My research at ITP has centered around different ways to make digital text archives living entities in the real world. In recent years we have developed the astonishing ability to store and catalogue vast libraries of prose, and yet we are still unable to interact with this text without removing ourselves from public space. The hermetic bubble of the library has been replaced by the hermetic bubble of the computer.

This segregation is by no means inevitable, even though the vast majority of written words encountered in public spaces today fall into one of two categories -- advertising and informational. That predominance cheapens text, teaching us to ignore and devalue the written word. Modern technology has the potential to call for a reinterpretation of "publishing" -- one in which written words and the thought they contain are not shared merely through books and libraries but in the everyday world as well.

These word modules are designed as a first step in overcoming that boundary. They are small, self-powered repositories of up to 255 single words, accessed simply and easily on a 16x1 LCD by scrolling the right wheel. The words, which can be inputted using the wheel and a single pushbutton, reflect the diction of prior users. Built-in formatting rules allow for only combinations of letters that are permitted in the English language. Moreover, each module is equipped with a serial connection, which when connected to a computer allows it act as a gateway to online archives -- the selected word is sent to a custom search engine, which returns a related word or set of words to the module's archive. That connection also repowers the device, supplementing the built-in solar cell.

The ultimate goal is for these modules to be deployed in the "real world" and travel autonomously, appearing in serendipitous, unexpected locations. Waterproof and self-powered, they will capture and reflect the linguistic peculiarities of the places they have visited. Grouped, they can be arranged in sentences or slogans; alone, the scrolling motion provides for interative sentences.

Final Semester Report

For this class, I have built a prototype circuit for a single word module. It contains up to 255 words, each inputted to a PIC18F252 in flash memory using a rotary encoder with a built-in switch and displayed using a 16x1 backlit LCD module (Truly MCC161A6-3).

Logic

Each word is stored in sixteen bytes of program flash memory on a PIC18F252, rewritten on the fly as new words are entered. That single chip is also pre-programmed with 2110 three-letter rules, representing all prohibited combinations, including "word start" as a twenty-seventh character. With the "word start" character, two-letter combinations are rendered moot, as all first letters are allowed. The specific algorithms and "POKECODE" routines are automatically compiled by a custom Perl program on my own server and then hard-coded onto the chip in the normal program flash write via the Epic programmer.

Given the previous three letters and a new letter request, the PIC searches through its memory for prohibited third letters. Those letters form an array which is accessed when the scrollwheel is turned, so that prohibited letters are skipped.

There is currently not enough memory on the 18F252 to store four-letter combinations without an advanced encoding scheme. The best solution would be to encode actual allowed words, rather than rules, but that requires approximately 200k of memory, 170k more than the PIC has.

Energy Efficiency

The circuit was designed with energy efficiency in mind; it runs on either 3 or 5 volts, depending on the power source. Over the course of the semester, energy consumption was reduced from 22 milliamps at 5 volts (110 milliwatts) to 2 milliamps at 3 volts (6 milliwatts). Those would be typical measurements when the module is not connected to a computer. The energy consumed "on-line", with serial and a backlight enabled, would be significantly higher, but drawn from the host machine.

The specific battery has not yet been chosen, but a small coin cell is the likeliest candidate. Ideally, a solar "thin-film" backing would be installed to complement the battery for long-term outdoors use.

Physical Design

The circuit will be printed on small custom boards, which along with the LCD would measure no longer than four inches by one-and-a-half inches by one inch. The total length, encased in rigid polyurethane rubber, would be no longer than six inches. The casting materials and design need further refinement before an actual circuit can be risked in the ooze of a rapidly-setting mold.

Serial Connections

I have not yet finalized what sort of serial connection would be best employed, although I now tend to think that some sort of USB connection would be optimal. Speed is not necessarily an issue given the relatively small amount of data being transferred. Infrared does not solve the battery-recharging element of the host connection. To be honest, I have come to think of the serial connection as a luxury while developing the project, choosing to focus instead on the more essential aspects of the design.

PicBasic Pro Code