DataCoin - better idea

Since I am a newbie (despite mining BTC since 2010) i am posting in this subforum rather than the proper subforum for this discussion...

Datacoin seems like a great idea for secure distributed storage. However i see couple design flaws with it and I think if these flaws are fixed we could have a really, really, really interesting coin to work with.

Current flaws:
Data is stored permanently. The problem with this is that the use of this chain for "secure temporary storage" is not applicable. I think it would be better if the "data contributor" had the option of overwriting the data, this way the chain would be useful for a much longer period of time.
The miner reward makes sense, but the final model for reward should be based on local storage and internet throughput of each client. Otherwise everyone will just use lite wallets and no one will store the data.
Data should be redundant, but each client should be able to specify the size of the chain he is going to store. This way if i choose to store 5,000TB of data chances are i will get more requests to serve that data to network, while someone who stores 100KB of data will have very few requests hence lower earnings.

Serving a block to the network "creates" a small amount of coin (100X) for the user.
Requesting a block from the network costs a smaller amount of coin (1X) to the user.
Requesting to overwrite a block with new data costs more coin (10,000X) to the requester. Only the block owner (data contributor) can request to overwrite the block.

Part of the block stores transactions, just like bitcoin or litecoin; and another part of the block stores user supplied data. User supplied data portion can be overwritten by owner, but transactions are stored permanently.
Updating existing block is done at difficulty level much lower than current difficulty level; this way blocks can be updated with new information relatively fast, but addition of new blocks to the chain is done just like BTC or LTC.

I don't have all the answers and clearly defined protocols; rather I'd like to start a discussion on this topic to see if we, as a community, can come-up with something worthwhile...

What do yall think?