My Idea Sucks - A Small List of Problems with My Idea

I have been the advocator of my idea so far, but here are a few problems I am currently working on, to ensure that the system will work as a coherent whole.

1. Security - security is, believe it or not, very important. We can further divide security up to
1. Encryption of information
   
2. Secure network/information transfer
   
3. Secure deletion of information - our system keeps everything you ever give it. You cannot delete anything really, so we need ways to delete information, references to and from that information easily and securely.
2. Provable correctness. It is important that we can show that what we are doing is actually being done - because we are using a more complex system, proving things may be more complex
3. Coherence - the computer/os/program pair makes the entire operating system on one hand a loose set of tools, but also a monolithic seeming program. This may make coherence a large issue
4. Scalability - well, CS has always been plagued by prophets that make things that work at the small scale, and then totally cr*p up when you try to scale it up. THis relates to coherence, time space complexity.
   
5. Support for newer things - I have tried to support the ability to make new inventions still compatible with the model by enumerating the current ideas, and extending them as far as possible. For example, I have suggested the ability of viewers to be more than 2d - to 3d or 3+1 d, I have suggested the ability for meta-viewers to be able to arrange spatially, temporially, or use other spatio-temporal methods to arrange things. I have also abstracted the computation into functions, which any turing machine shouldbe able to do. However, there are often unforseen circumstances. For example, quantum computers, different chips might not support lazy function fusion or lazy evaluation etc, different output methods that do not rely on sight or sound (smell, touch, taste, direct brain manipulation of senses a la matrix?)
6. Speed - believe it or not, speed matters in the real world. I have tried to fix this by saying that the code will all be high level intent based code, that will be rewritten (like JIT compilation) to use optimisations, like code morphism (e.g. if a=0, then the code can be so much simpler.. etc)
7. Simplicity - while I try to adopt the onion-layer-knowledge approach, where only when you want to know about something do you have to know about it, it might be inevitable impossible