A Brighter Future
for Crowd Work
Skill ladders may help crowd workers to “skill up” as they work.
But what other technical innovations will lead to better opportunities
for crowd work?
By Jeffrey P. Bigham and Kristin Williams
Work is changing. Crowd work, with its open calls, short contracts, and non-existent support for worker development, is proliferating. Work won’t look exactly like today’s Amazon Mechanical Turk in the future, but elements of it (often the least desirable) seem likely to influence the future of work if we don’t figure out how to stop it.
There are many angles one could take on this problem. For instance, an on-going
concern about crowd platforms is they are low-paying, so we could create better policy
to protect workers. Workers generally have little power on today’s platforms; we could
facilitate worker organization to better enable them to protect themselves. These are
valuable and necessary endeavors.
This article is about what systems
builders can do to create a brighter future for crowd work.
It is tempting to believe what we
need is to design a new platform from
scratch that represents our new future—this may in fact be one route to
a better future—but building new platforms is risky. It takes effort to build the
platform, attract workers there, and in
the end you might still end up with a
platform that you later realize misses
some core component of what you need.
We are less interested in the crowd
platform than what we build on top of
it. Many crowd platforms can be seen
as being “crowd complete,” a loose
play off of “Turing complete,” meant to
convey the idea that given a certain set
of functionality, any crowd platform
could be used to create any other.
Many of the tools we have built with
our collaborators use Amazon Me-
chanical Turk, Upwork, and other plat-
forms, in order to take advantage of the
built-in capabilities of each. Amazon
Mechanical Turk makes large numbers
of workers available via an API, whereas
Upwork offers workers with advertised
(and rated) expertise in different pro-
fessional areas (e.g., programming,
or visual design). But, platform affor-
dances don’t dictate how they can be
used. Sometimes we find experts on
Mechanical Turk (for instance, Turkers
speak hundreds of languages [ 1]), and
sometimes we use Upwork without the
customary personal contact with work-
ers via an API [ 2]).
Like many, we have taken inspiration from “The Future of Crowd Work”
by Kittur et al [ 3]. In particular, we
have adopted their idea of creating
“crowd career ladders” and supporting worker learning into our “skill lad-