I'm lucky I get to write about Ashleigh's journal. For design, layout and user appeal it puts my journal and the journals of the rest of my Physical Computing class mates to shame. I actually have a hard time writing about it because I tend to explore it, looking at her pictures and reading things that are off-topic to this review. This review is supposed to be about the project ideas, so I'll move on to those.
The first idea listed is the Interactive Painting/Collage. I think it might be challenging to make the video and collage both easily viewable when you overlay them on the angled Plexiglas. But if it can be made clear, I can see how it could add significant interest to the collage. The second idea, Video-Revealed Story Game , is a good idea for an integrated, interactive interface for video presentations. But, the videos and the context of their content will be critical in the overall success of this piece. I have seen slightly similar things, so if your going to follow through with this idea I suggest you hurry. It may become too common in the future, which would be an indication it's a good idea. Object Memory Box might be a little too obvious if you intend it to be "art". It sounds like the user will find something they would expect to find. I like Office Voyeur 's surrealistic play with the ideas of public/private space and identities. This is the most interesting of your ideas. I really want to see it. A Night in the Life of an Insomniac suffers from a similar problem as Object Memory Box. The results of the users actions are very predictable and I don't think that's a good thing if you're trying to present this as an installation or sculpture. Insomniac 2 , on the other hand, should be more rewarding to the user and I don't think it is the same concept as your first Insomniac. The first Insomniac seems to want to retell an historical narrative, while Insomniac 2 is dealing with the users identity through volume and shape in real-time. Of course, the videos you choose will make a significant impact on both of these.
This journal can't be found on its server so I'll review the next person's journal.
All of these ideas are very general and don't have many specifics about how they might work. But that's true of my ideas in my journal. They all sound worth doing and I agree that ideas 2 and 3 would go together. Idea 2 seems like a precursor to idea 3. Unfortunately I don't know much about machine learning or artificial intelligence so I can't make any useful suggestions about how to approach that aspect of idea 3.