I have to say, I think the typography / lettering / label placement of the NatGeo map is significantly better than the Imus map. It's easier for me to see what's what, and it looks less cluttered somehow, probably helped by having freer choice of orientation. But I won't deny that there's more information in the Imus map.
As I read I jumped ahead to look at the pictures and was all impressed about the attention to detail on the map on the right - look at how "Kettering" is curved so it fits. The labels are clear and not fuzzy.
I was impressed with his map.
And then to my surprise I discover that his map is the one on the left!
Maybe in a large printed version his is better? I don't know. All I can tell from the photos is that the computer generated one is better.
The label for "Kettering" in National Geographic's map is distorted (to "fit"), set in smaller type, crosses two road paths, and eliminates more smaller city labels than Imus'.
The irony of this is that these are all very common constraints to graph layout algorithms --- simple type, minimized crossings, label size and placement --- and the label you called out as "better" pessimizes them.
Add to that the fact that Imus' map subtly indicates Kettering's position in a population center (the yellow blob) and gives topographic and foresting information for the area with a clear contrast between the western and eastern sides.
Be careful. The Imus map is deceptively detailed (again: forestation levels), and is not shown in its best light in a PNG clip next to a simplified National Geographic map. You're right; the Imus map looks muddy and harder to read. That's because it's doing more, and is thus not as well suited to a laptop screen (or a clip in a web page) as National Geographic's.
Natgeo maps are largely hand drawn on a computer. Labels are manually placed. There are a few industry rated label placement engines (Maplex), but for published maps, manual placement is needed.
I believe this might be due to sharpening issues and noise and fuzziness due to compression in the photograph/scanning of the Imus map. When scanning/photographing text one must sharpen more aggressively than usual, and also use higher quality/lossless compression.
It seems the author of those digital pictures is not skilled enough as to faithfully reproduce paper documents digitally.
Seems a little unfair to judge the piece of work on a screen when it is probably better viewed "in person". Also a fairer test might be to try and use the map for its intended use rather than criticizing specific attributes directly (if you see what I mean). Anyway, monumental effort, 2 solid years work. I hope its a success for the guy.
I agree that the National Geographic map looks better superficially. To my eye the labelling on the Imu map is more functional. You are drawn to larger or more important places first, and other labels are less intrusive. The approach adopted by National Geographic is much less effective in this respect.
The greatest paper map of the United States I've seen is not this one (though I ordered this one), it is "USAtlas"[3] created by Richard Saul Wurman using early Macintosh design tools like Adobe Illustrator 88 and Aldus PageMaker 3.02 on a Mac Iici.
Wurman is cited in "Building Legible Cities 2 Making the Case"[1] and, oh yeah, created the TED conferences[2].
The biggest innovation for me as a frequent long distance driver in the late 80's and early 90's was having every page the same scale. But the clarity of information on a given page or city was unprecedented, was easily glanceable while driving, and is still unmatched to this day though Google Maps' data view comes close while offering more details. But this was hand drawn.
Here's Cincinnati. Notice the state borders vs rivers:
Wil Tirion plots star maps. At one stage he was using Illustrator on a Mac. Graduated from a drawing table and Rotring pens. I find his maps easier to use in the field than a laptop or netbook with 'red screen'. Its the overview aspect I suppose.
Quite a lot of published maps pass through Illustrator or similar at some stage for careful human retouching and intervention. For larger scale (quantity, not map scale) stuff then you tend to have a source data set containing the raw geospatial data and feed that through a series of processes to handle styling, annotation placement and white space management, and even division into map sheets.
Engineering maps tend to be generated far more directly from the source data, and as the quality of things like Google maps improves it gets harder and harder to meet people's cartographic expectations without human intervention.
The Imus map does look nicer, but as a computer geek I am more interested in how one could capture the heuristics he uses and better automate the process. Surely all (or most) of the label placement, typography etc. ideas he uses could be captured in a set of criteria that could improve the state of the art in computer-mapmaking. This is similar to what Knuth did to maths typesetting.
>Surely all (or most) of the label placement, typography etc. ideas he uses could be captured in a set of criteria that could improve the state of the art in computer-mapmaking
Yeah. The following quote really annoyed me:
>He used a computer (not a pencil and paper), but absolutely nothing was left to computer-assisted happenstance.
Well, if Buddy spent two years writing the algorithms to dynamically place the names on the on-line maps, how is that placement "happenstance"? Just as much effort went into it as went into the hand-drawn map...
I'm originally from Cincinnati, and the National Geographic Map makes better choices about what to show. Norwood and Newport are important. The Roebling bridge is not.
I'm in Chicago and agree generally; National Geographic's choice of small town labels is better curated. My guess is that Imus worked from population statistics and optimized to fit the most "people" in the labels he chose.
That said, it seems straightforward that National Geographic would have better-curated detail (they presumably spent millions to build theirs with a team of people). Meanwhile, the actual map layout Imus constructed is pretty amazing, and National Geographic's is merely legible (look at the river label in Cincinatti again, and the confusion it caused on this very thread).
After reading the article, I'm amazed that it's only $13.
I was thinking to myself how much I wanted it, but assuming that it would be selling for artisinal prices, and I'd have to think about whether I wanted to spend a couple hundred bucks on a map.
This map is small: 4x3 feet. I would like a large map, maybe four times the area (8x6 feet), but with the same density of information. One could represent four times the number of towns, streams, roads, etc. I want to hang it on a wall, and be able to read it for hours, always discovering new places to visit.
I tried going to the author's site (https://imusgeographics.com/) to see what the map costs just out of curiosity, but at this point the site seems to be down due to exceeding its quota. Too bad.
I just emailed the author of the article (Seth Stevenson) to see if he would let the mapmaker know about his map being mentioned on Hacker News. Maybe we can get some contact info for those of us interested in purchasing the map. I know I would be interested in one.
According to the WayBack Machine, it wasn't available for sale on his site the last time it was updated in 2008 (however individual states are relatively cheap: $10 - $35 depending on quality).
I Googled the URL and found a link to https://github.com/patdunlavey/imusgeographics.com
Seems this gave me access to the entire site - from behind the scenes.
If you go up one level you get the rest of Pat DunLaveys sites that he manages.
Appears to be a lack in security and decisions.
I only see one little image at the head of the article, but the text refers to what seems to be a few closeup shots. I bet it's Slate's damn tablet view dropping them on the floor. God I hate Slate's tablet view, it's slow and makes reading HARDER.
edit. Found the Normal view link. Yep. Tablet view is missing images. YAY.
Looking at both maps, I see only one thing: they are both so much worse than new interactive maps... There's too much information and not enough "whitespace". They're so generic, it's hard to find some specific thing on them. If I'm looking for highways, I'm not interested in areas of Chicago. If I'm looking for known places in Chicago, not only is 5 items too short, it's forcing even more information on the map where it's not really needed. Regarding the careful placement... let's see for example: According to Imus's map Plymouth is north of the river, while according to National Geographic it's south of the river. In reality the river goes through the city. If the place is not right, what's the meaning of a "better" placement of it's label?
Regarding typography, I don't like the italic text on Imus's map - once you have many lines in random directions around your labels, it's not trivial to say if some 3-4-letter name containing many round letters is italic or not (is ORD in Chicago italic?) Here it's trivial to figure out from context - when you're trying to determine a size of some city, it's not.
Yes, I'm being negative about this map (maybe a bit too much), but apart from art, is there really a good reason to produce maps like this anymore? I don't agree with the angle this article takes:
> For one thing, that zooming capability means the makers of a single digital map are forced to design dozens of differently scaled versions. This severely limits how much time they can devote to perfecting the layout at each zoom level.
It's actually better because it can show the same information after zooming in, but you can still read the "important" information while zoomed out. Reading any text on this map while standing back from it must be a much harder exercise.
> Imus argues that you can’t truly understand a place if you only use zoomed-in maps on teensy screens. [...] Looking at Imus’ big, richly detailed map offers a holistic sense of what America looks like—how cities spread out along rivers, forests give way to plains, and mountains zigzag next to valleys.
Give me an option to turn each category on and off and I'll see the relations much better. Being able to filter out noise would give many more possibilities of exploration than a "generic map with absolutely everything on it". If cities and rivers are what you're interested in, many maps will provide you exactly that information. Trying to figure it out from a map with 10 other layers is harder.
In short, I disagree with this article trying to prove that "old-style" paper maps can be more useful or readable than interactive, zoomable maps with customisable layers. In my opinion those are always more useful than paper maps. </rant> Not to discredit the work that went into this map of course. I do appreciate this map as art and see what the author was trying to achieve. I just disagree strongly that it's useful nowadays.
I don't think the author is trying to say paper maps are better overall than interactive, just that there are aspects (some of them aesthetic) that paper maps do better, and it's good to be mindful --- not up-in-arms, just mindful --- of what we're losing.
The only clear claim of superiority I got from this article was that large paper maps are better at conveying the gestalt of their subject, which seems like a straightforward enough argument. Not a novel argument; Tufte made a cottage industry out of it.
I don't understand why a large paper map is better at conveying "the gestalt of a subject" - I appreciate the work Imus did from the perspective of craftsmanship/artwork - and, indeed, it might be a great paper map - but, I'm guessing that 95% of the readership of HN uses digital maps for more than 90% of the time (and, who knows, it might actually be 99% and 99%.) The fact that your map will place your location on it, give you a reasonably decent satellite overlay, and is zoomable to the level that you need it, and will put _dynamic_ layers like weather, and/or traffic with _real time_ data. There is a reason why I haven't thought of turning to a paper map in 5+ years.
It would be interesting to see where a paper map performed better than a digital map.
> and, who knows, it might actually be 99% and 99%.
It seems like I'm part of the 1%, at least once in my adult life :) I'm working on a personal project that involves placing medieval (East-European-based) villages on a Google Maps mashup. In order to do this I rely on 2 paper road-atlases of my country, one published in 2007 and the other one published in the 1930s (I use the latter because lots of villages had their names changed in the meantime or they had just plain disappeared).
At the beginning I did try to rely solely on Google Maps to locate those villages, but there were many cases in which the villages' names were just placed wrong, or they didn't show up at all (not to mention GMaps' latest bug involving diacritics, which makes letters like "ț" or "ș" show up badly on their maps).
Also, having a more general, zoomed-out view of the area you're interested in is really, really useful. It gives you a better understanding of the geographical context (are there mountains around? forests? is it a densely populated area?), which I could never reproduce when using digital maps. Maybe this is all caused by the fact that I'm sort of a map freak and I'm not using maps only to get driving directions, my mother actually bought me a World atlas when I was 5 or 6, and I've been in love with maps ever since, but then again, this article seemed to be targeting people like me.
A little OT, that 1930s road-atlas I mentioned has a very interesting story: as a result of the Ribbentrop-Molotov pact my country lost part of its territory, which used to be mapped in the atlas I own. After WW2 ended and the "friendly" Soviet Army installed a Communist government in my country, it was not politically safe anymore to have those (now Soviet) provinces as having been part of my country, so a kind person actually used scissors and glue and cut out said provinces from the atlas, by very carefully following the new, Soviet-approved border. It all kind of looks like this: http://imgur.com/n6c9y and http://imgur.com/qvcvM .
Just curious, what is the mashup you're working on? When researching historical places, I've often wished that Google Maps natively supported a 4th dimension so that changes over time could be recorded. Everything from "I want to see the businesses on this street as of 2001" to "Zoom out and see European boundaries as of 1100".
> Just curious, what is the mashup you're working on?
Putting on the map the villages of Wallachia (http://en.wikipedia.org/wiki/Wallachia), a historic Romanian province, by order of when they were first documented in the official archives. The project sits here: http://sate.maglina.ro/ , it's based on Django+Solr (there's a "Filter" menu at the top right), it's in Romanian, of course, but it's more or less self-explanatory.
I also created a couple of (I'd say cool :) ) heatmaps based on the data, which you can see here: http://sate.maglina.ro/heatmaps/ , giving a general idea where the early medieval settlements had a greater density. (for the heatmaps I used gheat, of course, http://code.google.com/p/gheat/ , which is a very, very cool project).
> I've often wished that Google Maps natively supported a 4th dimension so that changes over time could be recorded.
I feel you on that :) But, I'll have to admit, it could turn into a very controversial thing. For example, one of the reasons for which I started this small project of mine is that lots of historians (Hungarians and Romanians, mostly) have debated whether present-day Wallachia was inhabited by Romanians continuously (as the Romanian historians insist that it happened) or if said Romanians emigrated to Wallachia from South of the Danube somewhere around 1100-1200 or later (the Hungarians' view on the whole story). I was hoping that by putting on a map the Wallachian villages by order of them being officially mentioned it could clear up some things, at least for me :)
Please don't judge digital maps by Google Maps... Be aware that you can easily make a digital map that fits your needs if you have the geo data (like eg from OpenStreetMap).
I placed a large 4'x3' map of the U.S. in my kids room. They're too young yet to have a concept of a map (computer or otherwise), but I want them to be able to say look at Calfornia, where we live, and take the time to look at each and every state in between to say the State of Maine. With a very large map you get much more of a feel for the position of states relative to one another (or countries for that matter) and then you can also zoom in (get closer) and see what lies within.
Yes, it is very cool in particular to do that on an iPad. But the entire feel is lost. If one looks at the whole US at once, most of the details are lost. Then as soon as you zoom in it is difficult to appreciate the spatial relationship between states as one "journeys" across the country.
That reminds me, I also need to buy a globe. Or is that to be replaced with an animated 3d monitor representation as well?
I used to love the globe my parents gave me as a kid. It had raised surfaces to indicate mountains. I remember tracing mountain ranges and wondering about how tall Mount Everest really was.
When a teacher showed our class a flat map and tried to explain how this Mercator projection wasn't truly accurate because the world is a sphere, I remember a bunch of students being confused. I was more fortunate. I imagined the globe back home and got it right away.
I love digital maps for the efficiency in performing the task of getting from point A to point B, and I love globes for helping to visualize the physical world as a whole.
With a very large, very high-resolution screen, and the same graphic/information design expertise, I'm sure the digital map will do just as well as paper.
The issue is simply that paper is (typically) more dense than pixels, allowing your eyes to take in patterns across a larger area, and allowing you to continuously and effortly adjust your vantage point.
Nobody could possibly argue that paper is superior to digital for weather or traffic. Digital maps are, in general, probably superior to paper maps.
Don't forget cost. That big dreamy screen rolls in at $500+ while the wall size map equivalent costs a mere $50 or less, requires no computer to be switched on and is instantly zoomable.
I used to have a Thomas Brothers LA basin wall map. This was a 5x7 (foot) map that went down to street level within the whole Los Angeles area. It was fantastically detailed and beautifully printed.
I could quite literally spend an hour studying that map, and I learned a lot from it. It would draw you in, in a way that digital maps can't quite do, at least not now.
I can also spend hours browsing Google maps, but it's more fiddly.
Besides the "draw you in" property, the wall map has the advantages of showing far-away regions at equal scale, and always showing places in relation to each other.
It's the difference between a reference book and a textbook: a reference book is better if you already know what you are looking for. In a mass of complex multi-dimensional data, it's easy to show it however you want, but difficult to find patterns between variables you thought unrelated.
At first glance Imus' map is more accurate as well. He has the location of the airport (CVG) south of I-275, and the location of Vandalia, OH west of I-75. NatGeo is wrong on both.
But like moultano said, he missed Newport and Norwood.
The OS maps are lovely, and interesting for some of the subtle detail they include. I noticed recently (while using an OS map as a background raster for some much more detailed data) that the roads on the 1:25000 and 1:50000 scale maps are drawn with accurate widths derived from the more detailed survey data.
The source data the maps are produced from is also pleasingly detailed, they sent man to the remote part of Scotland where my dad lives to properly survey all new houses. Openstreetmap has that sort of data for central London, but not for anywhere like the whole country.
Bing Maps UK has OS Explorer and Landranger coverage as an option. It is extremely impressive, as the sheets (which are drawn to the UK National Grid) are subtly distorted to the projection used by the Bing mapping (for example the grid lines are not all straight up) and routes can be calculated against the layers - even though they're raster data.
The first closeup, of Cincinatti is an odd choice because the Imus map does not color the river on the border blue. It was not clear to me that it showed the river at all.
The Chicago closeup has a confusing combination of a time zone line and dotted line. Still it is nice to have the time zone lines.
Paper maps are my favorite thing to put on the wall. My favorite right now is a map of the Appalachian trail. A tall, very narrow map, it cuts diagonally across the traditional map of eastern America, giving a very different perspective.
Joey, actually the blue line on the National Geographic map doesn't depict a river whatsoever, it's an arbitrary color for a state border.
Because the river actually IS the defining border for the state, on the Imus map the river is highlighted in blue underneath the dark-green that depicts a state border.
Reminds me of a post (from a now defunkt blog) explaining the design decisions of google maps that gives it superior readability compared to yahoo and bing:
They're fairly basic and common sense Edward Tufte tips but it's interesting to see what a difference a collection of tiny UI tweaks make in the aggregate.
Really well done. One of my cherished wall adornments is a Buckminster Fuller Dymaxion Airocean World map. I'd really like to see what Imus would do with that projection.
In the sense that a map like this is the product of decades of full-time cartography experience, experience which is inevitably going to get less and less available as more and more of the work is automated, it seems fair to be concerned that the art is endangered.
Side note: this article is a good example why websites should Not make a touch version. While the article looks more iPad-like in the iPad version, it also strips out the map example images, making half of it quite pointless. And off course switching to the desktop version of the site sends you back to the touch version for the next page, as it detects your user agent...
Reading this, I am most shocked that there are still that many mapmakers -- the age of exploration is long past and I guessed that the few updates that are necessary (renaming the cities and updating country borders, add a new street here and there) was pretty much done by a couple of guys on a computer, somewhere.
I think you have to take into account that making maps are about choice and map designers spend a great deal on figuring out what should be on those maps and how, not just what data do we have access to.
Just because there are web-developers does not mean you don't need a designer :)
Even if you exclude the presentation side of map making most utility companies have a small department whose job is simply to maintain the landbase data, getting updates from multiple sources (developers, local authorities, etc.) converting and sanitizing that data, and making sure it's all there as a reference for the engineers so they can plan how to wire up a new estate and connect it to the existing network. There are various national (and international) spatial data initiatives to come up with standardized data models for all of this to make these processes easier, but I think that may increase the appetite for data. I know of applications pulling together hundreds of such sources, and I think they are going to become more common.
There are a lot of different things to take into consideration when creating a map with many scales for online usage.
Most heavily used digital maps (e.g. googles, bing, open street maps etc.) use pre-rendered tiles. However this is besides the point, as at each scale (or set of scales) usually a different data source is used so that a specific amount of generalisation can be applied.
For example, a small scale map (showing a larger land area) would show not only fewer streams than a large scale map, but would also not show every little bend and curve; this allows for the appearance to be less muddy. However as the user zooms in they would expect to see higher detail in stream direction.
Often this is still something that is difficult to reproduce with an algorithm, often being left to selecting different datasets based on scale.
The article says he used a computer to draw the map. Does anyone know what software he used? Is it something specialized or just Photoshop? I couldn't find it in the article.
I could relate to a lot of his painstaking efforts after building the maps for Stormpulse "by hand" albeit digitally. Hundreds of little decisions all along the way.
