EveryBlock’s greatest achievement will be in the two layers that Adrian describes.

1. The first layer is the army of scripts do the job of “[compiling data from] public APIs, private APIs, screen-scraping the ‘deep Web,’ crawling news sites, plus harvesting data from PDFs and other non-Web-friendly documents.” I don’t know of any other website that’s doing the grunt work of compiling from multiple city and municipal data sources like this.
2. The second is the data storage layer that handle arbitrary data types with selective attributes and geocoding.

In short EveryBlock is going to itself become the quintissential API for hooking into all types of geocoded data and information. While it’s a bit early for Adrian to worry about funding (EveryBlock is currently funded by the Knight News Challenge prize), I wouldn’t be surprised if it’s eventually snapped up by Google or Yahoo. This is seriously useful local data that’s going to be huge as localized search/info becomes more and more crucial to web services.