A blog post about Web Mercator that didn't make the cut.
My opinions have changed since then. I no longer judge Mercator-bashers for being high and mighty about projections, I judge them for deeming technology a side-effect when it's actually the ruling term in the equation. If you don't understand how tiles work, and how they make maps a fundamentally different technological process than WMS, then it will be impossible to understand why Mercator is so popular. That is; Mercator didn't 'beat' other projections, XYZ tiles beat WMS.
Maps are the new typography and "Web Mercator" is the new Comic Sans.
Now it's true that Comic Sans is less than a great font - it's been mispurposed into being used at huge scales and with anti-aliasing that it for which wasn't designed. The saving grace of the Comic Sans debate was that everyone knew at least one font that was better than Comic Sans.
And so, Web Mercator has become a trendy thing to criticize, and it's easy to construct an argument against it, though they fall into three categories:
- A newspaper / institution / government requires all maps to be in X-small-town-arbitrary-projection-and-datum
- I think Mercator is 'bad' for various reasons, mainly dissolving down to "it enlarges some countries" or "it doesn't look nice"
- My newspaper / government is actually producing sub-centimeter data and wants to print out maps that we can use rulers on.
Let me count the ways.
It's true that Mercator royally distorts size. That isn't the point of the Mercator projection: it's for worldwide, zoomable maps that are useful in navigation and at many scales. In schools we have equal-area which are much better for learning about what the globe looks like, if you can't afford a globe.
But this unending faith in perception belies the reality that people can't properly judge the area of circles, much less many-shaded, crooked, discontinuous countries. Is the 10% area distortion of a cropped Mercator map really to blame? And couldn't one just save the development budget and use a list if they want to know the sizes of countries in descending order?
But how many people ask for a less distorted map in the actual places they use maps? By this I mean people - not people in group #1 who are blindly following a 'delivery requirement' or those in #3 who somehow own printers.
As far as I can see, we could stand to have more distorted maps: because the real problems here are:
- Geographical maps are rarely an efficient way to present data in small spaces
- Most people are using maps on their cell phones, seeing data on an 'infographic map' that's similarly constrained, or looking for driving directions.
By #1, I mean that opening Yelp and searching for food trucks gives you clusters of close points along a single street and empty space - around 10% of the display is put to good use.
Why do I write this? Besides trying to push the debate beyond just culture arguments, it's that the current approaches seem uncreative. We've seen that Apple is working on paper napkin directions, and the tools are there to give them a shot, but nobody is doing it.
I've tried to contribute something with Project It Yourself, but it's only one angle of the picture - we need some ideas for dynamic maps. The scale adaptive projection developed at Oregon State is neat, though blending between existing projections as the answer to future needs seem quite like putting wheels on a horse.