The New York Public Library holds a massive collection of historic maps, and the institution wants the public to help make sure these maps are precise and completely correct. The library has taken steps to make the collection much more accessible, digitizing 20,000 maps in high resolution and offering interactive tools like the Map Warper, which layers old maps on the current grid. Now, NYPL Labs is really digging into the details with its Building Inspector, an attempt to create a "time machine" of the individual sites in the city. The Inspector first launched last fall, giving users the ability to correct building outlines on maps, but the library rolled out a revamped version last week that allows users to add street numbers, classify buildings with colors, and fix incorrect building outlines. To test drive the system, we turned to Blurred Lines columnist Keith Williams, who is something of an expert when it comes to old maps and the ever-shifting boundaries of New York neighborhoods.
The finer points of "rectifying" and "vectorizing" old maps is something we'll leave to the geospatial programming wizards at NYPL Labs, but you can read an overview here. Suffice it to say that dealing with old maps is hard: even when the maps were drawn and colored perfectly, age does funny things to 19th-century paper, despite being stored in perfect conditions.
As it's been doing with the Map Warper for a few years now, the library is relying on "smart, motivated citizens" to help it classify old Sanborn maps, which showed individual buildings for the purpose of fire insurance. Don't worry about getting things wrong: the program shows the same image to multiple users, and won't take it out of circulation until 75 percent of them agree on their assessment.
If this task sounds boring, let us assure you that it's actually rather fun to see what used to exist in the city...
...particularly when the descriptions are questionable.
At your fingertips are four different
games types of inspections:
Check Building Footprints
Did the program's algorithm properly identify a building? Sometimes it works, sometimes it doesn't. Part of the problem lies with old maps: dotted lines often look solid, and creases throw the computer for a loop. Your job here is to give feedback to the code created by NYPL: did it do its job?
Fix Building Footprints
This is my favorite, because it reminds me of editing GPS maps on RunKeeper after the satellite has trouble seeing my watch through buildings. The code tries to recognize a building's outline; if it's not right, fix it!
This bygone church appears to have some buttresses, which confused the program. Delete the excess points and align the ones that remain to the shape of the building. If you need an additional point, drag one of those plus signs. Here's the final result:
Although Google now has a way to recognize almost any text (perhaps rendering human-verification services like CAPTCHA obsolete), NYPL's software is not so advanced. As the amusing intro video states, you can help ensure our predecessors could receive pizza delivery.
The final tool helps the program determine the type of building at a given site. These old fire-insurance maps use five colors: pink (brick buildings), yellow (frame homes), green (hazardous materials), blue (shops and stores), and gray (outbuildings like sheds). Sometimes colors fade or resemble one another, so it's your task to provide a human eye to the proceedings.