If the central coordinator goes offline, everyone’s file is inaccessible during the central coordinator’s downtime. Note that no one’s file gets lost because of this: files are still safe within the Swarm, they are just temporary inaccessible. This is certainly a single point of failure as regards accessibility, but so it is for any other cloud platform on Earth. In other words, if you store your stuff on Google Drive and Google goes offline, your files go offline as well. You may say that this is not an issue with blockchain-based cloud platforms, but actually there are many more accessibility problems, not to mention the enormous issues in terms of scalability and usability.
Furthermore, notice that even if the central coordinator went offline, no one’s privacy would be violated, because we do not store anyone’s encryption key on our server. Long story short, if you want to access my files, attacking the coordinator doesn’t work.
Speaking of centralization, at this stage Cubbit is in a blurred middle ground between centralization and decentralization, i.e. the network is distributed, but has a central coordinator. We’ve decided to take this path for two main reasons: first, the usability is much better this way (no need to know anything about cryptocurrencies/blockchain/securing your private keys/syncing the blockchain/full nodes vs light nodes / hard forks & soft forks etc.); secondly, this way we can deliver right away.
Our goal, however, is total decentralization, i.e. to distribute the AI coordinator across the network.
The necessity to make requests through a coordinator actually makes it faster. In fact, without the coordinator the location of everyone’s chunks would be distributed, and the system would be much more complicated (and slow).
According to the data we’ve collected during the beta test, the possibility of your files going offline due to too many hosting Cells going offline is one in a million, and has never happened, not even once. Plus, the downtime of the Cell is not an issue since the Cell only hosts one chunk per file, and if more than 6 hosting Cells go offline at the same time, the central coordinator tells the network to create missing shards out of the already available chunks, virtually reducing to zero the probability of losing a file.
I hope I have covered each and any of your questions, if not please let me know.
