dit’s algorithm is guaranteed to satisfy envy-freeness up to one good: No
participant would envy the bundle of
goods received by another participant
if the former got to remove at most a
single good from the latter’s bundle.
Chores (tasks) division. This application is akin to goods division, but
divides chores or tasks that participants dislike. This is useful for assigning call shifts to doctors and nurses
(in fact, the application was added to
Spliddit due to requests from medical
practitioners), or for splitting household chores among roommates. Spliddit asks each participant to report
how much they dislike one chore over
another, and finds a way to split the
chores that achieves Pareto optimality
together with “equitability.” The latter
mandates that all participants must
equally dislike the bundle of chores
they are assigned.
Credit division. When a group of
people collaborate, a key task is to
divide credit for the outcome. In the
absence of external information, we
must ask the participants themselves
to report their perception of the relative contributions. Spliddit asks each
participant to divide 100 points among
the rest of the participants, and uses
these reports to divide 100 percent
credit among the participants in a
way that satisfies impartiality: Participants cannot obtain more credit by
misreporting their view of the relative
contributions of others. This incentivizes each participant to report truthfully, resulting in an accurate division
of credit. It was envisioned that the
primary use of this application would
be ordering authors of a research article by their contribution, but based
on user feedback, it has been equally
useful to teachers for grading individual students in group projects, and to
companies looking to assign performance-based bonuses to employees.
Taxi fare division. Imagine a group
of friends hailing a cab after a party,
and each dropping off at a different
location. Clearly, splitting the total
cab fare equally isn’t fair. This appli-
cation finds a fair split by implement-
ing a classic concept from coopera-
tive game theory called “the Shapley
value,” which is the unique solution
satisfying a combination of important
desiderata, and, among other contri-
butions, led Lloyd Shapley to receive
the Nobel Prize for Economics in 2012.
CONTRIBUTIONS TO SOCIETY
Interactions among groups of people
necessitate fair division of resources
or costs. While extensive research
in fair division theory has produced
beautiful and provably fair solutions
to many problems, people are unaware of them. Also, some of these
solutions are hard for humans to compute by hand. Spliddit was designed
to implement these solutions on an
online platform and allow people free
access to them.
Spliddit has initiated
the process of
bridging the gap
and practice in
which is part of
the larger field of
social choice theory.
For the past two and a half years,
Spliddit has stood by its mission:
˲ “To provide easy access to carefully
designed fair division methods, thereby
making the world a bit fairer.”
˲ “To communicate to the public
the beauty and value of theoretical
research in computer science, mathematics, and economics, from an unusual perspective.”
During this time, it has attracted
more than 100,000 users and solved
more than 35,000 real-world fair division dilemmas. Spliddit’s attractiveness stems from its three key attributes. First, it provides a quick and
hassle-free way to settle a dispute.
Figuring out individual contributions
in a collaborative project can take
days, while dividing an estate or settling a divorce can often take years
and result in significant acrimony. In
contrast, Spliddit elicits participants’
inputs in a single shot and finds a
solution in seconds. Second, Spliddit’s solution is provably fair, which
incentivizes people to not only use it,
but often reveals their inputs truthfully. For instance, in the rent division
application, roommates who are unaware of the preferences of the other
roommates may want to reveal their
preferences truthfully to ensure that
Spliddit finds a solution in which they
Figure 2. RoboVote.org helps groups of users pick optimal choices through voting.
Figure 3. This shows a selection of comments from the users of Spliddit.
"I have just used spliddit to share the rent of a 10 people house. And I was very impressed
with the final prices it came up with."
"Thank you very very much for your brilliant website."
"Not for nothin, but I think I love you. Great approach."
"I love your app!"
"Love your tools!" "Great site and apps."
"Great app :)"
"I greatly appreciate your Spliddit website!"