MaidSafe uses the same type of distributed hash table as Bittorrent (Kademlia). Although it isn't the same protocol, it has roughly the same principals. The SAFE network is going to feature SafeCoin rewards for fast seeders, and for popularity of public files. If an artist uploads his or her content first, they'll earn SafeCoin (from the unissued pool) the more popular that content is. So the public gets free content and content producers get paid if they are popular.
Isn't MaidSafe supposed to be somewhere to store data in a safe and secure place? It's also built on the Masteroin protocol.
Torrents are far less risky than storing your own personal data online.
Anyway, good luck to them.
It's not built on the Mastercoin protocol. Only the MaidSafeCoins run on the MasterCoin protocol, but they are only temporary tokens to be exchanged for real SafeCoins 1:1 once the MaidSafe network launches. Once that happens MaidSafe has nothing to do with MasterCoin anymore.
MaidSafe has more in common with Bittorrent than with Bitcoin, since it uses a similar system to structure the network of data and nodes (the Kademlia DHT). The difference is that when seeding with Bittorent you choose which torrents and thus which data you are seeding. When seeding for MaidSafe, the network decides which data you seed. It wil gradually send you encrypted data chunks of 1MB in size for you to seed. Every chunk is also sent to and seeded by at least 3 other nodes. When a data chunk is downloaded by someone, you will essentially have a race with the other 3 seeders, and the one who finishes transferring it to the downloader first will get a chance to "mine" a SafeCoin.
So when you upload a personal file to MaidSafe for secure storage, your local client will cut up the file in chunks of 1MB and encrypts them. Your client will then send those encrypted chunks to it's manager group on the MaidSafe network, which will pass it along to the chunks data manager groups. Such a data manager group will pick 4 random seeders from the network which are tasked with storing/seeding the data chunk in question. The data manager group will monitor the availability of the seeders and if one goes offline will pick another seeder to also store that chunk. Your client will also put in your network account the hashes of the 1 MB data chunks and to which file they belonged. That way it can later retrieve those 1 MB chunks from all over the network, decrypt them, put them together, and that way restore the file.
Note that when I am talking about a "manager group", it is a group of nodes whose network ID have a mathematical relationship called 'XOR closeness' to the ID of your client or the ID of a particular data chunk (this 'XOR closeness' is the idea behind a Kademlia DHT). These ID's are calculated in a deterministic manner, so you cannot choose which ID your client or which ID's your data chunks will have. That assures that such a manager group is a group of completely random nodes. Such a group needs to achieve a 90% consensus before it can take any action on the network. Nodes in a group that do not follow the rules are de-ranked and eventually kicked out of the group and blacklisted. A node needs a set minimum rank to be able to gain a meaningful vote in such a group. You gain rank by providing the network resources (seeding and routing network data). You lose some rank when, as a seeder, you go offline for extended periods or when you lose data chunks you were supposed to store and seed. A higher rank results in higher SafeCoin rewards for seeding, so rank determines you rate of income as a seeder. You definitely want to maximize it.
As nodes leave and join the network constantly, these groups also change. Since you cannot pick your own network ID, and since you need to provide "Proof of Resource" to gain rank to vote in a consensus group, and since the groups are in a constant state of flux, it is deemed extremely costly to stage a sybil attack with the goal of dominating a particular manager group. There are more details to how the groups are formed, but this is the gist of it.
The end result is that, like in a blockchain, you'll have to outperform the rest of the network by a significant margin to be able to break rules, like double spending. The main advantage that this approach has, is that it's not only decentralized, but the work load is also fully distributed, which gives it virtually unlimited scalability.