John Foreman - Data Scientists at Work - page 118

or external user is the result of the thought process of someone already having
thought through and processed the problem in a particular way, which means we
have to figure out if the way they're posing the problem is the actual problem.
Going back to the elevator bank example mentioned earlier, maybe the
problem is posed as, “I need you to build an optimization to speed up my
elevators,” when, in fact, the problem is, “I want more people to get on the
elevators in any way possible.” The solution would be very different depending
on which problem you solved. Once it's understood that people just need to
get through, the right solution is to put mirrors in there so that people pack
in tighter. So you have to be very careful that, when people request features
or tools, you actually solve their underlying problem.
Furthermore, part of being a company in a very competitive space is coming
up with stuff maybe people aren't asking for yet, but as soon as they see it,
they're going be like, “Oh, yeah, that's much better.” So there's also this notion
of creativity. We have to make sure to understand where the space is going,
not only in our industry, but also watching what's happening elsewhere. This
way we can learn those metaphors and think about how to apply them to
our world. So that's where a little bit of the black arts for data science comes
in—with this creative piece. It's not just about understanding the techniques,
the data, and the technologies available to me—that's the base line. It's about
taking the time to dream and get creative, instead of waiting for people to tell
me what to do.
So there's the notion of really actively listening to figure out what people
are actually saying, and there's also the notion of hearing what's going on
in the world and understanding how it affects my world. The French word
for it is bricolage , which is taking these disparate things and combining them
into something new. In my case, it's about thinking how people have solved
problems in other industries and how that applies to what we are working
on. Sometimes you go down some dead ends, where you realize that for our
customers and what we're trying to do, this has no applications. Other times
you strike gold and you realize, oh man, if we did what they're doing, if we
could actually create this, it would be awesome for our customers.
Gutierrez: How do you think about the toolset, and what technologies you
are currently using?
Foreman: I'm very conservative in the way I think about tools. What has
struck me in the data science world is that the tools are wagging the analytics,
which is to say that a lot of organizations get attracted to new, bright, shiny
tools rather than thinking of what problem they are trying to solve. They say,
“Ooh, we really want to use this tool,” and then they come up with a reason
to use it. Vendors probably drive a part of that. Vendors are there to sell you a
tool for a problem you may or may not have yet, and they're very good at con-
vincing you that you need it whether you actually need it or not. I frequently