in 2021, i found it fascinating that technological discourse globally
spoke about gestural-input being the future of
human-computer-interaction (1
& 2),
but no one discussed by how much? so, when tasked with writing
a research paper during my third year of undergraduate study, i focused
my enquiry around trying to quantitatively compare a
head-tracked-interface with a graphical one.
at the time, i saw an increasing use-case for head-tracked interfaces in
environments such as hospitals, where people were asked a standard set
of questions with binary answer possibilities (such as the
osha-covid-screening-questionnaire).
covid-screening in india. source: india today.
i used a blob-tracking algorithm to develop a head-tracked-interface,
where a person could tilt their head to choose between either of two
displayed options (such as yes or no).
for the graphical-interface, i used a google-forms form on a laptop.
other usability gimmicks were stripped off from both interfaces, to
ensure that comparison was restricted to the efficiency of input
modalities.
the idea was to make multiple people complete a subset of the
osha-questionnaire (3/16 questions), using both interfaces,
one-at-a-time. by recording them doing this task, i could compare the
total time they took to answer each question via both input modalities
(finger-on-trackpad versus head-movement).
a person doing the same action (pointing to an answer choice) via both
input modalities.
then, smaller computing-actions to answer each question were broken down
via an adaptation of the goms-keystroke-level-model — originally
proposed by jef raskin in
the humane interface. jef proposed that the total time taken to complete a task on a
computer system is the sum of elementary gestures that the task
comprises (such as moving the mouse pointer, clicking a key, moving
between the mouse & keyboard, et-cetera).
in the given time frame and conditions, i was only able to perform this
experiment with 5 people.
findings:
there was a ~0.5s decrease in input time for each question,
after participants comprehended it, on the head-tracked interface. this
suggests that perceptual-interfaces might be more efficient in
situations where time to complete a task is vital.
next, people had to perform ~1.6 lesser steps, on average,
while using the head-tracked interface (according to the ‘steps’ defined
on the goms-keystroke-level-model). these steps include tasks like:
scrolling, removing their fingers from the trackpad after scrolling,
pointing to the answer choice, et-cetera. this finding can be
interpreted as lesser effort / energy being expended by people who use a
perceptual-interface. however, the exact difference could not be
accurately measured.
lastly, via follow-up questions, i established that all of the
participants found the head-tracked interface
easier to use. however, a couple of them said that using
the trackpad felt more ‘natural’ (since they were accustomed to it),
which presented an interesting hypothesis to pursue in the future — how
much does digital literacy impact the interface-effectiveness comparison
between a perceptual and traditional interface?
so, i concluded my paper empirically by stating that
perceptual-interfaces work better (faster & with reduced effort) than
graphical ones, albeit with the acceptance that my study was done in a
hyper-controlled environment, with zero error rates, with a
lesser-than-ideal sample size.
