The Keyboard Versus EyeQChat

Updated 9th September 2025

EyeQChat is a set of technologies that make using computers easier for everyone.

Please note, the working title for what is now called EyeQChat was Eucotua. That name is still present in the accompanying application, and in some of the images taken of that application within this document.

Resources

The following resources support this introduction to EyeQChat:

• The YouTube video introduction to the system is https://accessibleshell.com/intro.
• The Microsoft Store web page for the accompanying application is https://www.microsoft.com/store/apps/9NDJ3B0B8SW8.
• The Microsoft Store direct link is ms-windows-store://pdp/?productid=9NDJ3B0B8SW8
• The dictionary and performance data is available in the CSV format file words.csv

You can make contact at contact@accessibleshell.com.

Introduction

The Keyboard Versus EyeQChat is an application that demonstrates how and why one of its core concepts of EyeQChat works. It does this by contrasting a minimal mechanical implementation of a QWERTY keyboard with a minimal mechanical implementation of an EyeQChat input control. Both the minimal mechanical implementations behave in entirely predictable and repeatable ways, which allows them to be compared.

A world class implementation of EyeQChat will be many times better that this the minimal mechanical implementation. A world class implementation of a keyboard may only be a little better than the minimal mechanical one presented here.

The Keyboard Versus EyeQChat also includes a stripped down implementation of the EyeQChat concept using Windows' built-in next word predictor so that you can get a feel for how using a full application using the concept may feel. The full application is (currently) called EyeQChat Chat and will be made available later into 2025.

The video introduction to this site can be found here.

The Keyboard Versus EyeQChat is available from the Microsoft Store website at https://www.microsoft.com/store/apps/9NDJ3B0B8SW8 or directly from the Microsoft Store app from ms-windows-store://pdp/?productid=9NDJ3B0B8SW8.

Using the application

The application consists of four tabs:

• Keyboard: Presents a simple 26-letter QWERTY keyboard with an Enter key and allows its performance to be compared to a EyeQChat control presenting four predictions.
• EyeQChat: Presents a simple four prediction EyeQChat control and allows its performance to be compared to a 26-letter QWERTY keyboard.
• Demonstration: Presents a minimal implementations of a EyeQChat text-to-speech application.
• About: Presents information about the application, including copyright and patent notice together with a link to the accompanying webside.

Keyboard tab

When the application starts up it displays a minimal QWERTY keyboard consisting of the letters A to Z together with an Enter key, 27 keys in total.

In addition to the key caption (A to Z plus Enter) the keys are annotated as follows:

• If there is any possibility that pressing the key will result in a word in the dictionary the weighted proportion of those words is indicated as a percentage below the caption. When the keyboard is initially displayed "T" has a proportion of 13.75%, "X" of 0.02% and "Enter" has no value because no word is zero letters long.
• The most likely key is filled with a cyan background Other keys are filled according to their likelihood proportional to the most likely key, with the most likely being most cyan and the least likely most white.
• The keys are ranked according to their likelihood and keys falling within the top 90% are shown with a bold border.

You can type by clicking on the keyboard buttons by clicking with a mouse or pressing the corresponding key on the physical keyboard.

• When a letter key is pressed an image of the key is collected above the keyboard.
• When the Enter key is pressed the collected keys are cleared.
• While not present on the displayed keyboard, pressing the physical Backspace button on the keyboard will delete the last letter typed.

When the letters collected above the keyboard form a word in the application's dictionary a comparison table is drawn. The table compares typing the word with the QWERTY keyboard and three sizes of a EyeQChat controls is drawn.

Additionally, if the letters collected above the keyboard are not a word but are the prefix of one or more words in the dictionary, the most likely of those words is placed into a comparison table.

Each row of the table shows the keys necessary to enter the word. So for the QWERTY keyboard that will always be all the letters of the word plus the Enter key. For the EyeQChat control it will be a sequence of zero or more navigation buttons followed by the selection of the word. While the navigation buttons are normally shown as arrows, the navigation buttons here are replaced by the upper and lower bounds they are navigating to.

For the word HELLO you will notice:

• The QWERTY keyboard with 27 keys takes six clicks to create the word.
• A EyeQChat control with nine keys also takes six clicks to create the word.
• A EyeQChat control with 15 keys takes only five clicks to create the word.
• A EyeQChat control with 27 keys takes only four clicks to create the word.

An attempt at a metric to compare the performance of the different sized EyeQChat controls is also displayed. The "keyboard effeciency" is 27*(w+2)²)/(k*c²) where w is the length of the word, k is the number of keys on the control and c is the number of clicks used to enter the word. This number is displayed as a percentage and for the QWERTY keyboard will always be 100%.

Using this metric:

• The keyboard scores 100%, because it is the benchmark.
• The nine key EyeQChat control scores 300% because, while it took the same number of presses to enter the word, it presented a third of the buttons to do so.
• The 15 key EyeQChat control scores 259% because it presented more buttons even though it took fewer clicks.
• THe 27 key EyeQChat scored 225% because it required fewer clicks than the QWERTY keyboard for the same number of buttons.

A second attempt at a metric is indicated in parathesis. This is the same as the preceeding metric, but without squaring the number of clicks used to enter the word.

EyeQChat tab

The EyeQChat tab contains a nine key EyeQChat control.

The buttons of the EyeQChat control are annotated in the same way as the buttons on the QWERTY keyboard.

The keyboard is on a tab control that also presents a minimal EyeQChat control.

To briefly reiterate how the control works:

• The current word predictions are displayed to the right. Clicking one of these words selects it.
• To the left of the word predictions are navigation arrows. The top and bottom of these selected predictions before and after those shown. The arrows pointing between adjacent words will select predictions between those words. When a navigation button is pressed, the control will be redrawn with the new selections.

When there are no words in the dictionary before the first displayed word, after the last word or between adjacent words the corresponding navigation button is suppressed.

When a word is selected on the EyeQChat control:

• A comparison table for that word will be displayed.
• The EyeQChat control will reset to its initial state. (The mechanical implementation of this control means it will always reset to the same initial state.)

Demonstration tab

The demonstration tab give a limited feel of what a full EyeQChat application may feel like.

Some of the limitations of this implementation are:

• It only supports words consisting of uppercase letters. It does not support words like "IT'S".
• The buttons are of a fixed size. You can change the number of available buttons by changing the size of the application's Windows.
• A maximum of 13 words can be added to the left side.
• The application does not attempt to fix the issue when there are too many words entered to fit in the available area. The full application handles this situation.
• No voice or font selection is available.
• The Windows Text Prediction Generator is not available on all computers. When it is not available the application will fall back to the mechanical implementation.

Statistics

Most comparisons of EyeQChat to a QWERTY keyboard will show that EyeQChat takes fewer clicks to achieve the same result. When first being used, more time may be needed to think than with the keyboard, but the process quickly becomes natural.

The table below measures the average number of clicks to enter words in various scenarios.

The data used to create these numbers is contained in the CSV files linked in the resources section towards the top of this page. The CSV files has the following columns:

• The rank of the word, from 1 to 143475 and listed in this order.
• The word itself as a sequence of uppercase letters.
• The likelihood of the word. The numbers in this column sum to about one.
• The number of letters in the word plus one, that is the clicks to manually type it.
• The number of clicks to enter the word using a nine button EyeQChat control.
• The number of clicks to enter the word using a 15 button EyeQChat control.
• The number of clicks to enter the word using a 27 button EyeQChat control.

The data in this CSV file is from an open source repository. The manner the weights were measured is uncertain, but they appear reasonably accurate for most purposes and replacing the words and weights with values from another source are unlikely to result in significant chages to the presented statistics.

Legal stuff

This appearance and embodiment of this application is protected by copyright and patent law.

Copyright © Accessible Shell LLC, 2025
Patent Pending