Time to solve (note that it's a semilog plot!):

 http://trac.astrometry.net/attachment/wiki/SolveSeries/solvetime.png?format=raw

 http://trac.astrometry.net/attachment/wiki/SolveSeries/solvetime2.png?format=raw

 Field

 Field + Quad

 Field + All Quads

 Field Rotated

 Field Rotated + Quad

 Field Rotated + Quad + Box

 Field Rotated + All Quads

 Field Rotated + All Quads + Box

 Index

 Index + Quad

 Index + All Quads

 Index + Field + Quad + Box

 Index + Field + All Quads + Box

Field

 http://trac.astrometry.net/attachment/wiki/SolveSeries/d16.png?format=raw

Field + Field Quad

 http://trac.astrometry.net/attachment/wiki/SolveSeries/d15.png?format=raw

Field + All Indexed Field Quads

 http://trac.astrometry.net/attachment/wiki/SolveSeries/d14.png?format=raw

Field Translated, Scaled, Rotated so that quads overlay

 http://trac.astrometry.net/attachment/wiki/SolveSeries/d9.png?format=raw

Index

 http://trac.astrometry.net/attachment/wiki/SolveSeries/d8.png?format=raw

Index + Matching Index Quad

 http://trac.astrometry.net/attachment/wiki/SolveSeries/d7.png?format=raw

Index + All Matching Index Quads

 http://trac.astrometry.net/attachment/wiki/SolveSeries/d17.png?format=raw

Index + Field overlay

 http://trac.astrometry.net/attachment/wiki/SolveSeries/d6.png?format=raw

Index + Field (All Indexed Quads) overlay

 http://trac.astrometry.net/attachment/wiki/SolveSeries/d18.png?format=raw

Index + USNOB

 http://trac.astrometry.net/attachment/wiki/SolveSeries/d10.png?format=raw  http://trac.astrometry.net/attachment/wiki/SolveSeries/d11.png?format=raw

USNOB

 http://trac.astrometry.net/attachment/wiki/SolveSeries/d12.png?format=raw  http://trac.astrometry.net/attachment/wiki/SolveSeries/d13.png?format=raw

Why Do We Fail?

Here's one:  Huge Diffraction Spike

There's a huge diffraction spike, so we select false objects in our cut. (The field is aligned vertically near the center of the field. Unfortunately it's a bit of a pain to get the field boundaries so I haven't made that picture yet.)

 http://trac.astrometry.net/attachment/wiki/SolveSeries/diffr2.png?format=raw

 http://trac.astrometry.net/attachment/wiki/SolveSeries/diffr.png?format=raw

Here's one where USNOB has a weird overdensity (double-scanned plate? plate overlap?) It's not an obvious show-stopper, but it's something weird.

 http://monte.ai.toronto.edu:8080/usnob/map.html?SDSS_FILE=1&SDSS_FIELD=3256&nf=300&view=usnob&ra=202.78&dec=8.41&zoom=9&HP=6&over=no