optimal typings. The calculation keeps their exact weights; displayed percentages are rounded only for presentation.

Selecting a word n-gram opens its occurrence graphs. Each page shows one corpus context and its alternative typing paths. Boxes show key events; an arrow connects events that can follow one another. The selected word span is marked in the output columns, and highlighted transitions show where the contributor occurs. Transition weights count matching occurrences rather than probabilities, so overlapping matches can count the same transition more than once. Key-down and key-up events stay in typing order while their labels align with the text they produce.

Opening contributor details or a word graph keeps the current corpus results in place.

The Typing graph toolbar action analyzes a short text sample with the selected layout and its costs. It accepts up to 128 graphemes and shows branching and reconvergence between optimal typing alternatives, aligned to output positions.

The graph analyzes the text you enter while keeping the selected corpus results and contributor page in place. Its labels show the text after the usual word and separator handling.

For the source format, see Authoring Layouts.