Rough summary for reporting back
15 of us in the methods session, a range of technologists, historians, social scientists, literature, geography.
For a methods group, we spent a lot of time discussing tools - tools as instantiations of methods. Humanists are starting to understand the implications of using different methods, how do we help people understand the implications of choosing different tools?
We discussed our understanding of 'methods' from our different disciplines: social scientists, geographers, historians. What's considered acceptable methods varies by field - framing work in appropriate strategies. Different fields also have different ways of representing uncertainty - is a line on a map a possible path or a definitive statement?
A lot of discussion centred around the question of literacy about tools - who knows when to read 'with a grain of salt', to understand lines or directions on maps as generic paths rather than actual statements. The more complex the visualisation we come up with the more the complexity of understanding it becomes a barrier. We need to guide people to understand what they're seeing - what's the digital humanities version of close reading for tools?
Returned to the issue of the siren's lure of the shiny tool. Both in terms of tools developed for related fields like the social sciences that look promising but fail to cope with the messiness of humanities data, but also because they're problematic as analytic tools because they can be misleading e.g. showing term counts without considering variant terms for the same concept. We didn't use this term at the time, but we discussed the misleading 'truthiness' of visualisations made with tools that don't convey and problematise the data underneath and the process of creating the final product.
Lots of discussion about how do we teach humanists to interrogate the original purpose of the tools? How can transparency about methods and data selection be built into tools? We talked about word clouds as an example of a tool that needs to be problematised - some entry-level versions are disconnected from the methods that inform other tools that look almost exactly the same.
As always, some discussion of the messiness of humanities data - it's both glorious and a problem for digital tools.
We also discussed whether tools need to be developed with a research question, or a class of research questions, in mind.
We reaffirmed the role of digital humanities as a space for people interested in problematising the instruments; the need to articulate methods and to understand how tools are operating to speak fully about your research methodology.
We thought method was a bridge between different experts (technologists, etc) and humanists; place for common understanding; people who don't want to get their hands dirty in each other's field there, they can meet on the bridge.
Method as good place to meet - the important steps should be transparent for everyone, not just one professor. Description of programme, steps behind it - should not be black box. Must be able to explain methods behind tools.
Discussed the notion of participatory design and reading... asking your reader to be aware of the nature of the tools in their own interpretive process.
How do we get to have participatory publications or 'interactive scholarly works' in academia? How are they reviewed, credited, sustained? And what can we learn from educators and museum people about active engagement and reading?
We finished on a positive note with mention of the DH Commons - find a technologist or a humanist to collaborate with, but also to find reviewers for digital projects.
[Copied across on Wednesday morning]
NeDiMAH space/time 'methods' breakout session notes.
Dodgy rough summary at http://spacetimewg.pbworks.com/w/page/55467008/%27Methods%27%20breakout%20session - once everyone's had a chance to check what's here, I'll copy it across to the wiki.
Please add your name, institution and project/interests here!
[Not everyone's names have been added]
Jacqueline Wernimont, Scripps College, regional digital humanities center, spatial and temporal methods across disciplines.
John Theibault, Richard Stockton College of New Jersey, college digital humanities center, spatial and temporal methods in history and visualization strategies.
Olaf Berg, Max Planck Institute for European Legal History, Frankfurt/M, Germany
Representation of geografical and time information found in sources digitzied at our institute. Especially text from the school of Salamanca to trace the development of moral and legal concepts over time and space (discovery and appropiation of what is know Latin America). And representing a collection of fotos of places and buildings related to legal practices.
Fredrik Palm, HUMlab, UmeƄ University
- Relational browsing of data where information is stored according to the model of the content owner. Allows seeing connection between attribute, filtering informaiton, aggregating information, bookmarking etc Examples http://ships.ddb.umu.se, http://www.sead.se
- Textometrica: human and computer supported proceses to work with co-occurency in reading and analysing text for internet culture studies. http://textometrica.humlab.umu.se
Daniela Schulz, Cologne University, Cologne Center for eHumanities (CCeH); various projects, e.g. medieval manuscripts; early modern travelogue (http://diglib.hab.de/edoc/ed000081/startx.htm); scholarly digital editions in general; looking for new approaches/ideas; "fancy" but above all valuable forms of visualizing/representing time and space
Christoph Kudella, University College Cork, Digital Arts & Humanities PhD Programme:
Research on early modern networks of epistolary exchange, esp. on Erasmus of Rotterdam; Usage of Social Network Analysis and Topic Modelling
What do 'methods' mean in your field?
What's considered acceptable methods varies by field - framing work in appropriate strategies
Discourse analysis
Messiness...
Shiny tool... when bring in messy historical data, find that it doesn't fit into the template. Tools should have ways of bringing in where data is from...
Develop tools that allow for a high level of flexibility.
Early versions of ArcGIS defined space in inflexible ways...
Tools developed for social science deal with very precise data... what happens when you don't know the exact location of a site, everything ends up being in the middle of Sweden. The tools affect the way you think.
How often do you really need to query by geography as opposed to querying by name? [But what about the ability to query across an area regardless of the level of precision - you could do it with gazetteers but is it easier to do with spatial queries?]
How many tools are made for humanists?
Gethin: not being able to create a shape file for the data they had - places that were areas rather than points, things without places attached to them, etc. And for time - events that were dated to centuries, or relative to other events or periods (reign of emperor etc) - those events didn't fit into the system.
Look at interpretations that people made previously e.g. in publications. People would disregard things that wouldn't fit into the system even then.
Karl: lower-case gis and upper-case GIS. Lower-case gis... trying to get a handle on the requirements of historical scholars. It's a slow process, projects that are addressing it are one-off, happening slowly. Saw last week a paper on modelling fuzzy intervals - but not yet implemented in any GIS. GIScience community has an interest in representing fuzzy dynamics for humanists.
Jacqui - TEI can represent fuzziness about places, levels of certainty; can use those in structured way in doc headers or drill-down within documents...
Mia: different fields have different ways of representing uncertainty - is a line on a map a possible path or a definitive statement?
John: who knows when to read 'with a grain of salt', to understand lines or directions on maps as generic paths rather than actual statements. The more complex the vis'n we come up with the more the complexity of understanding it becomes a barrier. Need to guide people to understand what they're seeing. Open question: at what point does that become second nature? Now network diagrams are tough - not easy to understand for humanists.
Network maps tempting, give you feeling that it's showing something true. Much network software is only designed for one node type - if you add others, you still get pretty pictures but they don't make sense. If try to adapt techniques from social sciences, you need to understand why tools were made or write your own algorithm. Shiny tools as dangerously tempting. Need to go back to basics and think about data modelling and how information is transformed into data - processes of selection, curation, etc. Need to talk more about that, less about which shiny interface to use.
Mia: how do we teach humanists to interrogate the original purpose of the tools? Or do we just need to problematise the use of these tools and question how they were chosen as part of the review process?
Kimberley - referring back to assemblage, emergence from this morning. More on process - deliberation and transparency re how data is selected, curated, represented. JB Harvey essay, maps, knowledge, power - drawing attention to the emergence of these visualisations.
Frederick - transparency about what excluding... what are the steps that you need to show the users? GraphViz... if understand enough of what software is doing, can explain it; if you don't understand it, how would you do that? What are the critical steps (where decisions are made)? Combine different tools to support research methods.
'Black box' - need to see parts of the process to understand the critical decisions
Mia: need to understand that technologists are making decisions when creating tools - either in the affordances of what the tool does or doesn't let you do, or in what's emphasised in an interface.
After the break... we've discussed tools as instantiations of methods. Humanists are starting to understand the implications of using different methods, how do we help people understand the implications of choosing different tools?
What about the relationship between method and theory?
Jacque: articulating the different problems with tools...
Start with word clouds...
Problematic as analytic tools because they are misleading e.g. showing term counts without considering variant terms for the same concept
Frederik - difference between analytic tools and descriptive tools...
Karl: tag clouds as pet peeve... people assume size is term count, not statistical. Visualisation that's very attractive (like maps or graphs) that lead to assumptions that are almost impossible to dissuade, have to wait for literacy on those methods to catch up.
Jacqui: difference between tool and how it's used out of the box... you can't get people to unlearn what they've learned about the 'out of the box' instance. Explaning nuances in use.
Karl: salience measures... problems with map reading are age-old, why would we expect other visualisations to be different?
Mia: is there value in listing things you'd want a humanist to be aware of for different visualisation types?
Karl: dedicated scholars will demand to know something about the underlying algorithms.
... What can we learn from visual studies?
Mia: is a basic text box
Jacqui - only gets a rich sense of what's behind a visualisation if interacts with it. Need time to manipulate it. Adds to the time pressure.
Mia: can we make analogies with close reading that would help people value the process of spending time getting to understand a visualisation?
John: how much guidance do we need to give to people? Hard to know if we don't know who's using and viewing these tools.
Reception of Culturenomics... initial reception was excitement, took a while for a more critical engagement with it. Difference between people who get in on the initial shiny toy excitement and those who stick around and use it. But are we even attracting the newbies?
Frederik - bringing people into the design process to help increase literacy (?).
Gethin: is it a question of audience - is DH for humanists generally or just for DHists?
Frederik -
Gethin - tools often developed without a research question. Jacqui - that's a real problem, because then people come along and try to use it for their research questions.
Do tools always need to be able to support a research question, even if not necessarily a specific one? Yes.
Jacqui - not everyone in a field is interested in problematising the tools of the field. DH ought to be a space for people interested in problematising the instruments.
How do you have a digital research question and a humanities research question?
Karl: seconding, thirding that. In scheming for data models that will be useful...
How does one do both the general tool and the specific project? How far can one get without someone having a specific question they want to explore.
[Mia, wondering - do tools for digital humanists work best when they start by scratching someone's particular itch? Or does that make the design too closely entwined around one person's way of thinking?]
Olaf: as historian, finding sources and mining data from them is only part of the work of a historian - the other is creating relations, a story about them... history that is produced is constrained by tools. Tools that pretend to be definitive are problematic (not sure paraphrased that correctly)... creating relations that are not deducible - created by the historian.
Is the scholar abdicating their role in creating the story?
Christoph: with new tools can work with huge amounts of text that wasn't previously manageable by one person. Can now ask new questions. Also makes visible where previous histories have left out material that doesn't support their story. The tools change how we write history.
Histories of famous people are master narratives that follow certain patterns. But now you can point to the empirical data you've collected and seeing how those old histories selected sources.
Mia: what do we want to take back to the main group?
Jacqui: discussion of articulating methods; need to understand how tools are operating to speak fully about research methodology.
Gethin: method seems to be a bridge between different experts and humanists; place for common understanding; people who don't want to get their hands dirty in each other's field there, they can meet on the bridge.
Method as good place to meet - the important steps should be transparent for everyone, not just one professor. Description of programme, steps behind it - should not be black box. Must be able to explain methods behind tools.
Mia: should tools carry the context of their creation and the underlying methodology with them?
John: tools undermine narrative, crafting of a story as primary method of exposition for humanists. Ludic element, playing with tools is a type of transparency. As a way of inviting the reader in, moral obligation to make the tools open so that people can explore them; write to say 'follow me down this path and you'll find fun things to do' rather than 'follow me to be entertained'. The user has the power to move things in the way that they want.
Jacqui: the notion of participatory design and reading... asking your reader to be aware of the nature of the tools in their own interpretive process.
Mia: how do we get to have participatory publications in academia? And what can we learn from educators and museum people about active engagement?
Karl: 'interactive scholarly works'
Playing with publications as way to engage people with tools and interrogate methods.
(New forms of publication that require active, interactive reading to unlock the scholarly arguments, narratives within?)
Karl: no venue for review of interactive scholarly projects.
Christoph: in Ireland, new PhDs in digital arts and humanities, still no decision on how one evaluates the digital part of their projects. Digital stuff can't get published in the thesis in the traditional sense. How do the printed thesis and the digital stuff go together? It's a new form of scholarship - how is it judged?
DH Commons - find a technologist or a humanist to collaborate with, but also to find reviewers for digital projects.
Comments (0)
You don't have permission to comment on this page.