Demos vs. applications

Here's one of the slicker demos I've seen in a while, courtesy of Ben Fry. What does it do? Try it, and don't forget the "zoom" option. I'll wait.

Slick, huh? Note how the overall map shows the US population distribution without explicitly showing features like cities or streets^*, and invites browsing and personal readings. Edward Tufte no doubt loves it. Note how, as you type, you see not only a bit of how the US Postal Service thinks, but also see another indication of population density. The '6' and '8' regions, in particular, are huge, larger than most of the world's countries, while the '0' and '1' regions, for example, are much smaller.

But ... how often do I want to get the location for a ZIP code, anyway? The other way around, sure. If I do, do I really need to see the area narrowing as I type? If I do need to see it, wouldn't I probably want a bit more context, for example state/county/city names instead of just dots and numbers? People apparently sell three-digit ZIP code maps for sales planning, the first three digits being important because they correspond to the postal service's sorting centers, but such a map would be worthless without the other information you generally get from a map.

Again, it's a great demonstration of how to display a moderately large^** data set. It's fascinating to browse, particularly because it's interactive rather than static. But am I likely to use it every day? Probably not. Is it likely to be embedded in some application that actually needs to decode ZIP codes? Again, probably not. You don't need to spend all that screen space on a fascinating map when a simple form will do.

One obvious question: What sort of task would a display like this be useful for? Dunno.

(*) Fry does a similar exercise with streets, though I'm not sure either I or Google Maps believe the suspiciously sharp-edged empty patches in the Midwest [Since fixed -- D.H. May 2015]. I'm also reminded of my previous comment on nematodes, of all things.

(**) I say "moderately large" because by definition there can't be more than 100,000 five-digit ZIP codes and in practice there are more like 43,000. Compare this to, say, 26 million segments of road in the US or 3 billion base pairs in the human genome. It's large enough to be visually interesting when plotted all together, but small enough that the applet can carry the whole data set with it rather than querying interactively, AJAX-style.  [From a Google point of view, this would be somewhere between "small" and "tiny", of course --D.H. May 2015]