This is a project I spent a year and half building starting in 2011. At the time I had basically given up on programming because it bothered by hands/wrists/arms too much to do it. After speaking with a friend who suggested I code with the Microsoft Kinect, I started thinking about how text editing could be done differently so as to be efficient with lower-precision input devices like that. The project is largely dead now (2018), though I do still believe there were some pretty solid ideas in it that I hope to bring into another project at some point.
Pretty much everything else on this page is sorta copy/pasted from when I put a site together and posted it on Hacker News in 2013. It was on the front page for while there, and spread around the internet a bit (I even found an article written on it and its author on CSDN (China Software Developer Network)! And someone's thesis citing it in Brazil.)—but it never really went anywhere, and I wore myself out working on it and wanted to move on to other things once I realized funding wasn't just going to fall into my lap by posting a video on the internet.
- We have entered a new era of interaction with computer hardware with the advent of multi-touch phones and tablets, and gesture recognition systems like the Microsoft Kinect or the Leap Motion—and this trend is likely to deepen and continue as wearable computing and virtual reality do the same. The idea behind Tiled Text is that text editing via these hardware interfaces needs to be revised: forming selections using a cursor that operates at the character level is no longer ideal. Tiled Text introduces a system that operates on larger lexical/grammatical structures.
- The concrete examples on this site are from GIDE (Gestural IDE—written on Tiled Text) which I think proves the concept/engine.
- Tiled Text is not just for code! It could be configured to edit English documents via these units: paragraphs, sentences, clauses/phrases, words, characters (GIDE manipulates classes, members, lines, tokens, characters).
- I have a scheme for inserting new text: how it works (Or scroll down.)
- I can be reached at westoncb@[Google's mail service].com
The technology: Java/OpenGL (via libGDX for Android compatibiliity)/ANTLR/Eclipse.
I talk about Tiled Text as it relates to mobile computing because that's where I see its externally most interesting application. But, fact is I wrote Tiled Text to get around a mouse/keyboard injury I have, and it is, more generally, software for escaping those devices. Things like motion sensors, joysticks, and multi-touch surfaces are relatively interchangeable as input devices enabled by the software. In the video I use a joystick. There are some differences when switching to multi-touch, but the basic idea is the same.
What's not There (Yet): Text generation: with sufficiently structured text, the Tiled Text engine will narrow down the possibilities of what can be inserted at any given point so that the user can select textual constructs from a grid of options, rather than inserting everything one character at a time (this won't work for natural languages; it's for apps like shells and code editors, and perhaps some more exotic things). The engine for doing this is at ~(70% - 80%), but it has not been attached to the UI yet. This and a fuller selection of editing commands (copying, searching, etc.) are the largest remaining tasks.
Here's how it's going to work from the user's perspective:
They press the insert button. They are presented with a list of the syntactically valid constructs for inserting into the document at that point.
They make a selection. The selected construct is entered. This mean's the current insertion point has changed and a new set of options is now valid: a new list immediately replaces the last one.
That's roughly it. The list is actually displayed as a grid, and each cell has an input symbol associated with it (so you don't have to "move over" to a cell, just fire the input symbol: for instance, with a gesture recognition system, putting your hands in a certain postion selects the third cell). The grid becomes almost like a keyboard at this point, except each symbol on it accomplishes a lot more than inserting a single character.
Years passed. I didn't understand compromise then and preferred the prospect of working in another field over being crippled in the one I loved; though I eventually finished a CS degree despite having no real hope of returning to software development. I became more academic in the interim. I thought a lot rather than just building things. I liked philosophy and science (and philosophy of science!) and suspected there was some mathematics out there that would be as good as software. After graduating, my project was to discover why people thought mathematics was beautiful; failing that, at least acquire some rudimentery facility with it. That went so so, made a lot of progressbut it wasn't coding. About a year in, though, while explaining to a friend why coding with the Microsoft Kinect wasn't a real solution (with existing IDE's), the kernal of the idea behind Tiled Text occurred to me.
And, like the warrior who has put killing behind him, I blew the dust off my keyboard and set to work: my wrists would hold out for a few months until I could build this thing.
The initial idea was an early form of the code generation concept (didn't understand what a formal grammar was well enough to see how it would relate; thought of making some kind of linguistic constraint description systemthankfully my dad put me off the path of re-inventing the wheel here), but it was augmented in the early days by a few epiphanies. Real, old-fashioned epiphanies: if I were religious I might have thought they were gifts. The first was on the cycling behaviorthe other half of the core concept. The second gave the UI concrete form (was still thinking of building it on top of an existing text editor before this). The third mostly related to the architecture, and led to generalizing beyond just code editing.
Then there was work. And more work. I worked three days a week stocking groceries and four days a week on the project. I did this for about five months, with constant wrist pain. Then I got Type 1 diabetes. I had medical bills in addition to student loans and was barely scraping by. Then I worked for another 9 months or so, with progress slowing some.
NOTE: this was writen in 2013, so some of the following is outdated.
I needed a change. I'd been thinking about moving to Boston for a while, inspired by Paul Graham's comments on it from "Cities and Ambition." And now, here I am, in Harvard Square, excited about whatever's next. I'd like to continue with Tiled Text and GIDE, but I realize that may not be possible. What I really want now is to get in an environment where I can learn from others. I've been working in relative isolation for most of the time I've been programming, but I know the only way for me to progress is working with experts. Of course that will involve typing/mousing, but: ironically, I have learned how to mitigate my wrist pain from mouse/keyboard while working on software to obviate my usage of those devices.
And, incidentally, when I would procrastinate work on Tiled Text, I would often go to math books. My studies at these times were very fruitfull and I think now that mathematics has a chance at contending with coding, and I have finally seen some glimpses of mathematical structures that should be called beautiful.