There has been an increased interest, focus, and analysis
of storytelling this year in the ﬁeld of data visualisation
“This has unquestionably been the year of ‘storytelling’” -
“Storytelling with data visualization is still very much in its “Wild
West” phase” - Nick Diakopoulos
“Storytelling: The Next Step for Visualization” - Robert Kosara
“Storytelling has been one of the big buzzwords in data
visualization the last year” - Moritz Stefaner
“There is more to information visualization than punchlines.” -
....But there are also some people questioning it
With many great examples of stories being told using
Snow Fall by the New York Times is a stand out.
Some of these broadly include “visualisations” with
varying levels of “interactiveness”
Interactive story telling is nothing new. Interactive ﬁction,
role playing games, etc. have all existed for many many
years. Indeed, during the 1970 and 80s researchers
explored the topic. Chris Crawford’s book on the topic is
almost a decade old.
Quokka from the late 1990s was well ahead of its time and
arguably was a better, more scalable example of rich,
interactive story telling, in real time, than one off pieces
taking 7 months to build within a major news organisation.
As an aside Quokka’s creative director was Eric Rodenbeck
who when on to found Stamen Design.
This presentation is not about story telling, but rather the
scalability of story telling and more speciﬁcally annotated
journalism in the ﬁeld of data visualisation.
Today, most visualisations with a journalism or story telling
are developed with a static, somewhat historical, set of
Within these data visualisations, the opportunity exists for
careful analysis and considered commentary to be created
that augments the visualisations, annotating salient points,
and assisting the viewer in interpreting and understanding
what they are looking at.
The annotation layer of graphics is “the most important thing
we do.” - Amanda Cox, NYT graphics editor
But what happens when data visualisations are built that
are going to be updated with data for the next 1, 2 or 6
years, when the capacity for expert analysis every month
across hundreds or thousands of schools, hospitals, train
stations etc. is not possible.
Could robotic journalism be used to auto-annotate data
visualisations through their lifespan.
Automated journalism, robot journalism, narrative science
are all terms that describe “automatically” generating
online articles about particular topics.
Taking statistical and raw data, the basic building blocks
of most visualisations, smart “expert in a box” algorithms
interpret and “analyse” the data, extracting salient points
that are then fed into a language generator that outputs a
brief, human-like article. Like spam comments only mathy.
In some ways it is the natural extension of the googl-ised,
twitter-ised world that exists today in which “trending”
stories are given equal or more weight on news sites than
deep journalism. The ultimate in snackable content.
Companies exist today that use artiﬁcial intelligence style
algorithms to transform data into stories and insights.
StatSheet takes sports scorecards, analyses them looking
for patterns, isolates key points and produces an article
highlighting these insights.
When you think about it, this seems entirely plausible
given the robotic, repetitive and repeatable nature of
Narrative Science does a similar thing but across the more
“serious” topics of ﬁnancial services and information
The size of the problem domain is already huge and only
growing, especially as demand and expectations for data
visualisations to outlive the year grow.
“It seems to me that annotations, as an integral part of a
visualization design, have received somewhat little
attention in comparison to other components of a visual
representation” - Enrico Bertini
We’ve all heard Borel’s imagery of monkeys in a room
typing Shakespeare, but in this new world of Big Data, is it
possible that automated journalism could be used to
annotate and enhance data visualisation.
To date, few examples exist of successful implementations
of this style of annotation in data visualisation, especially
beyond the pre-school level “You have done 45% better
this year”, “Your performance is good”. But the gap is
There have been some initial efforts at automatic “labeling” but
much less on humanised text.
Schooloscope by Berg is a great example of beautiful
aesthetics and design along with an humanised textual
annotation that summarises what the visualisation is
So, what do you think? Do you have examples of
automated annotation in data visualisation? Is it the
future? If so, utopian or dystopian?
We welcome your thoughts.
This is a slide pack was prepared to help develop thoughts
around automated storytelling and journalism in data
visualisation. The thoughts and points are not ﬁxed in stone.