I've been experimenting with full text search engine [1], on the
Geograph archive, and realised could expose a feed exactly like the
real site, which others can play with.

Example:
http://www.nearby.org.uk/geograph/find-syndicator.php?query=Trafford+...

If you using a text search against the archive, then this interface
might be preferable - note it doesn't do location searches like the
real site (but does do some location, see [1]), so don't use that type
of search against it - just plain text. (it uses the same feed creator
class, so the results should be identical so the same parser should
work against it!)

If the above search doesn't return any results, could further try
prefixing it with ~ as that will do a 'match any' search - but still
sorts on relevance - more likely to get it wrong, but should find more
images

http://www.nearby.org.uk/geograph/find-syndicator.php?query=~Trafford...

Hopefully once get Sphinx to compile on the live servers, will migrate
this to be a full feature. (the one on nearby will probably remain as
a development version)

Feedback welcome....

Barry