I’ve recently finnished the first version of my chapter on the Dasher project. I hope that you’ll find this interesting and that I’ll get as much valuable feedback on this post as I did on the previous. Your feedback is invaluable to me.

Chapter 2: Dasher

This method requires some introduction. I’d recommend that you check out the video below to get a feel for how it works:

Demonstration of Dasher

I am working on a web based demonstration of the input system. The idea would be to quickly demonstrate it, without requiring any installation. Along with that, I’d like to demonstrate some of my suggested improvements.

A practical text editor for tablets / smartphone?

I believe that a dual Dasher system would be a good fit for modern tablets. The left thumb could for example provide for autocomplete by providing words presented by a Dasher Algorithm, and the left for character input. I’d say that this setup would feel quite natural – also it mimics what games are doing, which is always a good thing.

To extend this a bit further to truly accommodate for the power users, think of it in terms of VI editor modes. The left thumb could provide content editing commands, whereas the other would provide for text, both with the cursor and tree, based on probability model (insert mode). Obviously not for everyone but perhaps really useful for people who can handle a more advanced text-editor.

I don’t know if this type of system will ever reach high levels of market penetrations however, apart from in niches. What I do believe however that it’s a good source of inspiration.

Morse code

Chapter 3: Morse code

While being nearly 200 years of age, morse code still has some interesting aspects. For one, it compresses itself by assigning shorter strings to more frequent letters.

I do happen to have an idea that is to create a demonstration of some ideas that I came up with while working with the sigma project.

What I find very useful here was to highlight letters that my rather simple prediction engine came up with. Basically a log function on the probability that controlled size and color of the letters. With this system, I was able to write Chinese in the old system around 20 wpm. Which is quite impressive given a huge set – and the fact that I don’t know Chinese. This has to be taken in context and while I don’t see people switching to this system, it is an example of how to apply these findings to deal with selecting symbols of large sets / alphabets.

The xkcd strip above inspired me to approach the ordering of the letters in a 2 dimensional plane. Placing them after each other as letters on a line doesn’t make sense whereas 2d distance increases more than it should between similar letters. Making it more natural for a human to scan. Tried and used in sigma project for the quad board.

Above you’ll find a video on a quick very nonscientific analysis on two methods I’ve investigated in the sigma project. The quadboard turned out to be less efficient than I had hoped, however for some cases such as huge data sets, it seems to have its applications. It can be used to make a rather clever LaTeX based equation editor for example.

Also in the video, you’ll find analysis on my worst-case method, that I made to demonstrate how silly a linear search is. Or quasi linear, as in the case of an onscreen keyboard. Paired with a dynamic view, it performs quite nicely even with a very rudimentary prediction engine.

Thank you for reading!

Thank you for reading and please do not hesitate to comment whatever thoughts you might have. Your feedback is invaluable and I’ll do a post summing up all the ideas I’ve got already. If you don’t have anything to say but like what I am doing please share this so that we can find as many people as possible who are willing to push the boundaries on how we type.

Discussion down below or at
Hacker News

Spread the word

Leave a comment

Your email address will not be published. Required fields are marked *