As an "IT Professional" you should know better ! Exactly how long do you think its going to take someone (or rather, as you well know, a number of someones... known as developers !) to (a) come up with a viable algorithm (b) code it (c) test and debug it ........ seriously man ! Its a non-starter of a hairbrained idea to think someone would expend so much resource !
Most image algorithms you would need are part of your basic photoshop or image tools already - there are some you could probably apply from existing freeware tools without more than 10 minutes testing a few sample images.
There are some simple algorithms as a simple sysadmin that I can think of off the top of my head, that no doubt the sorts of people who work in image analysis and geo-mapping software could probably easily do:
1. Simple divide and conquer - take large areas of ocean, divide into uniform sections. In a parallel process apply a simple averaging of each section to reject any that are all ocean or all cloud. Feed the remaining images up for crowd sourcing in the manner tomnod are. I would not be surprised if they are not doing this already.
2. Same process, but use edge detection (as mentioned above) to find interesting sections and feed them up for more analysis.
There are probably more advanced methods that could be used to identify shapes or eliminate other noise, but the above could certainly cut down the workload and its hardly a difficult thing to do with an elastic platform like Amazon or the sorts of environments Google and Facebook have at their disposal.
I think its a bit naive to outright reject something that is quite technically feasible and hardly difficult for someone used to working with this sorts of technology. Especially at the sorts of places that deal with satellite imagery day in day out.