Chris Wiggins - Data Scientists at Work - page 14

and even nanoscience be born as a field in the time that I've been a prac-
ticing academic. My first research project in the 1980s was in chaos, which
at that time was being born as a new field. There's a famous book on this
by James Gleick, at that time writing for The New York Times, called Chaos:
Making a New Science . 6 It's not that new fields aren't created in academia. It's
just that it's so damn slow compared to the pace of the real world, which
I think is really for the best. There are young people's futures at stake, so I
think it's actually not so bad.
So I think the future of data science is for it to become part of academia,
which means a vigorous, contentious dialog among different universities about
what is really data science. You're already starting to see work in this direc-
tion. For instance, at Columbia, a colleague of mine, named Matt Jones, who's
an historian, is writing a book about the history of machine learning and data
science. So you're already starting to see people appreciate that data science
wasn't actually created from a vacuum in 2008. Intellectually, the things that we
call data science had already been sort of realized—that is, that there was a
gap between statistics and machine learning, that there was sort of something
else there. So I think there will be a greater appreciation for history.
Part of what happens when a field becomes an academic field is that three
main things occur—an academic canon is set, a credentialing process is initi-
ated, and historical study provides the context of the field. An academic canon
is the set of classes that we believe are the core intellectual elements of the
field. The credentialing process, which is another separate function from aca-
demia, which can be unbundled, is initiated so you can get master's and PhD
degrees. Lastly, historical study occurs to appreciate the context: Where did
these ideas come from?
As the names and phrases people use become more meaningful, then you
get the possibility of specialization, because what we have now is that when
people say “data science” they could mean many things. They could mean data
visualization, data engineering, data science, machine learning, or something
else. As the phrases themselves become used more carefully, then I think
you'll get to see much more productive specialization of teams. You can't have
a football team where everybody says, “I'm the placekicker.” Somebody needs
to be the placekicker, somebody needs to be the holder, and somebody needs
to be the linebacker. And as people start to specialize, then you can pass. You
can have meaningful collaborations with people because people know their
roles and know what “mission accomplished” looks like. Right now, I think it's
still up for grabs what a win in data science really looks like.
6 James Gleick, Chaos: Making a New Science (Viking, 1987).