Grapl was custom-designed to host the particular research undertaken by Stanford Professor Thomas S. Mullaney that is presented in The Chinese Deathscape; it was later expanded to include the work of colleagues from around the world, including Christian Henriot of Aix-Marseille Université and Jeffrey Snyder-Reinke of the College of Idaho. These scholars were inspired not only by the phenomenon of grave relocation in modern China, but also by the possibilities offered by the deeply integrated, data-enhanced, interactive map-based publication platform that we had developed in collaboration with Mullaney.
On Dev Relocation in Contemporary Digital Humanities
Grapl is largely the creation of digital humanities software developer (and an author of this colophon essay) David McClure, who came from the University of Virginia Scholars’ Lab in 2014 to join Stanford University and the CIDR team.  While at Scholars’ Lab, McClure worked as lead developer on the popular and versatile map-based suite of tools for Omeka (itself developed at the George Mason University Center for History and New Media) called Neatline. At the level of interaction design and intent, Grapl flows directly out of the work that McClure did on Neatline between 2011 and 2014.
Neatline provides a general-purpose mapping and exhibit application designed to meet the generic presentational needs of a wide variety of digital humanities projects. But in actual use, these scholarly projects have often included some sort of long-form narrative that was intended to accompany the map. Most commonly this accompanying text has been something like a journal article or a book chapter; at other times it has been something less formal, like a blog post. But rarely, if ever, has Neatline been used as a companion for a book-length text.
The combination of images, maps, and longer-form text always leads to some awkwardness: How best to integrate map-based digital exhibits with long-form narrative? In the past, most authors have simply published the two entities separately, using some sort of project overview page that links to each component individually.
But this often seems like a subpar solution. The division between the text (with its extended argument) and the map (with its geocoded data or primary source materials) is awkward: rarely is it feasible for the reader to read the prose narrative and engage with the interactively mapped digital component of the project at the same time. Especially in cases where the textual portion of a project makes close reference to the interactive map, the result is a fragmented, disconnected reading experience. Hypertext links, for all their power, are essentially paths to elsewhere—not integrations. Popup windows, for all their immediacy, obscure even as they illustrate.
What is really needed is actually very modest: simply some way to “quote” the map more directly in a text, in much the same way that a literary scholar, for example, can quote a novel or a poem by embedding its text directly within that of the scholarly argument. Such quotation is a fundamental building block of scholarly communication—a kind of scholarly primitive that cuts across many forms of interpretive work in the humanities and social sciences, and one of the most basic ways that we draw connections between our arguments and their evidence; it is the essential, vital link between primary and secondary sources.
But if the object of study is a map (or is something mapped), rather than another text, there has been no equivalent way to seamlessly include, reference, or even just directly point to it from the text. One can describe the map (and its features), or summarize it, or include snippets from it, in the way a film scholar might narrate a plot point or reproduce a still image. But this still feels less than ideal: not only is the extent of the “quoted” object greatly reduced (a reduction that may be perfectly acceptable in textual quotation, after all); more importantly, its dimensionality and thus even its very essence are reduced. Its visual aspects are impoverished or elided; its temporal aspects are frozen in a snapshot; its exploratory possibilities are truncated. Thus, in the abstract, our principal task in creating Grapl was to integrate fluidly the two-dimensional space of a map (or even its three-dimensional space, if we count the interactive, temporal component) into the one-dimensional axis of a textual narrative: to allow fully fledged, rich quotation of one medium within another, differently dimensioned one.
These challenges eventually led to the development of an add-on feature for Neatline called Neatline Text, which made it possible, in a very simple way, to connect sections of a text document interactively with objects plotted on the map in a Neatline exhibit. Neatline Text was not especially sophisticated, and rather more “bolted on” than well-integrated into Neatline; in any case, it was intended more as a proof of concept than as a fully formed application.
Grappling with Text and Space
Beyond these simple, one-to-one connections, though, Grapl also enables experimentation with higher-level interactions between the text and the map. To augment the granular linkages between individual word sequences and spatial locations, we wanted to provide a less granular, more schematic connection between the narrative and broader “regions” on the map. In Grapl, larger sections of the narrative—for example, a group of three to four paragraphs, or even an entire section of an essay—can be linked to entire, author-defined regions of the map, which are displayed as boxes subtly overlaid onto the map, with a title that often mirrors the section title in the text and is visibly connected with a dynamically drawn line. These two-dimensional section markers provide a high-level “railing” that guides the reader both as she moves through the text narrative and as she surveys the map. When the text is scrolled to a different section, leaving the map focused elsewhere than on the spatial region corresponding to the current section of text, a clear but unobtrusive tooltip is displayed prompting the reader to click and transition the map to the appropriate region. And, if the map is moved far enough away from the region assigned to the currently visible section of the text, the reader will again be prompted to “re-sync” the map when the cursor next hovers over the text.
This set of features speaks to a core goal of Grapl: to strike as even a balance as possible between giving the reader freedom to explore a data-rich spatial environment at will, and providing a clear, simple, recommended path through the spatial and data content which is linked clearly, but subtly, to the content and organization of the accompanying textual narrative. Grapl always tries to make it easy to follow the recommended path—by clicking either on linked brief spans of text or on larger-scale sections of the text as they come into view—but, unlike some other spatial reading platforms (e.g., those designed primarily for journalists), it doesn’t enforce a specific path through the map by automatically updating the map state according the scroll offset of the text. Grapl’s goal is to make it easy both to follow the recommended route and to leave the marked path for more personalized exploration; it also, of course, enables the reader easily to pick up the narrative trail again.
A Threefold Way: Data, Map, and Long-Form Text
Unlike some other spatial reading platforms, Grapl also tries to avoid compromising the structural integrity of the narrative text. Many digital reading interfaces break up or fragment the text in different ways—for example, displaying it as a series of isolated paragraphs that scroll into view under various conditions. In a way that we feel is more appropriate for an academic context, Grapl instead instantiates the view that regular, unbroken, long-format text is still an ideal technology for presenting complex, non-trivial scholarly arguments; thus it attempts only to enhance and integrate the text with the interactive map, rather than to change wholesale the experience of reading a scholarly narrative.
By the same token, Grapl seeks to take greater advantage not only of the map per se, but also—and particularly—of mapped data: the richly informative set of data points that serve not just as supporting evidence for the scholarly argument, but also, in a sense, represent the core of the project as a computational undertaking. In the case of The Chinese Deathscape, of course, this is data about grave relocations: their scope (counted in numbers of graves), the dates they occurred, and, most crucially, their locations—as illustrated in this example from Mullaney’s essay: a total of 2.5 million grave removals during 2012 in dozens of townships in Henan Province. This rich subset of data, comprising a large number of mass grave removals across the province, and documented in individual notices (which can all be seen in the “Graves nearby” section of each record), is mapped and visualized in a manner that highlights its significance and scope; its interactivity leads to more exploration, beyond a simple text with a static illustration, which would serve mainly to tidy up the “2.5 million grave removals” statement, and greatly lessen its impact.
Although Grapl in its current form is tightly bound to its subject matter and geographic focus (which will be apparent in many aspects of the illustrations used for this very essay), it is our hope that the software might later serve for other sets of geospatial, temporal, and otherwise quantifiable and computable data points (though note the discussion below of why we can’t rightly call Grapl a “platform”). Unlike some map-based storytelling platforms that use maps primarily as illustrations of a text, or those that present maps primarily as exhibit pieces and text snippets primarily as captions, Grapl is intended as a unified, coherent threefold path linking long-form scholarly text, data-driven maps, and richly mapped data—a framework for the publication of both texts and datasets, held tightly together by the spatial logic of the map.
Cycles and Recycles
In an ideal world, Grapl would be amenable and appropriate not only to the depiction and analysis of Chinese grave removals, but also, for example, to the mapping of Raskolnikov’s crimes, punishments, and perambulations around Saint Petersburg, or to a rich depiction of Leopold Bloom’s odyssey around Dublin through the hours of June 16, 1904. It would be flexible enough to present richly mapped and deeply explorable cradle-to-grave (-to-grave-relocation?) biographies, or to host long-form narratives bolstered with interactive maps that reveal the time and space of any other historical process: of troop movements in battle, of national borders shifting in time, of explorations and crusades and peregrinations. In fact, we believe that it is suitable for any of these uses; we simply haven’t tried any of them!
But we found, as we prepared The Chinese Deathscape for publication and incorporated additional scholarship into Grapl, that we had in fact left some things out: although we initially designed Grapl as a focused, lightweight framework for Mullaney’s rich archive of point data (representing gravesite removals, their numbers and geographies, the notices that served as source data, etc.), we had not made allowances for higher-dimensional geospatial features such as lines and polygons, which, it turned out, were necessary to represent the spatial data required by some of the other contributions in this collective work. That is not necessarily to say that not including support for lines/polygons was a mistake! In principle, it’s generally best to start simple and expand as needed. Going forward, however, the ability to incorporate a wide range of spatial data formats would of course be essential for a complete textual/geospatial platform—but, given time and resource constraints, this first cut of Grapl was designed around one particular data set. We were able to adapt it, post-factum, to interact with and display polygons, but the current solution is, we admit, inelegant and provisional.
The “Graves Platform” as Not Quite a Platform
In many ways, these difficulties reflect a fundamental challenge of trying to build frameworks rather than project-specific solutions, which is what Grapl has (so far) turned out to be. Reusable frameworks need to provide a broad set of features to support diverse use cases—but these use cases can be difficult to anticipate in advance, and large feature sets are often at odds with the engineering imperative to build focused, reliable solutions that do one thing well. Perhaps the best way to manage this tension is to allow specific projects to drive new development in a bottom-up fashion, in the way that successful projects like Omeka have evolved in response to concrete user requirements. In this sense, our hope is that the current pain-points can be taken as guideposts for where to take the project next.
Beyond expanding the range of supported formats and data types, one might also imagine incorporating not a historical or time-based map, but rather a different sort of two-dimensional image—say, a painting, or microscopy, or an architectural design—which can also serve as canvasses for location- (or coordinate-) based data; is no less amenable to long-form scholarly analysis; and no less dependent on “rich quotation of one medium within another” than a map-based historical phenomenon like grave reform. Such applications, while easily imaginable, are certainly not within the realm of Grapl’s current capabilities.
But we’re pleased with the work we’ve done, with the challenges we’ve overcome in doing it, and with the small improvements we’ve tried to make in the context of the already outstanding publication platforms in the digital humanities world. We hope that Grapl will be seen not as a competitor to those platforms (upon which we ourselves certainly still rely, like many others in the digital humanities community), but rather as a set of contributions, both technical and conceptual, to the digital scholarship ecosystem.
Finally, we’re exceedingly happy to have developed Grapl in a long and intense collaboration with such a thoughtful, innovative, and engaged scholar as Tom Mullaney, and in support of such fascinating works of scholarly inquiry—by both him and his coauthors—into the rich and heretofore untapped histories of The Chinese Deathscape.
Source of image at the top of this essay: Stanford University by Arthur Lites. Courtesy of Stanford University Archives.
 Following a nearly identical personnel relocation, the 2017 pre-publication work on Grapl and The Chinese Deathscape was done (in part) by yet another CIDR developer, Scott Bailey, who had likewise come to Stanford from the University of Virginia Scholars’ Lab, where he had also worked on Neatline. Scott joined Javier de la Rosa in this work. See the “People” page for more on the team behind this present work.