It comes from the fact that the corporate information is not liquid, difficult to move through entities, and that we think this won't change. Therefore we accept the reality of "the rocks". What can be changed however, is the ability to extract wisdom from those rocks by fine tuning every aspect of their privacy, and not trying to move them away.
  Rockchain is not a Blockchain. It is a distributed network of private data nodes with access rights managed by a public blockchain. Rockchain is blockchain agnostic and will work along with several public blockchains, using a distinct module for each blockchain. Rockchain through DappBox Relay can advertise private networks content on the public blockchain and pass queries of content access through the relay.
We did not reach the minimum cap and refunded the token buyers. We're still continuing the project on the founders foundings, lowering the scope of the vision to what can immediately be deployed in production for real use case. Also, the failed ICO pushed us to open to other blockchains than Ethereum, going towards the most practicle and fast choices for us.
We see too many open source forks with a few tweaks in the blockchain area, so after our failed ICO we do not intend to open source all the code yet. We know we need to "Proof of Work" our framework, that's why we'll release a free SAAS interface to test our RockEngine computation performance on private data. For the DappBox, as it is reusing open source parts, all those will be packaged as open source modules, while the DappBox still intend to be closed source. In the longer term, there will be an open source DappBox and a closed source DappBox Enterprise version.
We do not orchestrate a computational marketplace between data providers, CPU/GPU providers and users. This is somehow centralized in our view. The RockEngine creates "computational links" between DappBoxes, meaning that one DappBox can execute computations on one or several other DappBoxes while respecting their privacy. Of course, "computation on private data" means our algorithm is providing "computational integrity", meaning that the private data node cannot alter the algorithm to return fake results. This is an essential requirements for anything related to computation on private data. Rockckengine provides thus Peer to Peer computation respecting privacy.
Rockchain DappBox focus is a decentralized access rights management on the public blockchain between private nodes. We do not intend to provide a safe redundant and cheap online storage in our core offer. The core offer of the DappBox is to allow peers to exchange content using public blockchain logic.
We differ from secret sharing approaches (sMPC= secured Multipart Computation) since our algorithm is more like a compiler deploying code on two nodes at the same time. The algorithm then executes on the two nodes that communicate by exchanging encrypted messages. With this technique, we mathematically guarantee that no private data can be leaked and that the private node effectively executed the correct algorithm.
Most of the real world applications have to create logic applying on private user (or application member) data. The users of blockchain applications can't deal, in the long term, with all their data being revealed to the world, without the possibility to delete it ever. I wouldn't like to publish my blockchain public address if people could analyze it and know how much money I lost on online pocker games for example. Real world applications need smart contracts that have both logic on the public blockchain data structures, but also on the private user data. Winning blockchains will mix the two approaches in the same logical framework.

Of course now you can deal with private data with IPFS hashes, retrieving encrypted data and performing tests on it. But then you do not have a strong guarantee that the user is not changing the algorithm on his side, and all the trust must be delegate to smart contracts, making the balance difficult to find between code impartiality and privacy preservation in current architectures.

Ultimately, the winning platforms are often those abstracting the complexity away.
Everything is written in GoLang for the infrastructure stack, while everything front end is Javascript (with a preference for React). The modeling language for Rockchain is Javascript, and it is then compile using a GoLang compiler.