 So you know who this man is, which he has innumerable accomplishments that are public, starting organizations, sending ideas out into the world. So I'm not going to say anything about those at all, I didn't say that they're interesting. What's far more important to me, I must confess, is in all honesty, what a fabulous colleague he is. Berkman has had the great pleasure of having Ethan for a one-year fellowship for many more years than the provost would like to know about. And it is only because he has come with so many fabulous ideas with such kind of commitment to the community of people, both inside this building and outside, to work on the ideas that we share and the common values that we have, both in the highest order ways. And as just a colleague and friend to so many of us. And this place would not be anything like what it is without Ethan's presence. He is one that, in many ways, defines fellowship for me. So, thank you, sir. That is incredibly kind. Part of the reason for Colin's kind introduction is that he knows that I'm actually planning on blaming this talk on him. I went to Colin about two weeks ago and I said, should I give a talk on something I actually know something about? To which Colin basically said, no, no, no, that's too easy. Talk about something that you don't know anything about. And so, hence the disclaimer. Which is to say that generally speaking on Tuesday afternoons in front of this room, you have someone who is astute and learned and deeply versed in the field. Talking about something that they have thought up thoroughly and they're wishing to present to the world. That is not today's talk. I'm instead gonna talk about something that I know nothing about. I'm trying to figure out, have been thinking about for a couple of weeks. And I'm gonna do this mostly because I'm hoping that people will react to this and say that doesn't make any sense or that makes sort of sense. And here's where you might go with it. Because it's a set of ideas I'm playing with. And so you are heartily encouraged to heckle, ask questions, etc., etc. Let me also say that one of the subtle themes of this talk is to try to get you to go visit the Peabody Museum. And can I get a show of hands on who here has been to the Peabody? All right, that is really good. If you don't have your hand up, add this to your calendar in the near future. The Peabody is sort of just down and to the right. And it's one of the absolute treasures, not just of Cambridge, not just of Harvard, but of the state of Commonwealth of Massachusetts. I managed not to go there for five years until I finally got dragged there by my best friend who said it's astounding we have to go see it. He wanted me to see the beautiful blown glass flowers that are there. But I actually fell in love with something entirely different, which is the Pacific Islands Room, which is on the fourth floor of the Peabody Wing of the Museum. For those who haven't been there, the Peabody is basically a museum, two museums. It was largely designed around 1900, and basically none of these innovations and museum design have ever caught up with it. It's basically a great way of seeing what a museum looked like at the turn of the last century. And so what it means is you have this amazing collection of Pacific Islands artifacts, largely collected not by anthropologists or ethnographers, but by oceanographers who happen to be in the South Pacific. And the labels are sort of wonderfully uninformative. You'll get a label on something that essentially says where it was found, what the date was, the name of the collector. No information on what it's made of, the function, anything along those lines. You get to work that out on your own. It's sort of a temple to early 20th century collecting. And so I fell in love with an object there. And this is not the exact object, but a very, very similar object. And this is called a rebelib. This is a form of map that comes from the Marshall Islands in the South Pacific. It was collected roughly around 1900. And I'm going to show a couple of these to give you a sense for how these maps work and why I think they're interesting and how ultimately they're going to connect to the internet. So bear with me. This is a related sort of map. Also the Marshall Islands, also the same sort of period of time. And this kind of gives you an idea of what's going on in these maps. Ignore the horizontals and the verticals. They're just there to sort of hold the thing together. These shells are islands. And around those islands, we have these curved rifts. And those curved rifts represent ocean swells. So what's basically happening is as the currents of the ocean are moving past this island, that island is creating an interference pattern. You get a very distinctive set of swells coming off that island. And you can see, you actually get two sets of them. You get the sort of dominant swell. You get the secondary swell that comes off of it. This sort of information was incredibly important if you were navigating in the Marshall Islands in a dugout canoe in 1900. You are navigating a chain of coral islands that's about 250 kilometers wide and maybe 300 long. They rarely come out of the water more than 10 meters. You can't see them more than about a kilometer away on the horizon. You have no other navigational equipment other than these sorts of charts and songs. And you sing paddling songs as a way to measure duration as you paddle from one island to another. So if these charts were designed to do were to teach young navigators how to figure out how you would find an island. And the answer is if you started seeing this interference pattern, the interference pattern between the dominant interference coming off the island and the secondary one, you knew you were fairly close to an island and you could follow those interference patterns and actually find your way to the islands. So this particular map called a Matang would probably have not been carried on a boat. This was mostly to teach navigators how you navigate. These rebulibs might well be carried on a boat because they show an entire island chain. The shells are showing where the islands are. The diagonals in this are showing you where sort of dominant ocean swells are. So you know where you are both in terms of the islands that you've seen but also in terms of the swells coming off of the islands and the sort of dominant swells in the ocean around there. Now we know about these charts mostly from one German captain, Captain Winkler. We don't know his first name, but he spent about 10 years in the Marshall Islands and he tried to find out as much as he could about these charts. And they turned out to be basically secret information. He had to go into closed huts with the doors and windows shut so that no one would hear the navigational secrets and he had to ply people with glasses of sack to get them to tell them about it. But Winkler wrote this paper in 1901 that basically represents the state of the art of what we know about these maps. And everyone who's written about them subsequently comes back to this paper now more than 100 years ago. One of the reasons we come back to it is that the accuracy of these maps is incredible. If you look at this map that's up here and you actually look at the map of what the Marshall Islands actually look like, if you orient yourself, Majuro is here, Majuro is there, tip this one over a little bit because North is actually there, you'll actually see almost all of the island positions are pretty much exact. And one of the things that's interesting to me about this is that our current maps of the Marshall Islands suck. They're really awful or at least the ones that are pretty easy to get your hands on, you zoom in, you quickly get the Google sorry, no more information there. So I actually think it's pretty cool that, you know, going back in time we may have actually known more about this 100 plus years ago than we get from Google nowadays, but that's not really what I wanted to talk about. What I actually wanted to talk about is the fact that these shell maps, these rubber libs, do a really interesting form of mapping which is mapping infrastructure. They're not just mapping the geography of this island chain, they're mapping something that isn't immediately geographically apparent, which is how the ocean is flowing between these islands. This was an utterly critical piece of infrastructure for people living in the Marshall Islands at the turn of the 20th century. How do you ride a swell from one island to another? Were you likely to have a more or less coherent sailing path? So having this information makes this map far more useful than if it were simply a positional map. And it turns out that these maps of infrastructure are something that are fairly common. I've seen a whole lot of them actually sort of concurrent with this one which is to say at the turn of the 20th century. So we've got lots of maps coming out, you know, all throughout the 19th century. What you start seeing as you start heading into the turn of the 20th century are some really interesting maps that focus on infrastructure. So this is an 1891 map, it's a German map and the focus of this map are telegraph cables. So we start getting international trans-oceanic telegraph cables in the 1860s. By the 1890s, we have telegraph cables that are reaching literally every corner of the world. And for someone who's sort of an Africanist by training, I'm fascinated by the fact that Africa's surrounded. We haven't surrounded Africa with fiber optic cable. But this is a really interesting historical legacy since telegraph cables basically were the precursor to running an effective colony. If you wanted to run your colony efficiently, it helped a lot to be able to have regular communication with whoever was running it for you rather than doing it every couple of months when someone came back with a report and a steamship. So you can think of telegraph lines as very closely paralleling spreads of colonization. It has a lot to do with the spread of the British Empire, which is why you've got Australia and New Zealand in there. But it's amazing to me to think that we can get such thorough global coverage by 1891. There's other forms of infrastructure that get very, very important. Steamship routes. There's an amazing wealth of steamship maps around about 1900. This particular one is from the American Express Company. It gives you what were the standard routes of the day, the distances associated with them. You can quickly start to see where some of these nexies, where some of these hubs of shipping start to emerge. And of course there are railroad maps. And if you're looking for an amazing collection of infrastructural maps, there's a gorgeous collection of official railroad maps, which is linked to on my blog recently. Although one of the interesting things you'll see about this is that a lot of these maps of infrastructure really peak around about the turn of the 20th century. We see them into the 1950s and then we see a real sharp drop off. These maps turn out to be in many cases surprisingly hard to find of contemporary current day infrastructure. And I'm interested in that and that's one of the things that I wanna talk about in this. But one of the great things to think about with this is just incredibly pervasive the railroads were at the turn of the century. And you think about how this contributed to something like the rise of Sears, where you end up with Sears-Rowbuck catalogs going out to every one of these little towns and suddenly the fact that you're connected by railroad means that you are connected to what nowadays we would think of as e-commerce in those days mail order. But suddenly the capability of having something that if it existed in a big city like Chicago, you could actually have out here in rural Kansas as long as you had that form of connectivity. So I'm really interested in the ways that the 19th century starts becoming a century where we have infrastructure that's connecting not only rural and urban, but is actual international infrastructure sort of for the first time. We see the rise of transatlantic shipping. We see London and Bombay brought together by the telegraph. We see major advances in canals in railroads. Refrigerated shipping suddenly becomes very, very important. That's one of the reasons that New Zealand gets connected. You suddenly have this ability to export produce which otherwise didn't make any sense. One of the ones that I threw up there that may not make intuitive sense is the opening of the Chukka-go mercantile exchange. This is important because the Merck is the first major agricultural exchange to standardize. And what standardization meant was you could make a contract with someone in Nebraska for winter wheat and you could say, I wanna buy 100 bushels of grade A at the following price and you actually had a decent chance that you were going to get what you thought based on the fact that there was an objective grading system. So getting these sorts of standard systems that come up with something like the Chicago mercantile exchange allows you to start having industrial commerce on this much more abstract scale than you'd ever had before. Now, on top of this infrastructure, we see a wave of globalization that in some ways actually outpaces globalization that we see nowadays. I think a lot of us sort of think of ourselves as being in this sort of unique moment of globalization. There are a lot of historians that point out that the wave of globalization from, say, 1880 to about 1910 far outpaced what we're seeing right now. And one of the simplest statistics on this is to talk about global mobility. If you looked at what percentage of people in the world were living in countries other than their country of origin, it crested above 10% in 1913. Despite jet travel, despite the internet, despite container shipping, despite all the stuff that we know right now as the legacy of globalization, it's a hair above 2%. It's actually really hard to see from the US, because we're actually at quite a global high point in the US right now. We're actually very close to where we were in 1890, which was our peak. The images, by the way, are Alice Island, which is also a turn of the century institution. This is sort of a wave that's all happening at the same time. So basically, I'm looking at maps of late 19th century, early 20th century infrastructure, looking at maps of steamship routes, of canals, railroads, and telegraph cables. And embedded within them are these really interesting stories about the emergence of integrated financial markets, the emergence of real multinational corporations that are being run from a central office rather than being run by a company agent who's essentially fully in control of them, migration to a scale that's even unprecedented at the moment. As we start looking at infrastructures of today, which is to say, container ship routes, airplane routes, energy pipelines, whether they're electric, whether they're oil and gas, whether we look at telecommunications grids, there are some similar interesting stories to be pulled out of them. I think some of them actually resemble these stories. I think some of them are stories that it's actually very hard to sort of get our hands around what they're actually going to look like at this stage of the game. So I want to talk a little bit about state of the art in mapping. And here I'm going to reflect the fact that I am an absolutely not a cartographer. I'm an internet user. And my sense of state of the art is the fact that we've all gotten completely obsessed with Google Maps, which has done something really, really interesting, which is it's let us get really high quality satellite imagery localized to almost any point in the globe, almost no delay, zero cost. And we've gotten downright used to being able to pull up locations and getting slightly time shifted actual real world representations. In fact, we're so used to this that you'll see roughly once a year a slew of articles complaining about what you can't get off of Google Maps. So IT security wrote this thing, 51 things Google Maps doesn't want you to see. The article was savage because it turns out that two thirds of them actually are visible on Google Maps. They've just done their searches badly. It also turns out that Google Maps, generally speaking, doesn't have any control over this. Google is buying satellite imagery from people. They didn't for a while have control over whether or not you could see Dick Cheney's house from the air. It was actually being blurred out at the satellite imagery level. But these are sort of wonderful lists to see. First of all, what are sensitive enough things that say the government of the US or the government of the Netherlands goes to a satellite imagery provider and says, I need you to blur that out. But also this sort of sense of entitlement. This idea that we are entitled to a view of the world, a map that's so good that we can get down to this sort of level. Now, to be very clear, we don't demand this of everything. I mean, here's the current best imagery I could find on Google of Lagos. Lagos is an itty bitty little city. It's 8.1 million people, meaning the only US city that is smaller than is New York City, probably the most important city in sub-Saharan Africa. And the vast majority of the imagery you'll get from Lagos is clouded out. And it's clouded out because no one's demanding better imagery of Lagos. The way this imagery gets taken is low flying airplanes go over, they take photos, and you get whatever cloud cover there happens to be. You might need to send a plane multiple times if you want to get a full cloud-free panorama, but that gets very expensive. And so when you see a cloudless view on Google Maps, which is probably what you're all used to, you're actually looking at fairly expensive data. In this case, you're actually looking at pretty cheap data because there isn't a lot of demand for it. But my point on this is that state-of-the-art on Google Maps is pretty damn high. In fact, it's high enough that we start having people raise privacy concerns. The gentleman in the foreground of this image is only known as Bill. Bill had evidently had something of a tough night. The night before, he went back to his house in Adelaide, Australia. He didn't quite make it all the way. Google Street Maps managed to capture him. He went to Google and explained that he thought this was an invasion of his privacy. They eventually agreed, but the image does persist on the net. This is, in terms of geographic mapping, a level of mapping that we're growing to expect, and we're trying to figure out how to adjust our sense of privacy around it. I'm interested in this question. What are we actually expecting out of our maps of infrastructure? A lot of us who've worked in the internet got very, very used to an enormous number of maps trying to explain what the internet was all about. And for those of us who started working on the net in the early 1990s, we saw lots and lots of visualizations like this. This was NSFNet. This was the National Science Foundation, Giant Academic Research Network. It was very easy to build these maps because the data was widely available. Everyone was playing around with visualization techniques. And frankly, in the early 1990s, it was pretty realistic to map the internet in a way that was comprehensive. You could actually get a sense of what was going on. You could probably learn something meaningful from the intensities on these maps and how things were connected. Pretty quickly, in net mapping, we hit a point of just complete incoherence. Very, very beautiful, but very, very hard to extract anything from this map that's meaningful. This is the Opti Project. This was the last major effort I was able to find of someone trying to map the net by going out and saying, can I spider all live Class C addresses? Who's attached to who? Can I run lots of trace routes out there? And you get the beautiful, filigreed visualization, which my guess is pretty much useless. And one of the reasons that I'm guessing that is that network mapping in terms of building these whole comprehensive maps to the internet dies off in roughly 2004. The projects where people are collecting a lot of these maps seem to have that roughly as a death date. One of the bigger projects, Chethwick and Burke, move in and start doing smaller private maps on a cost basis for people. This Opti Project flourishes around 2002, 2003. Their last map, which this is an excerpt of, is in early 2005. My guess on this is that we reach a level of complexity on the internet where maps are no longer particularly helpful. If anything, they're sort of obscuring more than they're actually telling us and figuring out how we architect these. That's a little different from most infrastructure maps that you're able to sort of pull out of the public domain these days. I spent about half a day recently looking for detailed maps of electrical transmission lines. And my rule was I had to be able to find it on the internet. I had to be able to access it. I didn't necessarily need it to be under an open license, but I needed not to pay money for it. And this was about as good as I've done. And this is a map that's now about 10 years out of date. It's a rough overview of the three major regions of the electrical grid with a sense for how big the connections are. You can do much better than this, but you're going to be buying maps from vendors. And these vendors aren't cheap. If you want to understand where oil and gas lines are, you're probably going to end up buying maps from a magazine called Petroleum Economist. And this is a magazine that has a several thousand dollar a year subscription cost with access to all of their maps. And they're not so beautiful. They give you oil and gas pipelines. They give you facilities. All this information is available. It's out there if you're willing to pay the money for it, but the money's substantial. Just that map we were just glancing at is roughly 160 pounds. I think if you buy a lot of them, you can get a slight discount on it. I complained about this on my blog. They sent me one. I thought that was a very nice way of sort of acknowledging the problem without actually sort of addressing the core behind it. If you're not willing to either complain or shell out hundreds of pounds, the alternative is to look for stuff in the public domain. And one of the best for this is a really wacky site out there called Cryptome. I don't know who has heard of Cryptome. Cryptome is this sort of wonderful, strange dark corner of the internet. It's put together by this guy named John Young, who sort of gained notoriety for putting up a set of maps intuiting where Dick Cheney's secret location was. And he's had sort of perpetual fights over what information he can and cannot put out there, what server it's going to be on. If you search for the eyeball series, you won't actually find them. You actually have to find this URL eyeball series. And it's this wonderful collection of maps that go from the extraordinarily paranoid, like looking at all the possible air bases that might shuttle people to area 51, to the very, very helpful, like where are proposed natural gas terminals within North America. But it's very strange. You start looking for these maps of infrastructure and you find yourself wondering if you're doing something criminal. And in fact, in some cases, you may be in danger of doing something criminal. Sean Gorman, by the way, Sean is the hero for this talk today. So memorize that face. We're gonna come back to Sean again. Sean was a PhD student at George Mason. And Sean was getting a degree in geography. And Sean had a pretty interesting research project. The idea was he wanted to take publicly accessible maps of the internet. This was 2003, by the way, before people stopped mapping the internet. So he wanted maps of fiber optic cable of US cities. And then he mashed up these maps with the locations of major American corporations. And as he matched the two of these together, he started seeing extremely sensitive choke points where if you wanted to shut off large chunks of American commerce, all you really needed was to dig in the ground and separate the cable. Needless to say, not everyone was thrilled with this research. And the Washington Post ran a wonderful story on this kid in 2003, essentially saying that as Sean started presenting his data, people started throwing more and more classifications on it. By the time Sean was finishing his dissertation, the only room he could work in was in a sub-basement at George Mason with two keypad locks on the doors and a shredder so that he could shred everything when he went out. In this article, there's this wonderful quote from Richard Clark essentially saying about the dissertation, he should turn it into his professor, get his grade, and then they should both burn it. Now, let's be really clear about this, right? No secret data, no top secret classification. This is publicly available data that's been mashed up. Information on economic infrastructure on top of telecommunications infrastructure. But by putting the two of them together, it was viewed to be so sensitive that he literally found his dissertation more or less classified. So there's an interest in shift, I think, in how people think about infrastructure. I feel like in these 19th century, 19th century and early 20th century maps, they're almost fetishized. It's almost a way of sort of looking at this and saying, here is the modern, here is this connected global world that we're going into. Now the reaction is this reaction of panic, right? How is this information going to get used to hurt us? The security concerns behind these maps of infrastructure are that they might represent a way to figure out where the vulnerabilities are in society, where one should put the wire cutters and take down civilization. I've got a slightly different concern, and my concern is basically that I believe we're living in a world that's best characterized by the connections we're making on top of this infrastructure. I think what defines who we are, who we know, what we do, what we buy, what we eat, so on and so forth, has a lot to do with what's made powerful by these infrastructures, made possible by these infrastructures. And I worry that basically, the only time we ever hear about these infrastructures are when they fail. And to give an example of this, and this is one of two examples I'm stealing from Esther in this talk, so thank you, Esther. There's been this recent standoff between Russia and the Ukraine over the flow of natural gas. And a lot of the natural gas to Europe flows through Ukraine. And so when Russia and Ukraine fight, the rest of Europe gets really, really cold because Russia's ultimate weapon is to simply turn off the flow of gas. And so as these stories start hitting the news, you start seeing these maps. And these maps are pretty wonderful. They are, unsurprisingly, from Petroleum Economist magazine, the only people that I've found who consistently have access to these maps. BBC, I'm sure, is licensing them and it's them putting them up. And for this sort of brief shining moment during which we're all shivering, we suddenly become aware of the reach and the extent of this infrastructure which has become absolutely essential infrastructure. If you're living in Romania or Bulgaria, in some of these towns, there's a single central heating plant which is getting all the gas from these gas pipelines. You shut down that heating plant and no one in town has heat. It's not like in New England where you can't pay your gas bill and you go hang out at the heating center. You basically lose the entire town at the same time. My sense is that the only times we pay serious attention to these infrastructures and the implications of them are when they break. And so I'm now going to make a totally inappropriate analogy and we'll see where that goes. I wanna talk about phrenology which also turns out to be a wonderfully 19th century topic. Before we go any further on phrenology, I wanna point out something very helpful which I think explains American politics. Right in the center of the brain is hope. I think phrenology actually helps explain the current erection. I think it's a nice way to put this stuff forward. Phrenology is a nice punchline to a joke nowadays but there's actually some really interesting things historically about it. When Dr. Gall started providing phrenological diagrams at the beginning of the 19th century, he was doing two things that were unambiguously right and fairly revolutionary. He was one of the first doctors to say the brain is the center of thought. That wasn't all that clearly established at the time. He went a little further and said the brain is a differentiated organ. There are parts of the brain that have different functions independent from one another or at least distinguished from one another. Different areas of the brain affect different aspects of a person's existence and cognition. Which also turns out to be true. The place that he got wrong was actually in trying to apply the scientific method. And so very scientifically, he recruited hundreds of volunteers. He measured their skills in great detail and he attempted to correlate them to characteristics in their lives. So if he measured brilliant people and they all had a bump over here, he concluded that that bump had to do with brilliance and it ended up on the phrenological map. And the reason phrenology made some sense at the time was that it was very, very difficult to study brain function. How do you figure out whether one part of the brain is associated with one aspect of cognition or the other? The alternatives you have were pretty ugly. And generally speaking, the way we figured out about brain function came from catastrophic brain injury. So poor Phineas Gage, who was working as a foreman building a railroad, railroads infrastructure, 19th century, building a railroad in Vermont was tamping down explosives when he made a bad hit on the hammer, caught a blasting rod, which came up through his cheek, up through the center of his head. And bizarrely enough, Phineas Gage then stood up and said, I think I require a physician, which was rather surprising to everyone who had assumed that instead he was dead. And what actually ended up happening was that the physicians were able to remove the rod, staunch the bleeding, bandage his head. He lost his left eye and he obviously lost a good chunk of his brain. This was about an inch and a quarter thick rod. But he was able to continue living and working for an additional 12 years afterwards. And what was really interesting about Gage is that once he sort of recovered, people started noticing that Phineas Gage actually didn't resemble the Phineas Gage that they knew. Phineas Gage had been a very well regarded, well respected railroad foreman, a guy with really sound judgment, a leader of men. And he was a son of a bitch after he caught a rod through his head, which a lot of people sort of assumed might happen if you caught a rod through your head. But enabled people to start making some conclusions about what the frontal cortex was involved with, i.e. temperament, judgment, making rational choices. What you ended up with was someone who was far more spontaneous, far more dangerous, ended up eventually leaving for Chile for reasons no one seems to understand in his biography. For a good chunk of the 19th and 20th century, what we learned about the brain came from these moments of catastrophic failure. We learned an enormous amount about what the corpus callosum does from severely epileptic patients where doctors would go in and sever the corpus callosum in the hopes of stopping epileptic seizures to pass from one to another. This is, in the grand scheme of things, a lousy way to make a map. A much better way to make a map is what we do nowadays. And we make maps of the brain by doing PET scans. We introduce radioactive oxygen or glucose. We look to see where in the brain gets activated during certain patterns of thought, and then we're able to make judgments about what area is involved with learning a new skill versus exercise with a practice skill. What's interesting with this is that rather than mapping the structures of the brain, the physical structures, we're now mapping by mapping flow. We're looking for the flow of blood, we're looking for the flow of oxygen. And my basic contention is that if you want to map globalization, you've got to find a way to map flow. If you just map this infrastructure, which holds us together, you end up fooling yourself a lot of the time about what's actually happening. The difference between what's possible on top of an infrastructure and what actually happens turn out to be substantially different. So if we're trying to understand air travel, we can look at a map like this one. We can understand where you can theoretically travel from London on BA. This map doesn't do a particularly good job of distinguishing between destinations. It's not going to tell you that it's pretty easy to fly to Corfu in the height of summer, but it isn't all that easy to fly there the rest of the year. It isn't gonna tell you which routes are very frequent, which routes are less frequent. It's not gonna show you where the traffic's actually going. For that, you need a map that actually shows you flow. So here is one of those maps. This is a visualization done by the Zurich University of Applied Sciences. And one of the things that comes out of this immediately for me are some patterns that wouldn't be intuitively obvious. Obviously, there's a very clear European east coast of US corridor, but there's also evidence of amazingly concentrated domestic travel within the US and within China. There's a corridor that I hadn't realized at all that goes between Brazil and Portugal. There's a 24-hour map. So this map is pretty fascinating. We may have to run it a couple of times. This is 24 hours. So we just saw late night with the red eyes going over to Europe. Now we're seeing sort of midday coming in the other direction. I find this map endlessly fascinating. I've lost like half a day sort of watching this map and trying to pull different patterns out of it. Things you don't notice until the third or fourth time through it. There are no south-south flights, none. There's like one flight a day between Brazil and South Africa. That's about it. So these patterns sort of come out over time as you end up watching these things that end up being very, very different than you would get simply out of mapping the infrastructure. Here's another example. Here's a map of San Francisco that's made purely as a flow map. And the way that this map was made, well, back up, have a second. This map was made by taking the data from flightstats.com, which is basically a global database of all air traffic. It's not what's actually happening, right? This is all simulation. So this is on a day when every flight flies as scheduled. This is what air traffic should look like. What's that? You can find it on my blog. Or you can, if you search for Zurich University of Applied Sciences and AirMap, you'll find it in no time. This is a map that's actually built from real life data. The folks at Stamen Design, which is one of the better known web design firms nowadays, did a partnership with Yellow Cab in San Francisco. And Yellow Cab for many, many years has had GPS receivers on all of their cabs. This helps you figure out if a cab has gotten stolen. It also helps you figure out whether a cabbie is goofing off or actually looking for fares. They agreed to share this data with Stamen. And Stamen took this data, which was basically every cab, every minute it's fairly precise location, and smeared it out into data traces. So essentially, intuitive the path from all the points and drew lines. And these lines over time revealed the city of San Francisco. What's interesting is that they reveal a slightly different city of San Francisco than you might otherwise intuit. If you look at a street map of San Francisco, you're gonna figure out some of these big avenues, some of these big streets very quickly. These are the two major highways coming into town. This is one of the major bridges coming out. You're not gonna pick out these traces, which actually come up quite clearly in the map. And the analyst there suspects that these are both a way of avoiding traffic, but these are also pretty common patterns of people going to the hospital. Which is a pretty common use of taxi cabs. You call a cab, you go and visit the hospital. So you start seeing patterns that are not apparent out of the street grid. The other thing you start seeing over time are blank spots on these maps. And blank spots can mean a couple of different things. Obviously if there's a park there, there probably aren't a lot of taxis going into it. But these blank spots actually are neighborhoods. They just happen to be neighborhoods where you have a very, very low chance of hailing a cab. Specifically that neighborhood down and to the right is Hunter's Point, which is a historically African American neighborhood. It's actually quite a vibrant neighborhood. You're never gonna see a cab there. Is essentially what this visualization is telling you. Despite the fact that you have a large number of people living there, no particularly good public transit, it simply falls off the map. And this is the difference between having this sort of street map which essentially is showing you what's possible and a flow map which is telling you what actually happens. So it's certainly possible that you can hail a cab at Hunter's Point. It just for a variety of reasons most of the time doesn't seem to happen. So one of the reasons I think we don't have a lot of flow maps out there is that basically to make a flow map you have to put people under surveillance. So if you think about sort of the most common sort of flow map most of us have seen today which is the traffic enhancement on top of a Google map, the way you figure out where the traffic gems are are by putting up a helicopter or putting cameras on the ground, watching the density of cars go by and saying there's a traffic jam at this location at this time. On that top blurry level that doesn't look a lot like surveillance. When you get it down to a level of hey let me see if I can find Colin's car on this map then it suddenly looks a whole lot like surveillance. And I think one of the difficulties with mapping a lot of this flow isn't just the security concerns involved with mapping infrastructure. I also think in many cases there's a set of privacy concerns that end up coming from it. But I think there's certain maps that would really help us understand the world if we could actually map them in terms of flow. It's possible, although not easy, to get a map of major shipping lanes. I have not seen a map that actually shows me where the volume of container shipping is going. And one way that you would build that is by doing many, many, many, many times what the BBC is doing. The BBC has a project called The Box where they've taken a shipping container, they've enabled it with GPS, they're tracking it as it makes its way around the world. And this ends up being a pretty interesting narrative. They've gotten a very good story out of this. They've gone through some of the major shipping lanes in the world. They've gone through the Gulf of Aden, which has lately made a name for itself. They've gone through the Straits of Malacca. They don't actually end up going through the Panama Canal, they end up going over land. It's a good story, but I find myself sort of wondering what happens if you expand this to a global scale? Instead, we have weird ways of sort of intuiting these maps. This is the live pirate map, which is not in fact live. And in fact, pirate is a fun term to talk about as well. But these are the incidents of piracy or breaking and entering on ships as reported to the ICC in the course of 2008. And this map doesn't really show you where the global shipping routes are, although there are some interesting things you can intuit from it. This piracy cluster exactly focuses on the Gulf of Aden, which is one of the key shipping routes out of Europe. And you see a really strong concentration on the Straits of Malacca, which is another massively important one. So one of my questions on all of this is when we can't otherwise measure flow, are the ways to sort of intuit it by backing your way into the data? So the basic distinction I'm trying to offer is that infrastructure maps, as distinguished from geographic maps, can tell you what happens in it, what could happen in a connected world. I can look at the airplane map and discover that I can get from Logan to Fiji in about two-ish days. It's not gonna tell me how likely I am to go from Logan to Fiji, how many other people are going from Logan to Fiji, but it will show me that it is in fact possible to do it. A flow map shows you how often that's actually going to happen. How many people get on the airplane at Logan and end up in Fiji at some point in the future as opposed to going to Texas or Budapest. An intent map, which I have no good examples of because I don't really understand how you map people's intent, would show you what people wanted. And just to distinguish intent from flow for a moment, I'm gonna steal another example from Esther, who points out that despite the fact that she has no special fondness for Frankfurt, she spends an awful lot of time there. And this is because she flies Lufthansa. She's often going either from Chicago or Boston to Budapest. And if you look through Lufthansa, you basically have the option of going through Berlin or going through Frankfurt, and anyone who's done it will tell you that Frankfurt's a little bit easier. And so you end up, if you map Esther's flow, seeing Frankfurt as a sort of central point to her existence, but you misunderstand her intent in the process. So when we look at these flow maps, we have to understand that some of what goes on in flow is an artifact of how the infrastructure works. Getting the intent often requires a little bit more of a level of abstraction there. The other breakdown, by the way, is that infrastructure and flow often don't match one another. We're all good market economy-based people here. We sort of assume that people who are providing transportation services or data services or energy services will see a market and will respond to it. And that isn't always the case. And particularly those of us who work in Africa know that this is almost never the case. So this label here is Wagadugo. It's the capital of Burkina Faso. Burkina Faso is desperately poor, has been screwed over by colonialism and now by economics and by natural disaster and so on and so forth. Lately it's getting screwed over by infrastructure. And one of the reasons for this is that if we looked at the flow of good services and people from Burkina Faso to the rest of the world 10 years ago, it all went out through Abidjan, which at that point was one of the biggest parts in West Africa, a stable, well-developed country. Unfortunately, Cote d'Ivoire has been at Civil War for quite a while now. The Civil War involves Northern forces who are seen as aligned with Burkina Faso. It's become increasingly dangerous for Burkina Bay to even enter Cote d'Ivoire. And so what we've got instead are people in Burkina Faso coming in huge numbers into Ghana, which is stable, peaceful, high infrastructure, into Benin, which is another country that's in reasonably good shape. And infrastructure that doesn't reflect this at all. If you look at Air Burkina, which is yet another failing West African airline, they've got amazing connections to Paris, to the colonial capital. They've got some connections to other francophone cities, mostly to Abidjan, no connection at all to the country that's now their largest economic power and frankly, lifeline. And this is unlikely to change, actually. Air Burkina is now filled so badly that they could bought by the Aga Khan Foundation. And the first thing that Aga Khan did was insist that they open up an air route to Akron. They're working on it. But in the meantime, actually, we have Ghanaian entrepreneurs who are looking at this and saying, wait a second, I understand the flow. I understand how many people are commuting between Akron and Waga, taking 36 hour bus rides. I see an opportunity here. I can buy small aircraft and I can actually build a successful business out of this. There's enough disparity between flow and infrastructure that if you can figure out how to produce that infrastructure, you can actually become very wealthy in the process. So here's why I'm interested in this. I'm interested in Fiji water. I'm always interested in Fiji water. I've been writing about this stuff for five years now. Fiji water actually comes from Fiji, which is sort of insane when you think about how difficult it is to ship from Fiji to the US. It's grown to become the second largest brand of afforded mineral water in the US, only behind Evion. And it's a ludicrously successful marketing campaign in commodity. But what I care about here is that it's a great example of what the French economist, Daniel Cohen, calls imaginary globalization. And Cohen's contention is that we're in a much less globalized world than we actually think we're in. It's far, far less mobile than it was, even in 1910 or so. And the reason we tend to think it's so global is that so much stuff that we encounter is global. It's very, very easy for us to encounter atoms from other countries. We've actually developed that infrastructure rather thoroughly. We're much less likely to encounter people from these other countries. And I would go further and add we're extremely unlikely to encounter perspectives, points of view, so on and so forth. But basically, Cohen's idea is that this global stuff blinds us to the places where we're not connected. And he's interested in this in terms of economics. He asks people to sort of estimate what percent of stuff you consume comes internationally. People tend to overestimate by a factor of three to four. There's still an enormous amount of domestic commerce. And you can actually see that in that airline flow map. You can see how much more important domestic travel is than international travel. I've started calling this the Friedman fallacy. Tom Friedman writes big sweeping books about how the world was globalizing based on visiting highly globalized people in highly globalized cities and highly globalized business class. And it's very, very easy to get blinded from that to fail to understand the ways in which we're not connected, which has basically been the focus of my research lately. So what I'm going with this is I'm starting to ask the question, what would it mean to build an atlas for this sort of world? And I started by referring to this as an atlas of globalization, and then discovered that there are half a dozen sort of anti-imperialist, anti-colonialist, anti-most-denational projects to do an atlas of globalization. And I couldn't care less about that. What I'm actually interested in is building an atlas of the infrastructures in the flows that sort of characterize life in a connected world. So I'm calling it an atlas of connection, and I'm starting to think about how one would actually go about building it. Fortunately, our friend Sean is already doing one of a sort, which is nice. Sean, when he got his PhD, he's become a big figure in the open source GIS community. And he's got a company called Forteus One, which has put out two very cool tools. One is called Finder, and it's big sets of data that have been geolocated for you, so you can put it on maps very quickly. He's put together a Java-based application called Maker, which lets you make maps like this. And this map took me about 40 seconds to build. And this is a map overlaying fishing imports and exports over one another. What you can see very quickly is that the Chinese export a whole lot of fish and import very little. The Japanese import enormous amounts of fish, export some a lot. Bizarrely enough, every African nation is a net importer of fish, including the ones with huge swaths of ocean coastline. If you know anything about the fishing industry, you know that there are huge trawler fleets that are often fishing illegally off the coast of these countries. It's become increasingly difficult for fishers in Africa to actually make a living on it. This map gets you part of the way towards what I'd really like to see, which is a flow map, which is to say, where does all that Chinese fish go? And when the Chinese fish this much more than they actually consume, where are they getting it from, where are the Japanese getting their fish from? It's a nice step in the right direction. Other maps that I think are going a lovely step in the right direction. This is healthmap.org, and this is a map put together by some folks at MIT, some folks at the Harvard Public Health School. This map goes out and grabs lots and lots of public data, news feeds, collections of health data from different countries, extracts the geographic information from it, throws it together into a map, and is able to make pretty good guesses at where there are outbreaks of disease. And so you can follow this and sort of follow reports of cholera around the globe. And you'll find the obvious ones that you know about Lex and Barbara, but you'll also see other areas where this is being reported. And in many cases, this is actually just being scraped from news feeds. But it's a lovely way of essentially saying, we have lots more sources of data than we ever have had in the past to figure out how to map. I'm involved with a project called Ushahidi, which is trying to invert who gets to make maps. Normally cartographers make maps, and it's a long historic process. We're generally better at getting geographic features than we are at getting, say, people. But Ushahidi essentially says, if you witness something, help us map it. And this is a tool that emerged for mapping violence in Kenya after last year's disputed election. It's now been used in South Africa to map xenophobic violence. It's being used in Gaza to map missile strikes and responses. It's being used in the Democratic Republic of Congo to help people, people who are on the ground affected by violence, send them reports via SMS and sort of collectively do this form of mapping. So I'm interested in figuring out how do we identify and map these sorts of infrastructures that we depend on in one fashion or another. I'm interested in figuring out how we build maps that do a better job of representing flow and not just infrastructure. And I have to stop and look at this map because it's so cool. This map is from 1974. It comes out of the Congressional Research Service. And this is an attempt to understand how electricity worked in the U.S. at that point. This is electricity consumption. This is production. And production is broken up by what it's produced from hydro-oil, gas, coal. And for states that are consuming more than they're producing, you can see who they're getting their energy from, who they're exporting energy to. I haven't been able to find anything close to this accurate for the U.S. national good. I actually found this looking for grid maps. My guess is that this sort of map, which actually shows you where the major transmission lines are, is probably many levels classified or cost several thousands of dollars. But what's lovely about this is that this tells you a whole bunch of different things that's otherwise very, very hard to intuit from a table of numbers or any sort of other way of mapping this. So I want to map flow as well as infrastructure. And I want my next slide to come up. I want to map what we know about and who we know about and who we're paying attention to. Those of you who know my work know that I've spent a long time trying to map media attention, what media pays attention to, what we choose to pay attention to. But I'm also really interested in ways that people are mapping their own personal relationships. That's a mashup tool that attempts to pull your relationships out of Facebook and construct maps of who you're most proximate to and who you're connected to. So ultimately, in the same way that when we look at these rubber libs, we learn a lot about what was important to people in the Marshall Islands. We learn about a way of understanding the environment that we're living in. We learn about forms of information that we simply don't have when we consider maps today. What would we get out of this sort of similar isles of connection? And I don't know. That's what I'm sort of having fun thinking about at this point. So thank you for your patience. It's hot in here. You have to speak so someone should crack a window though. Yeah, that'd be good. So you've got links to all this stuff on your blog. I put up a couple of posts in this recently. What I think I'll probably do tomorrow is post these slides with all the resources behind them. There's a couple hundred other resources. There's a huge community of people who are essentially doing strange maps of one fashion or another. And there's, in fact, the strange maps website. But what's interesting is I haven't found many great collections, either of flow maps or specifically globalization connected maps. There's a neat project that got started at Princeton to sort of do some maps about globalization. I don't know if you wanna say something about it. A lot of these maps are incredibly pretty, but as you described, they require a lot of data and analysis. So he ran into funding problems, never really continued. But I mean, my guess is that the first thing we would do if you were actually gonna try to build an atlas is you would just build a website. And you would sort of start pointing to what already exists out there and what you can lead people towards. And I'm in my disorganized fashion starting to sort of build some links. And every time I write about this, I find 10 more things that people sort of bring to me. Didn't mention the word network once. Or at least maybe once, but I didn't hear it because I studied social networks. Is there a reason for that or are you? Because some of the stuff is, yeah, social network stuff. I mean, I recognize the taxi map that has appeared in that. Any particular reason? No, I think that's probably a big blind spot in how I'm analyzing this at this point. The way that I've been trying to do this analysis is suggesting that there's infrastructures that we don't pay much attention to. There's behaviors that take place over those infrastructures that may not map all that neatly to what the infrastructure connects. I think my contention would be that networks help explain what the behavior is on top of that infrastructure. But I don't think I've gotten that far on sort of building these things out yet. And Judith is gonna now tell us what I just missed on that. Of comments among people on a social network site over the network map. And so I think there's a whole use for that there because part of the interest in something like that is those links are basically imaginary constructs. All these network maps are. So the whole question of what are the behaviors across networks to take it to the next step beyond the structure of saying, here's something with a connection, but once you wanna understand the meaning of these connections where the whole, feel not just of social networking sites, but of understanding networks and sociology in general, needs to do is understand how do you start making sense of the behaviors that occur among the people? Yeah, yeah. So I think it is a really important question. I think one of my questions in all of this is that most of the network mapping that I've seen sort of assumes that every connection is as easy to make as every other connection. And I think one of the interesting things about bringing infrastructure into this, you're sort of describing on that, laying it on top of the infrastructure, is that infrastructure may reflect in those cases much harder barriers, barriers of language, barriers of culture, barriers of nation, which are the stuff that I usually am focused on when I'm sort of thinking about network theory. So. At least I think that too, it's really interesting. Yeah, so I think my hope is that I'm gonna get there and maybe in another month or year. You got a question? Yeah, I'm very sympathetic to this. My goal right now is to collect all the information people pass on Facebook amongst themselves at the college here. But there does seem like a tower of babble air to this. Of course, if we could collect all the information on everything that moves, potentially be very smart. On the other hand, we'd be tasked with the relevant stuff. And at the same time, since collecting a lot of this information is very expensive, it's no surprise it is proprietary where it exists. So what do you really want if you could have your three wishes for any data? So one of the things that any map does is simplify. And maps probably are more valuable in what they leave out than what they actually represent. So I think the catastrophic failure of those network maps, right, those amazing sort of leaf-like network maps are that they're not leaving anything out, right? Essentially, you've got someone proving that their visualization tool is strong enough to show you literally every connection. Whereas if you could actually find a way to sort of pull the stems out of those maps, you might find something pretty helpful. One of the maps that has had a lot to do with what I worked on for the last five years is telegeography's map of undersea cables. And this is sort of the earlier map of how the internet works in terms of connecting one country to another. And the big thing that you get out of this map, which I use in almost every other talk I give, is showing that West Africa is connected to the rest of the world by a single cable. East Africa isn't connected at all. And so you start asking questions about how are Africans represented in social networks, and you have to start asking your question, essentially saying, well, East Africans are approaching this at a much higher actual cost to participate in these networks than anyone else. So somewhere between that, which is a really simple primitive, you're on or you're not, here's the one cable map, and this, which has reached a level of basically useless visualization masturbation, there's something in there that actually shows you what the opportunities are, what the barriers are. So that's one thing that I want. The second thing that I want is ways of being able to take those infrastructure maps and getting a sense of what's utilized, what's overutilized, what's underutilized, what flows happen outside of that infrastructure. And that's incredibly valuable commercial data, right? Don't you think there are 100,000 bankers? I think there should be 100,000 bankers trying to get that data. Certainly, if you want to figure out where to put a gas station, having the traffic flow is a really helpful thing to do. My sense, just taking a very cursory look at what's available on this, is that it's pretty easy to get the infrastructure stuff if you're willing to pay for it. It's possible to get some generalizations on the flow stuff, but I don't see a ton of flow map projects. So what I'm hoping for is sort of an approach to doing this which either involves unlocking some of that data or finding different collective collaborative ways of putting it together, because just based on my very shallow dive, it's pretty hard to navigate at this point and it's pretty disorganized, yes. This is an actually very nice part of the country. Yeah, so that's the first thing. The other thing is I'm working on electronic inclusion and exclusion topics is actually with regard to the ageing society. And what we're doing is actually coming up with a social network map of elderly people offline and online and see if we can find, you know, by comparing those maps, if we can find a clue to inclusion and exclusion. So, you know, when I read a paper that was just a few weeks ago, I had this moment of truth, like, wow, that's how we can do that map, right? But basically, it's, you know, when you have a research problem, when you run into a problem that includes some kind of, let's say, network or geographical or mapping problem, you just have that problem and then you try to come up with a solution. You know, that's basically how we did that. Did you find any information that guides you when you want to do a map? You know, some kind of meta guideline, how to do good mapping? I think there's a lot of good information out there on sort of what the basic principles are of what you might and might not represent in a map. There are fewer tools out there to make good maps. There's some very high-end GIS tools that are extremely useful for making maps. There are a couple of sort of simple mashup tools, but they're good for some things and not for other things. One of the things we're finding around the Media Cloud project, we're building a lot of maps within this. We're ending up using Google's visualization interface. It produces very pretty colored maps of the world where you can color each country, but it actually only does it on this really tiny little scale. And the reason we're using that is that the other alternatives to it are professional mapping libraries that are very expensive. One of the things that I'm really psyched about with the mapping project that I showed off, the OpenGeo system, is that it lets us do sort of full-screen-sized country color maps. I guess I feel like not only is a lot of this data locked up, not only are a lot of the end products locked up, but I feel like the tools at this point are actually pretty hard to approach. What's interesting is to see how much sort of pent-up mapping desire there is. Google Maps came up with this really simple and frankly crappy mapping interface that sort of says, here, have a map. You can put pushpins on it. And people reacted to this by going crazy and sort of found thousands of different ways to put pushpins on maps. There's so much more you can do if you can color, if you can add sort of area boxes on things, particularly if you can start mapping flows from one to another. So I think it's not just techniques. I also think it's tools. If you wanted to build this atlas, this would never be someone's sole project. This would have to be a collaborative project of people mapping what they thought was interesting based on data that they could access and then based on tools that they could get their hands onto. I really like the distinction you're making between infrastructure and flows. And I'm curious if you have an explanation for why there's a gap between the two. As you were just discussing, there is a financial or potentially a financial interest in bridging that gap and yet that gap seems to persist from what you're describing. So what's going on there? So infrastructure has a lifespan to it. And if infrastructure is built rationally at one moment in time, it may not be as useful at that next moment in time, right? So rationally, if the government of Burkina Faso built a railroad, rationally in the 1980s they would have built it to go down through Cote d'Ivoire to Abidjan. Unfortunately, really through no major fault of their own, that infrastructure would be functionally useless at this point and you have to figure out another way to go on infrastructure. I think there's a thousand and one reasons, a lot of which are sort of idiosyncratic to the situation why infrastructure and flow don't meet one another. I actually think a lot of the time it's an interesting reminder that we're far less rational than we think we are about building these things. The reason I got obsessed with it, this is basically chapter one of the book that I'm trying to write. And the reason I'm sort of obsessed with infrastructure and flow is that I think we tend to look at an infrastructure map that connects the whole world and say the world is connected. And when you actually look at the flow of it, you discover that we may all be theoretically connected but the actual flow is pretty modest. It's certainly theoretically possible to have these South-South connections, watch an airplane map though and it just doesn't happen in real life. You've got this infrastructure and it gets much, much more heavily used in some places than others. And so I sort of started investigating this topic with sort of a great deal of skepticism about these maps of infrastructure, essentially saying I think the maps of infrastructure are fooling us, we gotta get to the maps of flow. Increasingly, I'm thinking actually the maps of infrastructure tell us a ton. They tell us what the flows have been historically. They tell us what people thought the flows were or thought the flows should be because that's another reason you build infrastructure. No one built rail between Beijing and Lhasa because there was tons of commerce happening between the two. It got built because people were hoping there would be tons of commerce between the two and it was a great way to sort of send the strong message that that's where we want the Han people to go. So I'm much more sympathetic to the infrastructure maps now but I think precisely the place to sort of put the chisel and sort of break them apart is when an infrastructure map and a flow map diverge from one another. One hunter's point is fully connected by a street map but no taxis go there. Intent part of the mapping, which I think is in some ways the holy grail of this mapping. I mean if you look at a business and let's say you're opening a shop and they wanna figure out what hours to stay open. It's pretty primitive the way that you would figure out the optimum hours and sort of maybe do some market research, watch the gas station unit, watch the traffic go by and you wanna be open when there's a lot of traffic and then maybe ask a few people, hey how are the hours working for you? And if enough people say open half an hour earlier that'd make a huge difference, that's what you do. So the same thing would go for airlines or anything else and that's what in many ways market research tries to do. I guess my question is how, is there any way, because that is gonna be really difficult to do on a large scale. Is there any way to measure or analyze the differences between the infrastructure maps that we have now and flow maps that are maybe a little bit more feasible so that we could, I don't know, is there any way to get to what the intent might be going towards or sort of tease out the intent from what we already have? So one of the things that I couldn't show during this talk just because I couldn't figure out actually how to get the video off of a proprietary format into it and maybe I'll ask the geeks for help with that. BBC has done a really gorgeous series called Britain From Above and it's basically a study of the infrastructures that make contemporary Britain possible. And they also do a taxi map, very similar to the Stamman taxi map and what's interesting in their mapping of it is that they focus on the city, right? The central financial district and as the work day starts you see the main thoroughfares light up and you can see where the big roads that you would want the taxis to be light up and then as they start glowing bright red you start seeing one by one every other street in central London light up. And what's going on is that there's so much flow in the city at that point that you can't handle it on those main thoroughfares so the taxi drivers immediately start going on to other roads that are lousy roads for taxis but it's the only other way to get through. And so you're sort of watching this and you're sort of watching on one layer of infrastructure sort of coming into being flow showing you this sort of profound intent which is how many people need to move around the central city at that point and you can start sort of peeling away those layers and figuring out how those different aspects of it work. One of the biggest things that comes out of it is that you understand both why we needed congestion pricing in London and how it's completely failed. But I don't have any good answers to sort of a systematic way of getting intent from flow except to say that you can tell certain aspects of flow that are always gonna have to do with infrastructure. You're gonna be able to look and see what are the frankfruits of the world as wonderful a city as it may be that people find themselves going through to sort of match their intent. But ultimately, mapping intent gets really hard because people's intent gets shaped by the infrastructure that they have to use. I really wanna go to Fiji but I really wanna go there for about two days. I don't really wanna spend a lot of time. I'm not much of a beach guy. I just wanna visit the Fiji water plant, take a couple of photos and get back. It turns out that the infrastructure doesn't make that possible and therefore my intent is gonna be perpetually thwarted by the reality associated with it. So I raise it mostly because Esther was reacting to one of my first blog posts on this and basically said flow's not intent. And she's absolutely right and I don't quite know where to go from there. I think flow's much closer to intent than infrastructure but it's still a big break away from being able to get that sort of theoretical map of if I really could do whatever I wanted, what is it that I would do? It's such a thing as everything everyone really wants to do. You're shaped by what the cost is to me. Although what's interesting about it is that if you could find an intent, not necessarily my Fiji intent, but if you could find an intent that's widely shared but isn't possible within the infrastructure, you could figure out a way to otherwise build the infrastructure. The reason that airline company is able to found itself in Ghana and suddenly make money hand over fist is that they're able to monitor flows and other media from Accra to Waga and figure out that that North-South route is very badly served. I think it's pure politics because it's such a web of bilateral treaties that shapes international flight needs. Although what's interesting and I didn't bring it out in the talk is that actually the real money maker for that airline is Accra to Tamale which is in the same country. It's just two cities that are very, very far from one another have a great deal of cultural distance, don't have historically good road connections and it's the first airline to sort of do the big connections. Last question? My last question is fine too. Thank you everyone. I don't know, we're running back. I clapped it twice, that's fine. Sorry to just pull the trigger a little late there. Something that I've been working on thinking about, I haven't been able to even conceive of how to study it yet, has been how these sorts of infrastructure maps would play out if you want to map the internet as experienced by an individual. Here's a situation where the flow is definitely not intent because intent is you give an address, you have no idea where the server is, it just goes and does what you're trying to do but there is a sort of, there is an actual spatial thing that's going on and additionally, I mean the complication that I'm looking at is in the Chinese situation which is what I study, you've got a certain infrastructure wall that's not a wall but anyway there's certain characteristics of the infrastructure that are completely or not completely but somewhat independent of intent and I just wondered if you thought at all about these. Let me give you sort of a parallel problem that I've had the chance to study and I'm now sort of trying to figure out how to write about and so far it's not in my draft because it's so complicated that it's actually really hard to explain to people but the very quick version says if you're in Africa and you're trying to decide whether to build an internet exchange point, right, if you're trying to decide whether all the ISPs in your country should talk to one another, you need to know how much internet traffic is domestic versus international, right? So it's not a bad thing to have all your internet connectivity coming from the US or from Europe if 95% of your traffic is CNN.com and Google.com but it's a real problem if 40% of your traffic is email that's being sent to someone across town where you suddenly get trombone routing, you have to go from one ISP into the United States and then back into Ghana again. We don't have the data and I studied this for about four or five years, it was very, very difficult to get ISPs to sort of give up this data so that you could actually sort of map the flow of internet traffic so you could make intelligent decisions about IXPs. What we ended up doing eventually was sort of saying it's crazy, there must be some domestic traffic, let's build an IXP, we build an IXP, people's costs fall by 30% and we say, oh, I guess we've got about 30% domestic traffic. It's a terrible way to map but it's something that sometimes you have to do to sort of back into this. So I agree with you entirely, that question of the fact that because some countries are constraining access to the net in some ways, the entire map of the net changes. There's some work being done here if you talk to David and to Hal Roberts, they're looking at ways to figure out where are choke points on the net and in terms of sort of intuiting where the big routers are by looking at how autonomous system numbers are assigned. This gives you a way of sort of building a map of how the traffic actually flows rather than these sort of abstracted maps that we're looking at in one fashion or another. So I'll happily make that introduction or David can introduce himself if he'd like to. But we're gonna stop here because we've got another meeting here at 2 p.m. But thank you all for questions and for listening. I really appreciate it.