So, as a result, each page appears as a list of topics, with an arrow pointing away from the topic, suggesting “click here for more information.” And, sure enough, if you click on the topic, via jQuery’s toggle(), automagically appears the contents of that topic. The result is that I give the user just as much text as the user wants. If a topic is done, click to hide it. Or click on the following topic to see both at once.

There are two obvious problems with this approach, however, and they have been annoying me since the page went live a few months ago.

• URLs are fixed. If I want to point someone to, say, my list of publications in the popular press, I have to say “Go to http://moacir.com/academics and then click on ‘Publications’ and then click on…” It’s a mouthful. I should be able to just say “Go to http://moacir.com/academics/blah blah” and be done with it. That url would give precisely the information I want to give, and it would let the user, then, snoop around the site.
• Uses JavaScript. The toggle is neat, but it requires the use of jQuery. If one goes to my site with JavaScript disabled, the pages are useless.

So the goal here was to solve both problems. I wanted it to be the case that, if one clicked on a topic, the toggle would fire, but it would also change the url in the location bar, so that the new url could be copy-pasted in an email. And this url should change if a second topic was opened so that it would be possible to see both topics open at once. And if a topic was closed, that section of the url had to be removed. And, then, I wanted a way to do all this (though minus the AJAX, of course) without JavaScript.

Turns out it was not so difficult, but it’s certainly not the kind of behavior I have seen on the web before, so, below, I explain what is going on. I’m a dilettante when it comes to coding, so if there are unnecessary redundancies, etc., please let me know. I know that, at the very least, the creation of the showmes array could be done with a def block, but I got a bit lazy.

The only solving these problems is at all possible, it seems, is because my website is not a set of static html files, but, rather, a web application. It’s built with Sinatra, a decision I made initially just for fun. Yet now, to add this functionality, I’m glad it’s a choice I made. If I had static webpages, then I would basically need a different web page for every possible combination of shown topics (or utilize extensive mod_rewrite magic?). For a site like mine, that isn’t so bad, but even at my scale, it’s outrageously inefficient. Sinatra, by being an application framework, lets me catch the GET request sent to the server and manipulate it accordingly. The task, hence, can scale, and I just made everything a whole lot easier for myself.

So here’s the previous setup. Each topic had an id that was some kind of string with subhead at the end, and it was of class trigger. The text of a topic had the same string, but with “body” at the end, and the class was toggle. As a result, the original jQuery was stupidly easy. For each element of class trigger, strip subhead off the end of the id, add body, and, when one clicks on that trigger, toggle the toggle-class div with the new id.

$('.trigger').click(function() {
	var togglediv = $(this).attr('id').replace('subhead', 'body');
	  $('#'+togglediv).toggle();
});

Ridiculously easy. No logic needed, no nothing.

So the first step is to change this code to add to or subtract from the url in the location bar based on whether the topic is being shown or hidden. That is, check if a thing is not visible. If it is not visible, add the name to the url and make it visible. If it is visible, hide it and delete its name from the url.

$('.trigger').click(function() {
	var stateobj = { foo: "bar" }; // Just some data for replaceState…
	var pathname = location.pathname; // the current url after 'moacir.com'
	var togglename = $(this).attr('id').replace('subhead', '');
	var togglediv = togglename + 'body';
	var togglepatt = new RegExp(togglename);
	var trailslash = new RegExp(/\/$/);
	if (!$('#'+togglediv).is(':visible')){
		// is the toggle not visible?
		// if yes, make the url without double slashes:
		if (trailslash.test(pathname)){
			history.replaceState(stateobj, togglename, pathname + togglename);
		}else{
			history.replaceState(stateobj, togglename, pathname + '/' + togglename);
		}
		$('#'+togglediv).show(); // and now we use show() instead of toggle()
	}else{
		// so the toggle is visible. Hide it and delete it from the URL
		pathname = pathname.replace(togglepatt, ''); // erase it!
		pathname = pathname.replace('//', '/'); // and the double slashes!
		history.replaceState(stateobj, '-' + togglename, pathname);
		$('#'+togglediv).hide(); // use hide() instead of toggle()
	}
});

Note that I use the history.replaceState() method. This is so that I can manipulate the location bar without making new GET requests. This is great for creating bookmarkable urls in an AJAXy environment. Because I use history.replaceState() and not history.pushState(), all this clicking around does not change anything if the user clicks on the back button. They will be sent to the previously GETted page.

That’s all the JavaScript necessary, which is great, since I hate using JavaScript. The rest of the work is done in Sinatra.

Once we’ve got this long url with all the shown topics added to it, if we paste the url into a new window, it will break, since the application does not know how to route it. In Sinatra, we’ll handle this by turning everything after the original url into a splat. For now, I’ll be working specifically on http://moacir.com/about/.

The basic route, then, tells Sinatra that we are expecting a splat after /about/ and that the delimiter will be a slash. The splat then becomes an array, called topics. Next we have a hand built array, called subheads, that corresponds to every possible togglename variable from the JavaScript above. Each main route (/about/ and /academics/, for now) will have a different value for this array. The third object will be a hash called showmes. The key in showmes will be one of the subheads, and the value will be either blank or style='display: block;', depending on whether that certain subhead‘s body should be shown or not.

get '/about/*' do
	topics = params[:splat].first.split("/")
	subheads = ["bio", "name"]
	showmes = Hash.new
	for subhead in subheads
		if topics.index(subhead)
			showmes[subhead] = "style='display: block;'"
		else
			showmes[subhead] = ""
		end
	end
	erb :about, :locals => {:showmes => showmes}
end

The penultimate line tells Sinatra to load the about.erb view and to send showmes to it. Now, in that view, I have lines like this:

<h2 id="biosubhead" class="trigger">Bio →</h2>
<div id="biobody" class="toggle" <%= showmes["bio"] %>>

and

<h2 id="namesubhead" class="trigger">Name →</h2>
<div id="namebody" class="toggle" <%= showmes["name"] %>>

It’s a bit clumsy, but it freaking works. Since the CSS loaded style for toggle class divs is display: none, I can pass a blank value for a key in showmes, and that means that div will be hidden.

Of course, now that the routing works as expected, this means that I can now add backwards compatibility for browsers which do not support JavaScript. In the original code, there are no anchor tags. The JavaScript toggles based on whether the topic object is of the trigger class. So, the first thing to do is add anchor tags to the topics. But I want to be able to send different anchors depending on whether the topic is to be shown or hidden. That is, if clicking on the subhead should make the topic appear, it will look like this:

<a href='bio/'><h2 id="biosubhead" class="trigger">Bio →</h2></a>

But if the topic is to disappear, it will look like this:

<a href='-bio/'><h2 id="biosubhead" class="trigger">Bio →</h2></a>

So let’s add a second hash in addition to showmes, now called anchors. This hash behaves more or less exactly like showmes does, and we add lines like this in the about.erb:

<a href='<%= anchors["bio"]%>'><h2 id="biosubhead" class="trigger">Bio →</h2></a>

Now we need to trick out the Sinatra code to strip out any parts of urls that include minus signs. So it would turn http://moacir.com/about/bio/name/-bio/ into just http://moacir.com/about/name/. This requires a little bit of clumsy string manipulation, since I have to match the part that includes the minus sign (-bio), strip it, then strip its positive cousin (bio), and then strip all double slashes, as usual.

The whole route reads:

get '/about/*' do
	if /-/.match(params[:splat].first)
		path = params[:splat].first
		/-[a-z]*/ =~ path
		parttohide = Regexp.last_match(0).gsub(/-/, '')
		path = path.gsub(/-([a-z]*)\//i, '')
		path = path.gsub(parttohide, '')
		path = path.gsub(/^/, '/about/')
		path = path.gsub('//', '/')
		redirect path
	else # as before, but note the addition of the anchors hash
		topics = params[:splat].first.split("/")
		subheads = ["bio", "name"]
		showmes = Hash.new
		anchors = Hash.new
		for subhead in subheads
			if topics.index(subhead)
				showmes[subhead] = "style='display: block;'"
				anchors[subhead] = "-#{subhead}/"
			else
				showmes[subhead] = ""
				anchors[subhead] = "#{subhead}/"
			end
		end
		erb :about, :locals => {:showmes => showmes, :anchors => anchors}
	end
end

Everything looks nice, except for one problem: the Academics page has nested topics! If I request http://moacir.com/academics/presentations/, then it is as good as getting the regular page, since presentations is nested within publishing. Similarly, if I click all those subtopics open and then close the publishing topic, they remain part of the url. Unfortunately, the only way I can think of dealing with this is with more string manipulation that is now hyper-specified.

These two if blocks seem to cover the two situations: ensuring that publishing is in the url if a subtopic is and obliterating all the subtopics if publishing is being closed. The latter can go inside the general /-/.match. For the former, I need an extra twist to the regexp so that while looking for publishing it does not get a false positive from selfpublishing.

get '/academics/*' do
	# Add this if block
	if /(poparticles|presentations|selfpublishing|cartography)/.match(params[:splat].first)
		redirect params[:splat].first.gsub(/^/, '/academics/publishing/') unless /(\/publishing|^publishing)/.match(params[:splat].first)
	end
	if /-/.match(params[:splat].first)
		path = params[:splat].first
		# Add this if block, too
		if /-publishing/.match(path)
			path = path.gsub(/(poparticles|presentations|selfpublishing|cartography)\//, '')
		end
		/-[a-z]*/ =~ path
# etc…

Almost done…

So it worked with JavaScript, and now it works without. But if I turn JavaScript back on, I lose the AJAXyness that was the point of all of this in the first place. I need to do two things here. First, I name the id and class in the <a> tag instead of in the <h2> tag as before. So what was:

<a href='<%= anchors["bio"]%>'><h2 id="biosubhead" class="trigger">Bio →</h2></a>

is now

<a href='<%= anchors["bio"]%>' id="biosubhead" class="trigger"><h2>Bio →</h2></a>

Then, in the JavaScript, I add a line that strips the href attribute from all trigger class anchors:

$('.trigger').removeAttr('href');

While I’ve got the JavaScript open, I’ll quickly add some logic to the $('.trigger').click() expression that obliterates the subtopics inside publishing:

	}else{ // so the toggle is visible. Hide it and delete it from the URL
		pathname = pathname.replace(togglepatt, ''); // erase it!
		// Add this if() clause
		if (togglename == 'publishing'){ // get rid of specific subtopics, too
			subtopics = ["poparticles", "presentations", "selfpublishing", "cartography"]
			for (x in subtopics){
				pathname = pathname.replace('/' + subtopics[x], '');
				$('#'+subtopics[x]+'body').hide();
			}
		}
// etc.…

Finally, I prefer urls without trailing slashes. I think http://moacir.com/academics is more handsome than http://moacir.com/academics/. So I’ll add a general route without the trailing slash. that assumes that everything is hidden. Notice the subtle addition of about/ to the anchors values. This is necessary to maintain the structure when JavaScript is disabled.

get '/about' do
	subheads = ["bio", "name"]
	showmes = Hash.new
	anchors = Hash.new
	for subhead in subheads
		showmes[subhead] = ""
		anchors[subhead] = "about/#{subhead}/"
	end
	erb :about, :locals => {:showmes => showmes, :anchors => anchors}
end

And, finally, (for real this time) I add some logic to the previous route at the top to redirect if the url ends in a slash at the root level.

get '/about/*' do
	# Add this if block
	if params[:splat].first.empty?
		redirect '/about'
	end
# etc.…

And that’s about all I need to do! Nothing all that tricky, and it gave yet another opportunity to do some serious problem solving while learning just a bit more about Ruby, Sinatra, and jQuery.

Of course, none of this probably works on IE. Life is tough.

What really matters, though, is that I couldn’t sleep last night from too much Diet Coke, and I was too out of it to do regular, real work. So I dreamt up this idea of having a Molly Bloom twitter bot randomly tweeting lines from the final episode of the novel. The pieces fit together in my head while staring at the ceiling in bed, and I spent much of today writing it.

So now I can unveil @MollyBloombot. If you @ the bot (I cannot quite call it “her,” because I created it, and my vestigial essentialism is uncomfortable about the trope of males creating female avatars online and in modernist novels) and say “random,” you will get a random stream of text. If you say “word” followed by some kind of word, the bot will tweet you back a line featuring that word. If the word is not in the text, the bot will try to guess a similar word based on the Metaphone algorithm. If it fails, the bot will get dreamily flustered. The same happens if you just @ it all kinds of jibberish.

If you @ it “info,” it will tell you about this post. If you @ it “help,” it will tell you what I just wrote.

If you follow the bot, it will (eventually) follow you back. Following it might be fun, since it .@s its string responses, meaning you can see what kinds of things people are asking it to tweet about as well as seeing what kinds of random tweets get generated.

If you ask it about a word that it finds, it’ll send a second tweet, not .@’ed, just to you describing a little about the instance of the word you chose.

Finally, it’s a bit poky, but, really, what else could you expect?

The bot is written in ruby and interacts with Twitter via chatterbot. The Metaphone algorithm comes from the text gem.

Google Scholar search. (click to enlarge)

Now if I search for the article on Google Scholar, using the search term “‘Under the Black Flag’ Mosby,” the top hit is to the article in question, appearing in Military Law Review in 1994. And if you click on the link, it takes you to HeinOnline, a web journal gatekeeper with which I was previously unfamiliar. It asked $30 for the privilege of browsing the site for the day.

So I did what I normally do in this case: try again, this time through my university’s proxy. Nothing changed, and the article remained elusive. But when I tried my Plan B, which is looking up the journal through the university’s library catalogue to see where they consider the electronic archive of the journal to be, I was forwarded to the Library of Congress, which offered not only the issue in question but an entire back catalog of Military Law Review for free, for everyone. No proxies, no nothing.

What gives? Google Scholar only links to the pay version of an article that is free from the US government?

Ah, but then look: there’s that link to “all 4 versions” of the article in the Google Scholar database. These versions include the HeinOnline one already mentioned, a citation of some sort, and two links that take you to LexisNexis, another for-pay gatekeeper. There is no link on Google Scholar for the free version of the article.² At least not that I can see. And if you search just on “military law review,” the first free pdf–offered by the military, not the LoC–comes only on the third page of hits. You have to click to the fifth page before the loc.gov pdfs start showing up in the Google Scholar search results.

I know publishers are pleading poverty left and right, but something about this situation stinks and stinks badly. If there’s a gateway to the free version of this article via HeinOnline, I was certainly unable to find it. The same is true via LexisNexis.

So fair warning: always, always, always exercise a little legwork before considering parting with $30 for an article whose value is uncertain from reading just the first page. And Google Scholar is certainly not providing the most useful results.

(I guess part of the issue here could be that the Military Law Review at the LoC does not seem to be indexed by articles, just by issues. Perhaps that is why the Google robots do not find them. Still, boo to paying for free things!)

1. Despite the fact that I’m pretty sure I don’t like For Whom the Bell Tolls or consider it particularly good, I could probably write a book about how crazy it is in important and interesting ways without repeating any of the work in Blowing the Bridge.
2. A regular Google search yields the free version at the bottom of the first page of hits, after the links to HeinOnline and LexisNexis.

Immediately, out came the haters. But Ravi’s two points are worth addressing: population density counts for something, as does bus access, right? Fine. But, most markedly, I’m explicitly not asking about transportation access as a whole. To do that would then also involve bringing in questions of commuting and the like–and I only coded Parisian (and near Parisian) Métro stops, not the system as a whole, and I only measured my Chicago maps for Chicago, not Chicagoland. It goes on. My question was simple: no matter where I am in Paris, what’s the furthest I am from a Métro station? The answer is about 700m. What would that same number look like in Chicago? This then explains why in Chicago I seldom feel close to an El stop, whereas in Paris, the Métro is unavoidable. So I was explicitly not asking the questions Ravi asked. And I’d argue that I didn’t ask them since they don’t change the answer.

So let’s tackle population density: First, in my area calculations of Paris, I included the two huge forests. “No one” lives there. If you take the forests out, then basically all of Paris is within 700m of a Métro station. That was the point of the initial exercise, and that’s why I’m hung up on this dumb number of 700m. So there you are: about 2.2 million of a possible 2.2 million Parisians live within 700m of a Métro station. The number of Chicagoans is, clearly, different. Some percentage of the possible 2.6 million Chicagoans live within 700m of an El stop. I’m going to guess that it’s not 100%. I’m going to guess it’s not even 84%, which would yield 2.2 million Chicagoans, making the number equivalent to Parisians. Whatever the number is, it’s much, much lower, as is evident from any acquaintance with Chicago. So even if I cut out O’Hare, the Stockyards, the Port, Beverly, whatever, it’s still simply not the case that everyone lives in the teeny little disks on the Chicago map. My point, however, is that everyone does live in the teeny disks in Paris. In that case, this is all self-evident: Paris is denser than Chicago in terms of population and train station coverage. Cheers.

As for the bus question, I’m not a crook explaining Chicago’s terrible transportation coverage in order to sell you a monorail. I’m making a comparison between like modes of transit. As I responded, in Chicago the bus system is obviously there, in part, to cover up massive holes in El coverage (as well as provide redundancy). As in, it’s evident that the bus goes to places that the trains just don’t. But saying that “Chicago ain’t that bad. We got buses, too!” is not the answer here, since even if El + bus + Metra provides public transit within 700m of everyone in the city, Paris manages a similar level of coverage with just a subway system.¹ If you add in the buses & trams, then public transportation becomes ubiquitous in Paris. Taking note, then, of the density of the bus system within the city, which runs not redundantly with the Métro, it’s possible that we start hitting numbers like 250m when it comes to the question of “how close is public transportation at all times?”

Five Parisian commutes. (click to enlarge)

Which brings me back to what spurred this little investigation: the anecdotal feeling that public transportation has a profile in Paris that it simply does not have in Chicago. I don’t think there exists a “what about…?” that makes my anecdotal suspicion turn out to be wrong. It’s just a question of how right it is.

Also, I can only get population data at the Arrondissement level for Paris, which is probably not terribly interesting.

Finally, for fun, I close with this quick little Google Map I made of the five possible commutes I can take between home and work. The shortest distance, in blue, is the walk. It is about 4.35km and takes me about 50 minutes. I could also take the 62 bus along most of the same route. It would shave time off the total, but I don’t know how much; I don’t time myself when I take it. The next shortest is the bike ride, which is in green and 4.55km in length. It takes less than 20 minutes. The more commonly taken bus solution for me is the T3 tram with a transfer at Porte d’Ivry to the PC2 bus. It’s aqua on the map. It takes less than 40 minutes and runs 4.96km.

Next are the two Métro solutions: in pink is taking the M4 from Porte d’Orléans to Saint-Michel and switching to the RER C to go to Bibliothèque Nationale. It takes over a half hour and runs 9.95km. I can’t believe that there was a time when this was my daily commute. Finally, in purple, is the University’s recommended commute, which involves taking the RER B from Cité Universitaire to Châtelet and swtiching to the M14 to Bibliothèque Nationale. It also takes over a half hour and runs 10.35km. Anyone who knows me well knows what kind of existential pain these last two commute options bring, but they do serve to make my ancillary point about the Parisian bus system: it’s there to cover up holes in the Métro system, which already doesn’t have much in the way of holes. But, yes, getting from the southern edge of the city to the southeastern edge can be tricky if you don’t have recourse to the bus or tram.

1. I think it is probably the case in Chicago that one is rather close to a bus at all times, considering that Chicago buses tend to run on the grid at half-mile increments. That means that you can expect to be .25mi (or 400m) from a bus at all times.

Vélib’ coverage in Paris. (click to enlarge / make useful)

When I made my Paris Métro map, the joke was that the next step would be the leap in order of magnitude between subway stations and Vélib’ stations. For those who don’t know what Vélib’ is, it’s the Parisian bike-sharing system that I’ve already described in great detail. But I knew there was no way that I would hand geocode each station (there are over 1200 of them), so I put the affair off. But this morning, I wondered if I could possibly reverse engineer a list from snooping around the Google Maps widget that Vélib’ offers. This widget I use every morning when deciding where I’ll get my bike and where I’ll park it. Luckily, a bit of sleuthery led to this XML file, which has the latitude and longitude of each station.

Once I’ve got that, the rest is kids’ play. Parse, parse, plot, plot. And now I’ve got this map here, which is so busy and full of information that it’s nearly impossible to understand. But one thing is certain: my initial guess was that one was hardly ever more than 350m from a Vélib’ station. That turns out to be about true. Not including the two Parisian forests, there are only handfuls of areas within Paris itself that are more than 400m from a Vélib’ station. It’s a remarkable amount of coverage that makes me realize how outrageously extensive this project was (and continues to be).

Still, there were a few kinks in the data that I found worth a remark. Two stations, 20018 and 20048, both in the 20th Arrondissement, had latitudes and longitudes that were, well, identical, and, also, in Algeria, prompting more than just me to make jokes about racist civil servants in the Mairie. What’s interesting, further, is that neither of the two stations appear on StreetView (I’m not about to bike up to the 20th to check in person). One, 20048, is marked on OpenStreetMap at the proper address, 110 rue de Bagnolet. The other, however, is listed at 2 rue Hartignies, which does not seem to exist. There is a 2 rue Harpignies in the 20th, but there is no Vélib’ station there on OpenStreetMap. Now if you go to the Vélib’ Google Maps widget and type in either station number, it will take you to Algeria and tell you that bikes are available there. Is this an error? Maybe. Is it coding trickery to prevent copying, along the lines of a trap street? Also possible.

Vélib’ wastelands in Paris. (click for Google Maps interactive)

Back to the coverage, though. In short, the map is so dense that I decided to experiment a bit further, so I created a shapefile featuring only the places in Paris that were more than 250m from a Vélib’ station. I then converted it to a KMZ, and you can open it in Google Maps by clicking here. I’m a bit proud of this interactive little Google Maps thing, so please have some fun with it, if you like. I used a quickly drawn outline of Paris for the clip shape, so it’s not a perfect fit to the Maps interface, but we can notice a few things right away about Vélib’ coverage in Paris proper. First, if you’re in a forest, you’re more or less completely out luck. This makes sense. Also, if you are in the middle of a railyard, you probably are rather far from a Vélib’ station. Most of the swaths along the northern edge of the city, as with Métro patches, are devoted to trains racing out of the city, so it’s not terribly necessary to have bikes where only trains go.

But I find it more interesting that along the Seine, it is hardest to find a Vélib’ stand where tourists are most likely to be: between the Louvre and the Palais Royal as well as near the Eiffel Tower. It’s not a complete dead zone, of course.¹ All of Champs de Mars is within 400m of a Vélib’ stand, and only a teeny bit of the park by Trocadero is more than 400m from a stand. But, still, Vélib’ing is trickier if you’re engaged in touristy activities, which is, actually, only fair. Vélib’ is useful for tourists, and they can use it all they want, but it should not be (and obviously was not) designed with the tourist in mind. The rest of the patches within the city largely correspond to either parks, railways, or hospitals.

I will close out this post with a link to a short presentation on the French ESRI site about Vélib’, which shows what a huge GIS project this was, involving planning at massively different levels of scale, such that the system works exactly as it should, despite a huge population in constant transit with shifting demands. Hopefully I’ll be able to supplement my own maps here in the future with Parisian demographic data or something else entirely.

1. Of course, Saint Michel and Hôtel de Ville do not suffer for lack of Vélib’ stands. My imagination says that the tourists drift more westward, though, once they have seen Notre Dame.

El coverage in Chicago (click to enlarge).

The last time I rapped at you, I talked about Métro coverage in Paris. I felt like Paris was exceptionally well covered by the Métro, and I used math to prove that basically one is never more than 700m from a Métro station in the city. How, though, does that coverage compare with Chicago? Would you be surprised if I said that Chicago ends up looking rather awful in comparison? How much worse, would you guess? If 700m is the maximum in Paris, what would you guess is the maximum distance from an el stop in Chicago? 1km? 2km? More?

After wrangling a bit with the GIS data available from the city itself as well as the CTA data from Steven Vance, I can build a simple buffer map much like the Paris map from the previous exercise, and the results? They ain’t pretty.

It is a bloodbath, in fact. Whereas in the previous exercise, the buffers were 250, 500, and 700m from the Métro, In this map on this post, the buffers are 1, 2.5, and 5km from the stop. In other words, everything that is dark purple is between 2.5 and 5km from a CTA El stop. Areas that are tan are parts of Chicago that are over 5km from a CTA El stop. But even if we just look at the 2.5km discs, we see that what feels like a third of the South Side is abandoned when it comes to the El. Pullman, Beverly, Rainbow Beach. All these neighborhoods are, literally, miles from the El. Even the parts of Hyde Park east of the Metra tracks (hot pink) are over 2.5km from the Green Line.

Ah, but what about the Metra, then? Surely some of these coverage gaps can be accounted for with the Metra, right? OK, let’s add them, as I included RER stations in my Métro map. And, fair enough, once you add in the Metra, nearly none of Chicago is more than 2.5km from a train station of some sort (dark blue for CTA, dark green for Metra). But, still, I’m using largely inflated scales for these maps to try and get something like the coverage of Paris. So, as a final map, I’ll dial the buffers back down to Parisian size: 250m, 500m, and 750m, just so you can compare, side-by-side, Parisian coverage with Chicagoan coverage.

At my university, a vice president has been arguing that there is no place for a GIS (geographic information systems) curriculum because now everybody can get that kind of data and everyone can make maps.

…

What is your sense of it? Is GIS really dead as a discipline? (And have I thus completely missed what’s going on?) What arguments would you put forward? What are the arguments stronger than the ones I have suggested above?

Thinking that GIS has no place in a curriculum anymore because of the proliferation of GPS devices and (let’s say) Google Maps strikes me as a rather ill-informed position to take, considering practitioners of GIS have not even figured out what it is (or what the “S” in it stands for). How can a moment be over before it is even a moment?

During the past two decades, geography journals have repeatedly flared up with arguments over whether GIS is a tool or a science (thereby becoming a (sub-)discipline), whether use of GIS automatically aligns the practitioner with the military/corporatist history of GIS, and whether such a thing as a qualitative GIS can possibly exist, thereby ”freeing” GIS from its allegedly quantitative and positivist roots. Recent articles by Marianna Pavlovskaya give general histories (and useful bibliographies) of the debates, from John Pickles’s broadsides in the late 90s to Mei-Po Kwan’s feminist rehabilitation of GIS in the last decade.

Furthermore, collected volumes published in the past few years, like Hillier and Knowles’s Placing History, published by ESRI (the publishers of the ArcGIS software package), and Cope and Elwood’s Qualitative GIS (usefully and succinctly reviewed by Daniel Lewis), published by Sage, give accounts of several GIS projects that could simply not be accomplished without GIS (as well as geography!) training that goes beyond hours of asking the internet for driving directions or geotagging photos.

What the vice-president seems to have in mind is what many have called ”Neogeography,” the sort of DIY punk geography that could be the equivalent of the cheap handheld movie camera or portable four-track recorder. But film schools did not close because of the cheap handheld (this seems a useful comparison to me), nor did, and this is vitally important, film studies departments or the companies that make large, pro cameras.

That is, neogeography is a new approach to the creation/collection of geographical data, but the old forms (census tract tables, for example) have not lost their importance at all–nor have they become, I suspect, more intuitive. Similarly, assuming that GIS is “only” making maps on Google Earth would probably be considered an insult by even the people whose chapters were rejected in the volumes mentioned above.

Additionally, for historians, the ease of creating maps of today’s world means virtually nothing when what one cares about is the world from over a century ago–a massively labor-intensive project documented, for example, by Anne Kelly Knowles in her effort to imagine, using GIS software, what General Lee was able to see from his post at Gettysburg. The data she used was not available to ”everybody.” She had to create the data by hand from historical topographical maps. That also means she had to know–have been trained–how to create that data.

Finally, I can give my own personal experience, which was that of a year-long course in GIS, for which a course in statistics (not ownership of a Tomtom) was a prerequisite. Outside of a short unit on using GPS devices to make a small map of campus, nothing we did in those 30 weeks fits the description of Neogeography given on Wikipedia (or is recognizable in the VP’s concern). An iPhone won’t teach spatial analysis, how to measure clustering, what a nearest neighbor is (and why that is or is not important), how to correlate income data from the federal census with crime or transportation data provided by the city, or how to answer even a basic personal (non-academic) question, like, ”where should I live if I want to live within 200m of the subway, within 20km of work, in a neighborhood with an average per capita income of at least $20k, and with < 20 property crimes in the past month?”

Hopefully this email has given some arguments (and suggestions for further reading) about how GIS (or geography) can’t be simply brushed off because of the ease with which one can make “mashups” on the internet.

I think it’s funny that JFK could have merged the idea of the “Global South” with the literal southern hemisphere. Randall Munroe’s snarky joke doesn’t change the metaphorical power of the expression.

But what a glance at Munroe’s own map shows is that the bulk of the land on Earth is located north of the Equator, so if we picked a speck of terrain at random, it’ll more often fall north of the Equator. What if, I then wondered, I got rid of the Equator and split the Earth in half by area, in effect making an “Area Equator,” so that a randomly selected point on land would have a 50% chance of landing in the “South” as opposed to the “North.” What might that world look like?

Enter Quantum GIS. I downloaded a shapefile of the world, and it had area already keyed in as an attribute for each country.¹ I was ready to calculate the areas of each polygon, but I’m glad I didn’t have to. Then I added polygons until the sum of the areas of the selected polygons was about half of the total sum of the area. Next, I chose an outrageous color scheme. The results:²

The world, split by area. (click to enlarge)

What’s interesting is that this map is not so terribly different than the map of the North-South Divide provided by Wikipedia. The main difference is that I include Australia, while they include much more of Asia. China by itself has a larger area than Australia, so subtracting the Aussies from my area list and adding China would already knock the swing out. But my point is, at this stage, strictly cartographical.³ One can now sort of see where the “Area Equator” of the Earth is.

So this doesn’t let JFK off the hook, but it might nuance the point a bit.

1. There were some errors. A handful of small countries and Eritrea reported 0 for their area.
2. I ran the test twice, with and without Antarctica. Mostly adding in that frozen landmass means that I have to deselect much of the Middle East, Pakistan, and, I think, Algeria. So it’s not terribly different. Remember: Antarctica is never as big as it seems on unprojected maps.
3. Another caveat: I have no idea where the “area” calculation came from, so who knows how reliable my results are.

ArcGIS in action. (click to enlarge)

[A lot of the below is meandering toward what I suspect is a rather obvious conclusion to hardened veterans of the digital humanities. Since I'm not one of those, my own shoes needed to walk the mile. Of what transpires below, what might be new is, quickly, how while there is a call for digital humanists to move past prose to include other forms of analysis (maps, in this specific example), geographers have a different approach to the post-prose moment, one steeped in skepticism over the value of visual representations of data. Are geographers scaredy cats? Or might digital humanists be overexuberant? Or some combination of neither?]

Aside from the “reflexive vs. positivist” opposition in Martyn Jessop‘s talk at the Institute for Enabling Geospatial Research and his similar article in Literary and Linguistic Computing (discussed briefly here), the opposition that caught me most off-guard in Jessop’s article was one that was reasserted a few times at Geoinst:

there are fundamental issues concerning the status and function of images in humanities scholarship, this includes the images produced by digital visualization tools. Humanists are used to expressing themselves, and assessing the work of others, through the medium of prose. There is a belief that the visual cannot be as rigorous as the written. It is seen, as, at best, a supplement to the written word and stands in a subordinate position.

Jessop continues to explain how images in “modern educational texts” tend to be merely distractions to break up the flow of the text as opposed to being an integral part of the argument. There are “very few instances where the visual is treated on an equal pedagogical footing with the written.”

This line of reasoning from Jessop’s article continued in his talk, and Anne Kelly Knowles referred to it as well, explaining that history tends to be verbal, whereas geography tends to be visual, setting up the situation in history where the text is privileged over the map, which requires more critical response.

Now it is certainly not the case that the humanities value the textual over the visual as objects of study. I know a few art historians, musicologists, and students of film who would spit milk over their keyboards upon reading an assertion like that online. But it feels true to say that, as a mode of scholarship, the prose work lays claim to the most prestigious form of knowledge creation in academe.

From my reading of the dizzying Digital Humanities Manifesto 2.0 at UCLA, I get the sense that this tension is a relatively well-investigated and argued one within the digital humanities.¹ Part of the appeal of DH, it seems, is precisely attacking the primacy of prose as the form good scholarship should take, which is reflected in the value given the curatorial (as opposed to the straight analytical). As the UCLA manifesto asserts,

[W]e are advocating for a neo- or post-print model where print becomes embedded within a multiplicity of media practices and forms of knowledge production… Digital Humanists recognize curation as a central feature of the future of the Humanities disciplines… Curation means making arguments through objects as well as words, images, and sounds… All of which is to say that we consider curation on a par with traditional narrative scholarship.²

What’s interesting here, to me, is the interest of the digital humanities to move toward the visual precisely when geography is having its own crises, especially among critical geographers, regarding the visual. The visual is attached to the “scopic regime,” an “ocularcentrism” derived from Descartes, who posited “an epistemological standpoint of early modernity that subscribed to the notion of a detached, objective observer capable of a ‘god’s eye‘ view of the world.” Barney Warf here is drawing a history of geography’s relationship with “capital accumulation [and] the rise of the nation-state,” a relationship helped by the illusion of the “certainty of visual knowledge.”

As the visual arts came to be dominated by linear perspective, so, too, did geography come to be dominated by the idea of homogenous, infinite, Newtonian space containing interlocked nation-states or other rigidly bounded entities, within the metaphor of the surface. “The rise of logical positivism in the late nineteenth century,” Warf continues, “added a aura of scientism to this view, mathematicizing it with the disciplines concerned with space such as geography and urban planning in the forms of isotropic planes, surface in which the distribution of social features is evenly distributed.” This scopic regime continued in geography, more or less, until the critical geographers began to break away from it and the quantitative revolution in the 1970s.³

But it is not the case that the critical geographers are asserting for “more prose” in their work. Instead, they were looking for ways to get out from the empirical burden, which, as Soja remarks, though producing “significant and useful factual knowledge about the objective, real world,” had a “tendency to fixate on materialized surface appearances and directly measurable patterning, creating an illusion of opaqueness that could block deeper understanding of the causal forces underpinning these surface expressions.” These “idealized visions of the world” led to a “luminous search for deep structures of causality as the imagined took precedence over the real.”

Soja himself, one of the fiercest proponents of a larger role of spatial thinking in attempts to understand the world, does not argue for “more maps / less prose” but for merely an approach that treats the spatial as an equal party in the trialectic of spatiality-historicality-sociality. In fact, it seems that it is the reliance on the visual that gives geography its static and synchronic sense, a rigidity (in comparison to history’s richness and dialecticity) that Michel Foucault suspects is the result of Henri Bergson.⁴

I bring up this brief history of geography since it shows how the emergence of GIS can be seen (and is often seen) as a reaction to the work of the critical geographers. Soja calls it a “defensive disciplinary response.” “Over the past ten years,” he continues, “the positivist and descriptive core of geographical analysis has refortified its centrality, sustained in part by large flows of financial support for the advancement of Geographical Information Systems (GIS).” The apparent intellectual offspring of the quantitative revolution in geography, “today GIS,” as Marianna Pavlovskaya explains in a 2006 article in Environment and Planning A, “sustains an industry worth $6 billion a year… and remains a corporate and state-sponsored technology widely used for profit making and control.” If GIS was not so appealing as a means of state and capital control, it wouldn’t be getting the funding it is today, especially in contrast with critical geography.⁵

GeoDa, now available for Mac and Linux! (click to enlarge)

But Pavlovskaya’s article does not to simply criticize GIS: she explains that the reception of GIS as the latest guise of state-control/positivism is misguided, and that, in fact, GIS is beginning to be used as a qualitative method, that is, one of the methods that has “become an accepted strategy for those advocating nonpositivist knowledge production and aspiring for emancipatory change.” GIS, Pavlovskaya argues, gives the illusion of precision and exactness, which subsequently gives the illusion of quantitative analysis. But putting something in a database doesn’t guarantee accuracy, just like relying on fieldnotes doesn’t guarantee sloppiness. Furthermore, computers don’t guarantee logic: one can behave illogically with them and logically without them. Both quantitative and qualitative approaches involve interpretive efforts at pattern detecting, at reading textual fields.

Next, and on this point I’m not sure I stand with Pavlovskaya, it’s not the case that the math used in GIS analysis is actually complex enough to qualify as “quantitative” or even “statistical.” A lot of it is, at its root, just counting or not notably different from regular human interaction with space. She writes,

In truth, most spatial techniques available in GIS are only marginally ‘quantitative’ despite being very illuminating. Using simple math (such as distance measurements or calculations between raster layers), they require spatial imagination skills (such as buffering or overlay) and logical thinking (such as combining layers in site selection of multicriteria evaluation). As such, these core functions replicate human spatial thinking about places and phenomena that is common to all geographic research. Overall, spatial analysis in GIS today is largely qualitative, visual, and intuitive despite its insistent labeling as a quantitative method.

She further offers that even cutting edge “quantitative” work in GIS using AI or Bayesian probability is just an “attempt to replicate human reasoning.” In my mere year’s worth of GIS training, we definitely started feeding legitimate statistical beasts, doing spatial regressions and clustering calculations. These don’t replicate human reasoning–in fact, they exist precisely to slow down human reasoning, which is often terrible at detecting randomness, or the lack thereof. Pavlovskaya certainly isn’t asserting that all GIS is qualitative, of course, but I know that, in my work, I felt like I was eating at the kid’s table until I started being able to attach p-values.⁶ This is my own bias, though, that I’ll unpack another day.

Pavlovskaya does approach the leading question of this post head on, however, in terms of visualization–the image over/with the text. Data visualization is what GIS’s core strength seems to be, as GIS can output maps quickly and efficiently, with both quantitative or qualitative data. She points to Mei-Po Kwan’s work on irrational responses to data visualization to argue that it’s “the most telling example of nonquantitative functionality in GIS.” The visualization is, in fact, quite the alluring song that attracts people–myself included–to GIS. If part of our charge, as digital humanists, is to move past prose, to visualize our data, the satisfaction of GIS is really right up our alley. And it’s easy to get striking results: “Visualization is so powerful a technique that often the manipulation of data within GIS does not go beyond querying the data and displaying the results.” Feed in a spreadsheet of census data, dial up a chloropleth, export to .jpg, and move on.

This is a bit flip, but I think it’s important to assert it. Visualization shouldn’t be an end in itself, and engagement with and understanding the tool of visualization deserves the highest priority. Pavlovskaya warns about how maps are tools of control. Maps over-assert their reliability by relying on a metaphor in the mind of the viewer in which space is scientific and exact, so, as a result, maps are true. Here I cue, again, of course, Mark Monmonier, who warns in his epilogue about using a map with the “dual role of both informing and impressing its audience.” After all, “a flashy map… touts its author’s sense of innovation, and cartographic window dressing in a doctoral dissertation… suggests that the work is scholarly or scientific.” With GIS, we have the added authority of having a computer that’s making the map, so the stink of truthiness clings even more formidably to the embedded .jpg.

Warf closes his own article on networks with a bomb detonated in the ocularcentrist modernist’s favorite street-corner café. Vision’s attachment to truth “is essentially a positivist assumption that denies the possibility of other ways of understanding the world.” Mobilizing Richard Rorty, he finishes by declaring that “once we abandon the positivist metaphor of the mirror as the basis of objective knowledge, we are led to the metaphor of the conversation, in which language, positionally, and dialogue are central.” Strangely, to me, this sounds like, in part, a call for less image-based analysis and more dialogue-based thinking, which gets recreated in prose.

I plan on not introducing any more new sources from here on in, so a recap is in order: there is a move to advance past prose-based scholarship in the digital humanities. This means curating various kinds of objects, this means incorporating non-prose forms of analysis (like maps) and data visualizations in general.

Visualization, however, is an approach to data that, along with its current big-budget exponent, GIS, is attached to quantitative analysis, and, hence, to forms of state and corporate control, power relations that move in direct opposition to projects in the digital humanities that are interested in the empowering capability of digital humanistic scholarship. Furthermore, visualization as an end to itself is still wrapped up in questions of power and control that, in my reading of the UCLA Manifesto, remain unaddressed, pushed aside to make room for unrelated emancipatory rhetoric.

On the other hand, GIS itself has not managed to live up to the hype surrounding it as a quantitative tool. In fact, the revolutionaries in the qualitative world can exploit its power for their own purposes.

But is quantitative work necessarily bad? Can’t there be a quant/qual matrix that people like? This is probably a terribly boring discussion that’s been had at every social sciences get together where there are as many bottles of wine as graduate students, but it’s still new to me.

I find it interesting, for example, that the UCLA manifesto separates quantitative and qualitative into historical moments–a sort of political/developmental timeline that Marx might be proud of:

The first wave of digital humanities work was quantitative, mobilizing the search and retrieval powers of the database, automating corpus linguistics, stacking hypercards into critical arrays. The second wave is qualitative, interpretive, experiential, emotive, generative in character. It harnesses digital toolkits in the service of the Humanities’ core methodological strengths: attention to complexity, medium specificity, historical context, analytical depth, critique and interpretation.

Now there’s a lot in this snippet that I think is very, very wrong (or, at least, getting carried away in the rhetoric of the manifesto).⁷ But it does pitch the digital humanities in a similar historical narrative as that of critical geography. Quantitative geography, however, has not disappeared, so we can’t talk of geography in waves as much as in branches. And considering the fantasy of the quantitative promise of GIS (which, pace Pavlovskaya, I still have), being encouraged to incorporate it into my digital humanities work certainly doesn’t seem like a full on, earnest effort to be, also, “qualitative” and “emotive.”

So I end with two messes on the table for starters: the political/control nature of visualization is unaddressed in relation to the pressure/encouragement to visualize and the role of quantitative work in digital humanities seems to earn the feeling of being old-fashioned or compartmentalized within a larger qualitative framework, at least within the framework of the UCLA manifesto.

Conveniently, I’m walking away from these messes, citing a lack of space on this page to continue. But I do have a feeling I’ll be returning to the UCLA manifesto soon enough. The tension over visualization, though, seems like it might be too complex for me right now.

1. I’m new around here, remember.
2. “Narrative scholarship” here, I think, means “prose scholarship,” not scholarship of narratives. But I’m not positive.
3. Soja on the quantitative revolution: “This increasingly technical and mathematized version of geographical description, however, differed only superficially from the neo-Kantian tradition that helped to justify the isolation of geography from history, the social sciences, and Western Marxism.”
4. “Est-ce que ça a commencé avec Bergson ou avant ? L’espace, c’est ce qui était mort, figé, non dialectique. En revanche, le temps, c’était riche, fécond, vivant, dialectique.”
5. For a tour de force of geography in the service of state control, I encourage one to read the opening pages of Gearóid Ó Tuathail’s Critical Geopolitics. The short version is that Ireland did not exist until the English crown needed to control it, so they sent in their surveyors to create an Ireland by mapping and dividing up the land.
6. On the other hand, Pavlovskaya mentions that even with the hard core quantitative work, “GIS technology has fulfilled its promise for quantitative analysis only marginally.”
7. These concerns are not appropriately addressed by the backtracking in the sentences that follow the quoted material.

I have been called a “provocateur” at work for supporting les fennecs. This is my office door. (click to enlarge)

Yet again, I’m putting off the “Fieldwork vs. Armchairwork” post, which began as a joke threat, but is actually slowly turning into a few ideas about methods courses from a total neophyte and non DGS.

In the meantime, I’m getting very excited about 64 other “vs.” coming up in the next six weeks, namely the World Cup in South Africa, which opens on June 11th, pitting the hosts against the uncertain Mexican team. I think the Bafana Bafana will need quite a bit of hometown bounce to beat Mexico, but I find that totally within the realm of possible outcomes.

If you fancy yourself a good prognosticator of possible outcomes, I welcome you to join my silly little World Cup Challenge Pool, where one can pick the results of all 64 matches and win potential prizes.

I like some of the tricks I pulled off with the website, and it is, I promise, the last thing I ever write in procedural PHP (at least, that is, until I feel confident saying “Give me procedural PHP or give me death!” In other words, there’s a lot of progamming pupu platters in perceived future). Furthermore, the site relates a bit to the themes of this week’s posts in that I worked on it during downtime at Geoinst, despite the fact that it was already live. Oops.

Charlottesville, VA.

Yesterday’s post on the tension between curatorial/service-y intellectual work and straight up analytical work was intentionally kept rather general, both for larger appeal and since I’m trying to figure out my approach to these questions in a way that’s consistent. Today, I’ll be a bit more specific, and this is sort of a warning about that.¹ I want to show how geospatial scholarship can be a sort of ground zero of these kinds of discussions and debates, though, about curating and analysis.²

Whether factually true or not, I got the sense that I was one of the only people at the NEH-sponsored Institute for Enabling Geospatial Research at UVA’s Scholar’s Lab who had never used Google Earth as more than a fun little toy.³ I’m still pretty uncomfortable with the interface, I have no idea what the program does, and I really can’t figure out what role it plays in my life. Most succinctly, finding the url for the following link is the most work I have ever done with the Google Earth API.

On the other hand, I was probably in the top quintile among participants in the use of ArcGIS. So how is it that I’m reasonably proficient, by humanities student standards, with ArcGIS, but completely covered in thumbs regarding Google Earth? Curating vs. analyzing. Webmapping vs. mapping (for the argument).

While trying to figure out, during the proposal process, if my approach to literature was crazy or mainstream, I spent a lot of time trying to find similar projects to mine online. Much of the work I found was similar to Google Lit Trips, a persistent whipping boy of mine. Google Lit Trips is a curated repository of data from multiple contributors (so, collaborative) leaning on the Google Earth API, available largely to enhance the experiences of the encoded text for K–12 readers. In other words, it features service- and pedagogy-oriented work that leaves the work of analysis and argument to the reader. There are many projects like this online, and we were even shown many over the course of the Institute, either as in-the-world examples or by the “curators” themselves.⁴

I have no epistemological quarrel with Google Lit Trips (despite the fact that they’re getting darned close to my datasets!) or similar projects, like (the now defunct?!?) Gutenkarte.⁵  At this time, I have not made any use of them, as I prefer (and have the luxury of doing so) to collect my own data and code my own representations of the data. Furthermore, though our initial actions are similar–we make note of where things happen in texts–the scope of the following step is dramatically different. Google Lit Trips compiles that data into a public fly-through that helps readers orient themselves with a text. Gutenkarte scatters place names from a text on a publicly accessible map. I, on the other hand, process the data (in private) to try and make a (public) argument with it. I play the role not only of the service, but also that of the sole reader, who then transforms into analyzer and broadcasts results of analysis to the public of my dissertation committee.

My argument with Google Lit Trips, and why it’s “my whipping boy,” is that I find it limiting as far as general expectations of what geospatial scholarship in the humanities can do.⁶ I don’t want it to be the case that saying “I do geospatial work on novels” will come to be understood as “I wrote my own version of Google Lit Trips.” This is why the tension between curating and analyzing I remarked on yesterday is still not entirely resolved.

To, me, in fact, putting the data out there by itself, as, let’s say, a table of geolocated and page referenced events, is almost irresponsible, since I feel an obligation to make use of my training to provide some kind of analysis. Sure, anyone can read a .kmz displayed on Google Earth, just like anyone can read a novel. But close reading a novel is a skill that, presumably, adds some kind of value that justifies the extra layer of literary scholar to the interpretation of a text. The same is true of a map, which is basically the same thing as a Google Earth fly-through. I have been trained to close read maps, and I think that’s a skill worth sharing.⁷

But it gets even more complex. Not only have I been trained to read maps, but I’ve been trained to make them. And then I’ve been trained to augment the data on them (“geoprocess”) in order to answer questions. Once I introduce other analytical tools (network analysis, geostatistical analysis), not only does the range of possible questions I can ask (and subsequently try to answer) explode, but the answers become much more… precise. I can start talking about “confidence” and “significance.” And then I can generate arguments, which lead to chapters, monographs, etc.

At some point in this chain of events, however, I stopped being interested in webmapping, in providing a service for people to “explore” the data I’ve accumulated. I turned inward, engaging in my own play of buffers, directional distributions, and nearest neighbor calculations to see what I could learn from that play. None of this looks anything like “so you wrote your own version of Google Lit Trips,” which is why I want to encourage a high enough profile for it, so that people don’t underestimate the usefulness of geospatial work in the humanities.

But this, then, explains why I’ve never used the Google Earth API, and it’s why I only found out about Mapnik for the first time last week. Making web-accessible, pretty, interactive maps has never been the focus of my work. In fact, my maps are pointedly offline, ugly, and frozen, which they have to be for when I start processing them.⁸ And even now, I’m not sure I have a great interest in changing my approach. So, at the Institute, I was less interested in the discussions of making webmaps aesthetically appealing than in finding out if there were free ways to recreate the ArcGIS Spatial Statistics toolbox with something like Quantum GIS (signs point to no). It’s also why what I was possibly most excited about at the Institute–as far as “I can use this in my work immediately!” value is concerned–was learning that Open GeoDa had become publicly available.

On the other hand, back when I was planning my dissertation proposal, and back when that project included geocoding the events of a couple dozen novels, I always assumed that, when I was done geocoding, I would make the data available publicly. Maybe not as a interactive webmap, but still. Why not, after all? I can’t claim proprietary control over mere facts that I collected (as we learned during the fair use presentation at the Institute!). The number of novels I’ll geocode (and the depth to which I’m coding them) has greatly shrunk as my project has changed, however, to the point where I stopped thinking about the data I’m collecting as of any public utility.

In other words, I always assumed some sort of service-y aspect to my dissertation work in addition to the analytical; I was already pursuing an argument via curation, via play, via iterative exploration of my own dataset. I just wasn’t calling it that, and once I stopped considering my data to be of interest to the public, I even stopped thinking about it as a potential site of service.

But that tension… that tension remains, since now I’m saying basically that I would leave the data available in my wake. That is, I’ll continue building a monograph, and as a side benefit, anyone with a web browser can see where all the activity in Dos Passos’s U.S.A. occurred. As a result, service becomes the cherry on top, which feels kind of wrong. Or, at least, a copout.

I think I’ll end here, despite the fact that it feels like my argument has more run out of steam than concluded. These pieces I’m writing this week–most everything I’m writing about the Institute–is going to be half-baked, since it’s more a question of relating a response to the events of the Institute to my own interests than presenting finished, tidy thoughts. Oh well.

1. I know, I promised “fieldwork vs. armchairwork,” but that will come later!
2. Ooh! maybe this will be “fieldwork vs. armchairwork”!
3. A variant of this arose during our Twitter argument over cartographical aesthetics. I was forwarded links to Mapnik, which I had never even heard of before. As I’ll show, I’m still not sure how I’ll ever use it, though I’m glad to know about it.
4. I don’t know that the people involved would agree to such a designation, which is why I isolate it in quotes
5. Gutenkarte’s domain has expired. Here’s Metacarta‘s description of it: “Ever read a book, and wondered where the heck it took place? With Gutenkarte, we combine books with maps to show where a story is taking place.”
6. This part is a bit strawmanny, but I think it’s an important discussion to have, and I sort of regret shying away from it at the Institute.
7. A quick example: everyone who has taken a statistics course has probably had some kind of exercise to show that humans are awful at detecting or creating randomness. This is true on the spatial plane, too. Eyes are crappy at separating clusters from random distributions, so that it can often be the case that untrained eyes (non-skeptical eyes) can straight up misread a map. That actually may be easier to do than to misread a novel.
8. I pretty them up a bit when I prepare them for public presentation. And by “frozen” here, I mean in contrast to “flat,” which is a distinction I’ve tackled elsewhere.

The Rotunda again. (click for original)

“It seems kind of absurd to expect a 30 year old to be able to produce a monograph,” one of the attendees of the Institute for Enabling Geospatial Research at UVA said after our dinner in the stunning Dome Room. We were chatting as a group, and the topic moved to how the dissertation as a project has changed over time. Robert Scholes, one person pointed out, catalogued the Cornell Joyce collection to earn his PhD. So no less a luminary than Scholes, whose Textual Power I had to read twice by the end of my second year of undergrad, did not write a monograph for his dissertation.

Then another person at the table mentioned the frustration that emerged over getting little appreciation for the amount of extra-analytical work that went into the dissertation project… databases, webpages, etc. None of it counted, presumably since none of it would go into the monograph.¹

Personally, I had never really considered these issues. I always figured the non-monograph dissertation was a sort of fluky thing that happened after the fact, like how Tractatus became Wittgenstein’s dissertation as a formality. Similarly, I had never expected that extra-monograph-y work, like a database, would (or even should) count toward a PhD. In framing my project in the formal document of the proposal, I knew I would need to build a database. And since the proposal is a contract to do certain amount of work, I knew that building a database would be part of the labor, which can include reading (novels, theory, archival material), writing (the dissertation) and other things, like learning a foreign language or programming language or acquiring some other skill. So if I didn’t want to spend time with a db, I should’ve invented a project that didn’t require one.²

The Scholes example, though, messes my preconceived notions up a bit. And during the course of the Institute, precisely during Todd Presner‘s great presentation on the work he has done with HyperCities, the tension in my preconceived notions was framed in a tidy grudge match: curating vs. analyzing.³

HyperCities calls itself a “collaborative research and educational platform for traveling back in time to explore the historical layers of city spaces in an interactive, hypermedia environment.” Pulling from various online repositories, the HyperCities project creates something like what “The Arcades Project would have looked like if Benjamin had Google Earth,” according to my notes. Layer atop layer of feature-rich historical data spatializes various narratives of the city that is being “hyperized” (my term). I instantly saw in it lots of potential for projects that friends of mine are working on (and duly let them know about it), but it left me cold regarding the very narrow silliness that is the argument of the/my dissertation.

Presner anticipated this response, if I recall correctly, by saying that the HyperCities project is “not an argument, but a curation.” He later added that curation can be its own argument, and I’ll get to that, but I want to spend a bit more time on the coarse opposition on the table. As a curated project, HyperCities allows the visitor to “explore” a city and a history as one object made up of a staggering number of embedded, spatialized narratives, with no real argument to be found.

And there’s a weird word in that description of HyperCities: “visitor.” I can’t imagine a dissertation having a “visitor.” It has, at best, “readers,” and one can usually count those readers on one hand. Further, the monograph dissertation makes an argument. It’s not a curation. Curation is more like Scholes’s dissertation, the kind of dissertation that, to me, at this time, sounds totally inconceivable as a dissertation.

Most of the participants, or so it seemed, at the Institute were faculty, so they were not, obviously, dissertating anymore. But wouldn’t their projects, which were often similarly more aligned along the curatorial side than the arguing (I’m going to use “analyzing” from now on) side, surely have to be eventually converted into the coin of the academic realm, the monograph that posits a thesis and engages in analysis to reach a conclusion? Generating these curatorial projects seem to be great for pedagogy and for “visitors,” but I’m surprised that they get folded into promotion.

In fact, a feature-rich website offered to the public sounds a whole lot like “service,” which, one presenter remarked, doesn’t count for much when it comes to promotion. I certainly think that’s a crummy state of affairs, and I’ve written before that I think that if the humanities wants to get its “we matter” mojo back, a great way to do it is precisely via service-oriented projects, especially those with a collaborative element. But those are wishes, and, as Steve Sanders famously remarked, if wishes were horses, then beggars would ride.

So as I wrote above, I was left cold by the presentation. I analyze, I said to myself. I don’t curate. And I still think that’s true, for what it’s worth, but Presner’s saying that “curation can be its own argument” requires a bit of expansion, since I think it muddies things up for me rather seriously.

If Presner described what he meant by the above, I didn’t jot it down. But to me, it reminds me of the role of the cartographer in making a map. Curation isn’t about collecting every little datapoint about every little thing. That would be a task for Arthur Danto’s “Ideal Chronicler,” and, as Danto shows, the Ideal Chronicler cannot exist. Curators, like cartographers, make choices. They have biases. A curation can’t have everything just like a map can’t show everything. Cue Monmonier’s axiom that not only is it “easy to lie with maps… it’s essential.” The map and the curation both reflect the choices of the person behind the product—and that person’s omissions, agenda, and so on.

But, amazingly, this sort of known incompleteness, even if it cannot be totally grasped, resonated with a second theme from the first day of the Institute, which was the conflict between reflexivity and positivism. Martyn Jessop, who presented on issues facing geospatial research in the humanities (a presentation similar to his article from 2008 in Literary and Linguistic Computing), closes a section of his article by asserting, “Ultimately, the most significant contribution of GIS to humanities scholarship may not be as a positivist tool but as a reflexive one.” In his talk, he repeated this sentiment, and both times I did not understand what he meant by “reflexive.” Seeing GIS as “positivist” is pretty easy (and a possibly wrong… or, at least, conclusion-jumping… move), but I had no idea what the “reflexive” side would look like.

So I asked. Jessop answered that it involved collecting the data and looking at it, thinking about it. Reflecting on it. As I understood it, this meant a strategy of collecting data and not considering the day done once R^2 is calculated. This reflexive move then returned in Presner’s presentation. HyperCities invites the visitor to explore (or, perhaps, better, “play with”) the data, setting up playful situations that generate aleatory encounters that lead to future arguments.⁴ This idea persisted through to the final day, when Bethany Nowviskie presented on “graphesis,” a term I don’t know, but that I’ve described in my notes as a sort of “sketching” or “iterative graphical ideation/expression/inquiry” that generates, through the play of iteration, previously unseen strategies of argument.⁵

So “curation can be its own argument” incorporates reflexivity, play, the aleatory, graphesis. It’s iterative and unpredictable. Exciting. Oh, and it’s service-oriented and open to the public. All this against “analytical,” which is closed, limited, deliberate. A precision strike. Maybe I should reconsider my earlier pride from declaring that “I analyze” instead of “I curate”?

And maybe, then, the dissertation that is only a monograph might begin to seem antiquated, selfish, and, perhaps, even, problematically élitist?

1. These memories are intentionally a bit fuzzy and uncertain, since nothing was on the record, but I don’t think that, if the people involved were to recognize themselves in what I’ve produced here, they would complain. The details aren’t terribly important, but what is important is to show that this discussion stretched beyond just the presentations I describe below.
2. “Waste” would be the currently appropriate verb, as the mess sits as an .mdb, waiting desperately to get converted to a real format with a web interface, but that’s a project for another time.
3. I think that all of my Geoinst-related posts should be presented in this combative style. We had human (language) vs. machine (language), now this, and next will be, um… fieldwork vs. armchairwork? Yeah!
4. I’m not sure it’s fair to suggest, which is what’s implied by setting up this “reflexive” or “play” mode against the “positivist” one, that there isn’t a certain amount of play and experimentation with data in the positivist social sciences, but that’s not a terribly crucial point here.
5. The role of the random in project planning, inspiration, and execution is way outside of my expertise, but it seems like any and all projects involve a certain amount of randomness in the form of contingency: the agents are at a certain place at a certain time, and the like. As such, there is no dissertation in the world that does not have some stage of spaghetti being thrown against a wall that then moves forward from the contingent circumstances surrounding which spaghetti it was that, finally, stuck. The issue here is whether the play gets suffocated—straightened out—in the sclerotizing effort of the PhD candidate to make something Serious that can get the candidate Hired.

UVA Rotunda. (click for original)

Should a foreign language requirement for a literary studies PhD be fulfillable by a machine language? Or maybe even by a methods course (like a course in statistics, GIS, or some other competence in computational technology)?

These questions have been on my mind since I flew back Sunday morning from an energizing time at lovely (since it reminds me of high school) UVA, where I participated in the NEH-funded Institute for Enabling Geospatial Research in the Humanities, run out of UVA’s impressive Scholar’s Lab. It was a great time, I appreciate the NEH for tossing the cheese the Scholar’s Lab’s way, and I especially appreciate all the hard work the UVA people did to pull off a seamless little three-day event.

Over the next few days (read: weeks), I hope to write up more about the various things that went on at UVA, but the one that has been needling me most came from a quick flash of Twitter conversation on the topics in the lede, prompted by Brian (@briancroxall), who wondered whether GIS methodologies should count toward a “methods” course requirement in a PhD program, as it already does at Penn’s History Department. This then led toward the suggestion that, perhaps, proficiency in a machine language should be accepted as fulfillment of a foreign language requirement in a PhD program. Ryan (@ryancordell) even wished he had spent the time cramming for a French exam learning Ruby, instead.

As I mentioned while recusing myself, I’m kind of a militant about foreign languages (the more the better), to the point where I would consider basic Spanish proficiency a requirement of any specialist of American (U.S.) literature.

But after thinking about it, I realized that the two languages, human and machine, are, bizarrely, and in most mainstream use scenarios, totally non-comparable. Human language proficiency, at the literature PhD level at least, is measured in ability to read.¹ The mere fact that my classmates can (and have) fulfilled their requirements by taking “Reading German” or Latin (dead language!) courses demonstrates that the key skill being taught is reading.

Someday being able to read a machine language, however, is never the goal of learning it. As Matt Kirschenbaum writes in his appeal to humanities students to learn to program (my emph):

Many of us in the humanities miss the extent to which programming is a creative and generative activity… Programming is about choices and constraints, and about how you choose to model some select slice of the world around you in the formal environment of a computer. This idea of modeling is vital.

From this promising beginning, Kirschenbaum comes out in favor of using machine languages as substitutes for human languages in fulfilling requirements.²  If one is studying contemporary American literature, in which code can appear, he argues, it has certain value.

But Kirschenbaum’s understanding of the value of the foreign language requirement seems misplaced.³ In the beginning of his article he criticizes those who view what we literary scholars do as little more than correcting spelling and grammar in comparison to those who think wrongly that computer scientists do nothing but fix bugs in code. So why, then, is he willing to imagine the foreign language requirement in such reductive terms when he says explains that (my emph),

Knowledge of a foreign language is desirable so that a scholar does not have to rely exclusively on existing translations and so that the accuracy of others’ translations can be scrutinized. One also learns something about the idiosyncrasies of the English language in the process.

Really? That’s it? Nothing about alterity, about imagining different conceptual schemes (pace Davidson), about forcing a disruption in one’s comfort zones (and comfort Weltanschauungs)? I should learn French just so I can take Massumi to task on how he continues the tradition of the translation of “agencement”? That seems a bit… thin.

Furthermore, when he pushes human and machine languages together, since knowing code helps one understand novels in which code appears, he is working against the very point of programming with which he opens his piece: programming is world-making. Sure, if I know C, I can read a novel that has pages of C. But I’m interpreting it and re-generating there, not generating tout court, as I would be with a “Hello World” program.

So the confusion in the Kirschenbaum piece gets reflected in the argument on Twitter: a willingness to compare human and machine languages, perhaps only since they are both called “languages,” emerges. This gets amplified, then, in the academic world; looking back at Penn’s history requirements, we see that they treat foreign language acquisition as a “competence” in a “technical area” (if I read the guideline correctly), akin to competence in GIS or statistics. Language is foregrounded–the technical option seems available more to US history scholars–but it is still treated as part of the same piece.

But history is not literary study; in an English department, I maintain, knowing French and python are completely different beasts serving two different masters in the scholarly process. The former relates to how stuff goes into the scholar (reading), while the latter relates to what comes out of the scholar (analysis). Exceptions are obvious, as one could publish in French and analysis usually involves quite a bit of feedback looping, but I think that I’m more or less right for the huge majority of cases.

So, on the one hand, I do wish that I could get the sense that computational methods training had greater appeal in the humanities.⁴ But, on the other, I do not think that it is entirely appropriate to view methods and foreign language as either/or objects in a course of study, when, as at my university, a student needs to show only proficiency in a single foreign language.

Given how my colleagues dismissively look back on their foreign language requirement (something like “I took ‘Reading German’ and remember none of it” is common), this distinction is probably unnecessary. In fact, as far as real-world skills are concerned, a much more lucrative future can be built up in a quarter-long course on Ruby than in the first quarter of first-year French. Human languages require care and attention, like Gabriel Conroy’s going to the continent to “keep in touch with the languages.” But though machine languages also benefit from practice (like any skill), it strikes me that it’s much easier to get back in the coding swing.

Still, maybe the question gets a bit more provocative if we consider the language requirement not at the PhD level, but, rather, at the undergrad level. Undergrads at my university can fulfill their math requirement by taking intro level CS courses that include vocational programming courses for web development, but the math requirement is certainly not the same thing as the foreign language requirement, which can be met in a dizzying array of ways. I cannot imagine that my university, which has made a huge deal over the past decade about expanding study abroad opportunities, would start accepting perl proficiency as meeting the foreign language requirement. And I’m not sure that’s bad.

Now it seems like we’re still comparing two things, human and machine languages, that can only really be compared in coarse ways, like through simple, macro-level questions of time management or speculation regarding future employability–two things, I think, that fall out of consideration at the PhD level (ha!).

So back to Kirschenbaum as I try to wrap up this wandering. HASTAC recently published a response to his article that reminds readers that one shouldn’t think that CS familiarity is demonstrated by machine language proficiency. This is certainly true: I know how to build and manage a GIS, but I don’t think I’m at all a geographer. I can code, but I’m not a computer scientist. And I don’t think my two years of formal French language study make me a scholar of French.

The article continues to discuss NLP, suggesting that the humanistic disciplines and CS have quite a lot in common. This is certainly true, too: my old CS roommate and I were often working on similar projects from different angles. But that then cuts out the final leg from the table that is Kirschenbaum’s argument. In describing programming as world-making, Kirschenbaum compares the coder to Jane Austen, a formidable world-maker herself of reasonable renown. Yet CSers working on NLP aren’t making worlds. They are investigating the world around us, just like us boring literary scholars.

1. I fulfilled the requirement with Russian by showing that I was reading Russian literature and engaging in literary discussion of the texts without translation.
2. He hedges, hardcore, by arguing that, in fact, it should be not only case-by-case but implicitly appropriate only for programs that require two foreign languages, which is not, as I mention, the case at my university, where one foreign language suffices.
3. I am working under a bit of a fantasy here about what the foreign language requirement should do for PhD students. I know the reality is just about always different.
4. My committee certainly had no issues with my pursuing further foreign language acquisition, or learning statistics, or taking a full year’s worth of GIS training. The promise to learn and work with GISes was written into my dissertation proposal, even!

Anyway, Jockers divides the digital humanities into two groups: those that work on digital objects and those who use digital tools on objects (that may or may not be digital). In my department, the former group (“A”) seems to circulate around the New Media Workshop, and the latter… well, they don’t really seem to have a high profile.

I like to imagine myself as part of the second group, the group B, to the point where I’m not positive I’d even consider those who are exclusively in group A “Digital Humanists,” but that’s probably not a fight worth having. Point is, I’ve got some of the cred,¹ but one glance at the latest Literary and Linguistic Computing indicates that I’m still a total n00b.

Still, the reason I bring this all up is that Jockers closes his post by pointing out that there is nothing quite as “new” about us Type B Digital Humanists as current outsider articles seem to be suggesting. As he writes:

I came to utilize computation in my research not because the siren’s song of revolution was tempting me away from my dusty, tired, and antiquated approaches to literature. Rather, computational tools and statistical methods simply offered a way of asking and exploring the questions that I (and others such as those pictured above) have about the literary field. What has changed is not the object of study but the nature of the questions.

That’s right. I mean, it’s tough to say straight that one did not pursue question x instead of question y without some kind of fantasy of “revolution”; there’s a reason a dissertation proposal includes a “state of the field” section, after all. But what I extrapolate from Jockers’s position is dead on: you can’t digi yourself up in a useful manner until your questions need digi techniques to get answered (cue my usual complaint about Wordles).

And that Wordle complaint is worth cuing again, since the point deserves being repeated over, and over, and over, and over…

1. Membership in the Alliance of Digital Humanities Organizations, utilizing tools like R, ArcGIS, and GeoDa in my work

The problem of the war machine, or the firing squad: is a general necessary for n individuals to manage to fire in unison?

Our applications for dissertation fellowships are due at the end of February, which means that I’ve had my sole chapter on the mind quite a bit lately, even while wasting most of yesterday playing Doodle Jump. So it was excellent timing that Ryan Cordell should come out with his post on using Cultured Code‘s Things to help manage tasks over at Profhacker earlier this week. Putting together a dissertation is such a crazy project with pieces all over the place, that I feel that it’s necessary to rely on some sort of software tools just to save on redundancy.¹

Anyway, Ryan makes the case for using Things as the general to get the firing squad to shoot in unison, the dissertation being both rhizome and a tree. But more precisely, his post is pretty excellent, because it provides a lot of use scenarios specific to academics–the sorts of scenarios that are completely obvious, but only in hindsight. Making a task with a deadline from an emailed CFP? Duh! Yet I had never thought of doing that, myself. Mostly, however, the post reminded me that not only had I spent a considerable chunk of change on Things, but that adding tasks can be done via the quick-entry popup in any application, a popup that includes references to the url you were looking at when you summoned it!

See, over the past half year or so, Things had become nothing more than the repository of the list of movies I wanted to see along with reasons why I wanted to see them. My inbox, instead, became my to-do list. This is suboptimal for three main reasons: first, even with Gmail labels, it’s hard to distinguish tasks from each other or see how some might actually be related. Second, an email won’t tell you when the deadline has passed. And, third, and most important, your inbox is constantly getting augmented by stuff that isn’t necessarily tasks, meaning that it is an unnatural blend of to-do list and, well, inbox. And that’s a mess.

So Ryan’s piece encouraged me to trim my inbox down to just four emails as well as separate out into discrete tasks the long lists I emailed myself about things I needed to do (read or think about) for my dissertation. Now, in Things, I can order them, group them, etc. And most importantly, I feel like I have some control over the process of thinking about a dissertation again.

Things is not perfect. Sharing the task database using Dropbox is rather risky. Further, syncing with the iPhone is a pain: it doesn’t sync when cradled–one has to open Things on the iPhone while it’s also running on a computer and both are on the same subnet. But why would I ever launch the Things app on my phone when it’s on the same subnet as my computer? If we’re on the same subnet, that means I’m at my computer! It would also be nice, and this may be possible though I just don’t yet know how, if I could have stuff automatically get sorted into certain areas based on the tags I give; everything tagged “dissertation” should land in my “Dissertation” area. And, finally, I would love if I could sync Things with Gcal, as right now I’m sometimes torn between putting something in Gcal and putting something in Things. Usually I opt for the calendar if the stuff is date-specific.²

The standard caveats apply about task managers (and any organizational tools, including DevonThink): it’s only useful if you use it, and what’s useful for one might not be useful for another. But for the time being, I’m back!

1. My handwritten notes of ideas I get in bed or something, for example, are often very, very redundant. Software helps eliminate that duplicated work by showing one that it is duplicated work.
2. Apparently, the application can sync with iCal. I think I knew this, but there was something about the implementation that I did not like. Still, this means that I can most likely use iCal as a bridge to Gcal, but that has been messy in the past.