Ashok's Assignment 13

I read the thesis "Tangible Interfaces for Manipulating Aggregates of Digital Information" by Brygg Anders Ullmer, for my thesis review. The basic idea behind this thesis is to offer an alternate representation of data interaction within digital systems. The main motivation behind this thesis is that syntactically, GUI interfaces do not necessarily overtly contextualize data and data constraints in a comprehendible or readily usable manner. The simplest example I can think of– with a GUI interface one is expected to use the same device to word process as one would be to play media or do scores of other functions.

To contrast with the GUI interface, Ullmer develops a TUI or a Tangible User Interface. All the components and data manipulations within the TUI are done physically, and thus, have some advantages over the GUI stemming from the fact that a user, by visual inspection and haptic manipulation can better understand the data, constraints, and various data manipulations that can be performed. The idea of a token (a shaped block) is introduced to represent data or a group of data (specifically “aggregates of digital information” which allows the physical system to be of manageable size and yet encompass vast amounts of data) and constraints are introduced that constrain the movement of the block to 1 dimension. For example, a block within a track can move left or right (along the track) whereas a wheel can fit into a single circular slot on a grid and can turn clockwise or counterclockwise. Within this simple physical scheme data can be digital, as in whether a block is currently within a given constraint (this is referred to as the “associate phase” in the paper), and analog, as in I can move this block a continuous amount within this constraint (this is referred to as the “manipulate phase”). Essentially, various constraints work with various tokens, and thus, the physical attributes of the token and the physical attributes of the constraints naturally contextualize the data. Tokens can be constraints too (as with the case of nested tokens) which analogues to inheritance.

Ullmer attempts to use physical token schemes to interact with Media data linked from the internet (“MediaBlocks”) and a Query type search associated with database systems (“tangible query interfaces”). The thesis statement aims to show that constrained tokens provide an effective means for interacting with aggregate data.

MediaBlocks introduces tokens that act as containers for online media –ie: the physical objects merely reference the media but do not have memory to keep the media. This allows for the containers to be “limitless.” MediaBlocks interface with media sources, which “copy” the media data onto the block, and media playback devices, where the block “pastes” the data into the device. Examples of input devices include digital whiteboards and a video camera. Examples of output devices include a network printer and a wall mounted display. Finally a media sequencer can be used to make multimedia presentations by interfacing with the physical blocks to allow for editing by block control.

The tangible query interfaces project uses multiple wheel tokens (as opposed to the block token) with each wheel representing a parameter of a given database search. Thus, by cascading various parameter wheels, one can physically access the the subset of data that fits their parameter specifications. A non-physical LCD parameter bar is also used to navigate the database (the parameter bars are raised and lowered much like histograms).

I thought this thesis was extremely interesting. The syntax of systems and attempting to create physical systems which are inherent to human understanding for a given task is something that will continue well past our lifetimes, and not just in the realm of computing or digital data manipulation. I especially liked the physical realization of media data manipulation exhibited in MediaBlocks. The only real criticism of the paper I have is the end section with “User Studies”- Part of me wonders if it’s necessary and if there isn’t another way to “quantify” or satisfy the requirements without trying to explicitly show that the TUI is better than the GUI which seems a little foolish. In a way, scientists need “results”, and so after unveiling the Ph.D. implementation a User study is necessary to prove the thesis statement made at the very beginning of the paper. It reminded me a lot of Ben Shniederman’s Science 2.0 slide from his presentation where he talked about User studies and their role in HCI projects. I guess I don’t have a criticism but more a thought– would there be a way, either by the author rephrasing the thesis statement or by some other mechanism, in which such a User Study wouldn’t be required at the end of the Ph.D. thesis?

Last modified 4 December 2007 at 2:54 am by Ashok.Basawapatna