I think that this is really the responsibility of WorldWide Telescope and socl.
In socl's case, they are perfectly capable of using Photosynth's HTML embed code. There is nothing stopping them.
In WorldWide Telescope's case, again, no one is stopping them from breathing in photosynths and panoramas from Photosynth.net. The data is all online and available and the data is very simple: point cloud data and Deep Zoom Collections. WWT already uses Deep Zoom for its own imagery, so there's no barrier there and they say that they support point data as well. Moreover, many photosynths and panoramas are already geo-tagged and geo-aligned.
There are two other upcoming Microsoft projects that you should be keeping track of:
Digital Narratives: http://www.digitalnarratives.net/
Read/Write World: http://readwriteworld/ (More information at http://bit.ly/readwriteworld )