They are looking for beds for guests who don't meet their own shelters' criteria. If a man in need of detox shows up at a shelter for sober women and their children under five, the staff has to call around and find an available bed elsewhere. Naturally, the phone line of the shelter for single men in need of detox may be busy all night, because at the latter shelter they are trying to find beds for a battered woman with two infants who showed up at the door.
Obviously, none of these shelter workers have time to create and maintain web sites - in fact, they're so busy calling all over town that they don't have enough opportunities to interact with the guests at their shelters.
While it's good for a shelter to have a web site - especially considering how many homeless people make use of internet access in public libraries - it doesn't seem seem like a cost-effective use of human resources for every shelter to create one from scratch.
To me, it's seems self-evident that if all the shelters in a metropolitan area pooled their resources to create one web site with the right kind of content management system, then each shelter could not only make information about its services available to the web-surfing public, but could also participate in an extranet that would enable them to track bed availability in real time for the many shelters (each with its own criteria for guests) in the region. In cases where a shelter has to turn away a guest, the staff could use such a system to reserve space at an alternative shelter, and to print out directions to guide the homeless person or family to the venue where accommodations are available.
Naturally, there would be obstacles to overcome in creating and sharing such a system. Particularly in the early stages of the project, it would take a great deal of cooperation, and willingness to invest considerable effort, money, and time.
However, this is mostly speculation. I would like to invite anyone with experience in web-based collaboration among shelters in a region to post comments about ways in which their model works well or badly.
