Interesting discussion.

I don't pretend to be a SAR guy, but general crowdsourcing strikes me as highly dubious. It would take a high degree of on-the-ground familiarity with the type of terrain, plus typical survival behaviours, to interpret aerial photos or video footage in an effective way. The observer needs to have a practised intuition for filtering out the normal and homing in on what looks out of place.

For example, I would be useless looking at swamp, jungle, desert, and coastal rainforest; but better with aspen parkland, boreal forest, or the Northern Rockies.

There are also a lot of terrains where seasonal snowfall, extremely spiky terrain, or the sheer density of vegetation changes everything. Shelters of natural materials are perfectly camouflaged in some situations. An on-the-ground searcher could almost trip over them before they can be seen.

It may be that, in time, machine algorithms will be more effective in spotting anomalies than human observers. This is comforting, I suppose, as the machines love us for giving them something to do (other than plot world domination).

My 2c.