This website was designed as part of the ANR PORTIC research program. This program studies the navigation and foreign trade flows in the eighteenth century when trade was largely carried out at sea. In addition to gaining a better understanding of these dynamics by cross-referencing two sets of data, a fundamental objective of the program was to find methods to qualify and deal with the uncertainty of the historical data by experimenting with suitable methods of displaying and writing the information.
In this framework, Working Package 5 of the research program involved a series of case studies based on two existing corpora of digitized historical sources: Navigocorpus, a database covering the history of navigation, supervised by Silvia Marzagalli and Christine Plumejeaud; and Toflit18, a database covering France’s foreign trade, supervised by Loïc Charles and Guillaume Daudin.
This website was designed as part of the second phase of the project, following a first phase of digitization to constitute the two databases, intending to take advantage of the work already accomplished and carrying out an in-depth study of significant data subsets. These studies shed new light on themes already studied within the discipline by providing new material and methods by means of the databases developed as part of the program. This initiative raised a number of questions:
- How can two sources be used together when some of their dimensions cannot be measured or are difficult to compare? How can the digitized data from the two databases be cross-referenced to acquire a deeper and more qualitative knowledge about the phenomena provided by historians specialized in the region studied?
- How can the survey and analysis take into account the considerable stratification of transformations, reductions and interpretative decisions that form the chain that connects the initial sources with the data drawn from them?
- How can the uncertainty or incompleteness of the data be taken into account, while proposing a scientifically sound analysis of the phenomena studied?
- How can the standard forms of historical writing –the narrative formed by consecutive sentences, the argument developed paragraph by paragraph– effectively serve the forms of diagrammatic and computational writing involved when creating a website?
- How is it possible to manage interdisciplinary collaboration and the division of expertise involved in this kind of multimodal writing?
These questions were worked out across a six-month period of collaboration that led to the development of the website and will be briefly recounted below.
From a series of questions to a week of investigation
The case study which led to this website was the outcome of intensive collaboration which we call a data sprint and which consists in gathering experts from several disciplines to work together over a short period of time on pre-determined research questions, based on one or more datasets prepared for the occasion.
Ahead of the data sprint, several months were devoted to preparing the data and the research questions. As regards data, the major task for the Navigocorpus and Toflit18 data teams was to clean and refine the data for the regions and periods studied, as the simple fact of planning a specific case study revealed new problems that had remained hidden until an in-depth study was implemented. Another major challenge was to align the two databases, i.e. to match the places and types of products traded or transported as described in each of the databases.
To prepare the initial research questions, we held a series of preliminary meetings based on collaboratively produced online documents that helped us identify pertinent questions with respect to the history of the region and the existing literature, but also with respect to what could or could not be explored and displayed with the data available. This work of adjustment and matching up questions and data ran from October 2020 to April 2021, allowing us to identify the most relevant exploration activities for submission to the workshop participants, and to avoid the pitfalls that would have led us to unsolvable questions
Following the preparation work we drafted a scientific questioning document that described the research questions to be examined and was shared with all the participants during a week-long data sprint. At the beginning of the data sprint, the case study was divided into two complementary lines of inquiry and two teams were formed with a balanced composition to allow interdisciplinary collaboration:
- The first team was appointed to study the dynamism of the Poitou region compared to France and the world in general (with the hypothesis of a decline compared to the rest of France during the eighteenth century).
- The second team was in charge of studying the relationships of complementarity and domination between ports in the region, with the aim, notably, of distinguishing the trade flows and commercial networks associated with them.
Background material was then prepared to help the experts quickly grasp the multiple dimensions of the workshop, and preliminary surveys were conducted to solve the most obvious questions and make the most of the wide panel of expertise available during the short and very dense data sprint.
Unusual research conditions
The data sprint was scheduled for early April 2021, and this coincided with a strong upsurge in the COVID-19 epidemic in France in general and in Paris in particular, which forced us to hold the workshop online. An online data sprint meant that we were deprived of the invaluable contribution of informal discussions and other possibilities for displaying and circulating information and occupying different spaces when working in the same physical environment. The situation was compounded by the French government’s sudden decision to close kindergartens and primary schools that week, adding a heavy extra-professional burden on a substantial number of participants.
The data sprint nevertheless took place, and we decided to consider the unusual circumstances not only as constraints requiring a drastic readjustment of our expectations, but also as opportunities for invention and experimentation in finding new ways of proceeding that departed from the methods initially planned.
Improvisation and adaptation
During the data sprint, the diversity of the participants’ competencies led us to reexamine the division of tasks. As a result, each person was given the opportunity to suggest research topics, manipulate the data, and examine the treatment and formatting of the results. The workshop also provided an opportunity to experiment with a large number of technologies used during the investigation, including the Jupyter notebook, kepler.gl, vega, and matplotlib. However, we were occasionally hindered by « Zoom » conditions, and specialized discussions (from a technical or historical perspective, for example) sometimes required groups to form once again along disciplinary lines with moments of transversal collaboration.
The data sprint format also implied an experimental, sometimes chaotic dimension, amplified by online communication problems. Our working days were therefore organized around a lighter schedule and in a format that sought to vary the rhythms and scales of interaction: a plenary update in the morning was followed by work in two groups that quickly divided into pairs or even individual work.
From a technical point of view, we opted for a permanent Zoom room open throughout the week divided into breakout rooms for work in smaller groups. We also filled a well-organized google drive folder throughout the week, in which all the activities were archived and recorded.
Methodological procedures were also implemented to facilitate the process of mutual understanding. Some of the participants, for example, acted as facilitators and were in charge of maintaining communication between the different groups by constantly circulating between the different work Zoom rooms. We also stressed the importance of documentation by including daily writing of « reports » completed as the day progressed, and by appointing « scribes » who were in charge of these reports and responsible for keeping as many traces as possible to allow asynchronous exchanges of information. These reports were eventually phased out, as they were too burdensome given the weight of the other activities. We also gradually cut down video conferencing time to reduce the fatigue associated with online communication.
Data sprint work sessions can sometimes cause a « bubble effect » in which each team can quickly become disconnected from the work accomplished by the others, leading to dead ends, duplicated tasks, and « blind spots ». This problem is amplified when working online. To meet this challenge, we scheduled progress update meetings at regular times during the day to share information on different scales: in small groups, in teams, and all together. In addition, informal visits and « Zoom surfing » made it possible to circulate ideas and methods to a certain extent.
One of the advantages of the data sprint format is that it assigns a specific time for interdisciplinary work. However, because work during this short time is intense, the expectation is that results equal to the efforts invested will be produced. In normal times, the data sprint is a time for exploration and investigation but also a time for generating results. However, in our case this principle was challenged in two ways: by online working conditions, which made us lower our expectations, but also by the very principle and objectives of the PORTIC data sprint which aimed at publication of finalized websites contributing in a relatively precise way to very specific case studies (and not data mining reports as is often the case with other data sprints). Faced with these challenges, we had to prolong our work after the workshop and change the time frame to achieve a publishable result.
The website, a pivotal tool for multimodal writing and defining a technical and methodological infrastructure
Based on the work completed during the data sprint, we began a second deeper phase of the project. In terms of content, the research topics initially selected were gradually reformulated. This work started at the end of the data sprint week, following the completion of a summary document based on the visuals produced during the week. Cécile Asselin then carried on summarizing and tidyng up the productions of the week.
In order to progressively draft a publishable document from the data sprint explorations, creating a story and arguments based on the two databases, we combined the three types of « writing » needed to develop the future website: computational writing for the new data files that grew out of data sprint investigations, diagrammatic writing for the visuals, and finally, of course, prose for the narrative and historical argument based on the case study.
We decided to divide the website into three similarly-structured sections: a « main visual » presenting the section’s main argument together with text based on the visual and expanding outwards, combined with « secondary visuals ». This visual and rhetorical organization induced an iterative work method similar to « ping pong » which was conducted in three stages:
- Immediately after the data sprint, historians first worked on clarifying the arguments of each of the three sections (April-May 2021).
- We then organized two intensive days to design the main visual proposals (at the end of May 2021) with a small team of engineers and designers (Cécile Asselin, Robin de Mourat, Christine Plumejeaud, Paul Girard, and Maxime Zoffoli).
- Finally, the historians returned to the text which had been enriched and framed by the information proposed in the visuals. As they clarified the text, the need for additional visuals arose and these were gradually produced to fit the text (June to September 2021).
This framework allowed us to progressively develop a work infrastructure adapted to the continuous development of the website. The website thus not only served as the goal but also as the point of contact between the different activities that arose after the data sprint workshop.
The website presented to the public was created using an automated protocol enabling daily updates as written content and further website developements emerged, which meant that progress was made jointly on its structure, graphics and the various contents (see its github directory for more information). The automated protocol consisted more specifically in retrieving content written by the historians and stored in a Google document, and data from the Navigocorpus and Toflit 18 databases in their respective digital spaces, and finally producing web-optimized data before updating the website.
This customized method, instead of using a CMS (such as Wordpress) or totally DIY content integration, allowed us to multiply iterations on the data, visuals, and text and develop a website by iteration with consistent and readable content. It also allowed us to avoid a website production approach that would have been too sequential with a first phase of « design » followed by a second phase of « development », and instead left room for discovery and dialogue with the different materials as they were being developed.
Thus, the method of preparing and assembling the website, which nevertheless required some optimization with the automation of certain tasks, enabled a more organic dialogue to develop between the different types of writing in this document: by simultaneously producing the website, the visuals, and the textual content to go with them, we tried to make each of these activities benefit from the others and be the source of mutual invention or confirmation, rather than consider them as specialized or coordinated tasks in a process of subordination (from history to technique, or from visual to text, or from argument to illustration, etc.).
A website designed to connect with other activities and audiences
This website is intended to be both a stand-alone product that presents a scientific narrative of the history of the La Rochelle region in the eighteenth century, and a building stone for a scientific and didactic construction yet to come. Its target audiences include history students, secondary history teachers, and advanced amateurs passionate about the Poitou and Charentes regions. However, the website can also be of interest to two further complementary audiences:
- Professionals working in historical outreach (for either an expert or general audience) who can use some of the visuals for initiatives aimed at a wider audience.
- Specialized historians who can cite these visuals in conventional publications in journals or monographs and use them for future investigations.
To this end, we have ensured that each visual can be separated from the narrative and can be accessed via the atlas page, and more technically we have provided each visual with its own URL.
Furthermore, the website source code, shared under a free license, was developed and documented to allow it to be used –in whole or in part– for future projects. Useful elements can serve for the next case studies of the PORTIC program but also for other similar projects. Finally, it should be noted that all the data used and the scripts that generated them from the entire Toflit18 and Navigocorpus databases (both freely accessible) are also included in the website source code, which means that our visuals and results can be fully reproduced.
This website is the result of a process of collective investigation and writing that brought together people from several disciplines: history, economics, computer science, geomatics, and graphic design.
The website texts were written by Silvia Marzagalli (Professor of Modern History, Université Côte d’Azur - CMMC, Principal Investigator of the PORTIC program), Loïc Charles (Professor of Economics, INED, Université Paris 8 Vincennes - LED), Guillaume Daudin (Professor of Economics, Université Paris-Dauphine - LEDa-DIAL), and Thierry Sauzeau (Professor of Modern History, Université de Poitiers - Criham).
The entire process of collaboration was led and managed by Cécile Asselin, an engineer intern who worked full-time for six months on the methodological and technical supervision of the entire case study, with the assistance of Robin de Mourat (research designer, Sciences Po medialab - scientific co-leader of the PORTIC program in charge of case studies).
The visuals were designed and developed by Cécile Asselin, Robin de Mourat, Christine Plumejeaud (geomatician, LIENSs laboratory - scientific co-leader of the PORTIC program), Paul Girard (engineer specialized in digital humanities - Ouestware), Maxime Zoffoli (graphic designer specialized in infographics), Géraldine Geoffroy (data librarian, Université Côte d’Azur).
The basemaps were designed by Christine Plumejeaud.
The data were prepared and checked by the Navigocorpus team (Christine Plumejeaud, Silvia Marzagalli), the Toflit18 team (Guillaume Daudin, Loïc Charles) and the medialab team (Cécile Asselin, and Robin de Mourat).
The website is also the work of those who took part in the data sprint: the engineers in digital methods and designers of the medialab laboratory of Sciences Po Béatrice Mazoyer, Benjamin Ooghe-Tabanou, Guillaume Plique, Héloïse Théro, and Pauline Gourlet; the researchers Pierrick Pourchasse (Professor Emeritus of Modern History, Université de Bretagne Occidentale, Brest), and Alain Bouju (HDR Associate Professor in computer science, Université de La Rochelle).
A special thank to Yvonne van der Does and to Alexia Grosjean for their professionalism and their efficacy in assisting us with the English translation.