Sia has changed substantially since the creation of this thread. Much of the information, especially in the early pages, is incorrect. A new thread has been created here: https://bitcointalk.org/index.php?topic=1060294.0

Sia is a new cryptocurrency that is being actively developed. We have released a whitepaper, and are in open beta.

About Sia's Technology:

Sia is a proof of work blockchain just like Bitcoin. A single feature is added that which enables clients and hosts to make storage contracts that will automatically be enforced by the network. More can be learned by visiting our website and Github page, and by reading our whitepaper (all available from our website).

About Sia's Economic Model:

Sia compensates hosts using a cryptocurrency. This cryptocurrency, Siacoin, will be easily exchangeable for bitcoins and subsequently USD. When the currency launches, 10,000 siacoins will be mined per day and distributed to hosts according to the volume of storage they offer to the network. The number of coins mined per day will decrease until the 4 year mark, at which point the network will mine coins such that the annual inflation rate is kept permanently at 5%. No coins will be premined.

Storage can only be rented using siacoins. This presents a major inconvenience for most people, so we will be providing a service that allows users to rent storage with USD, transparently exchanging dollars for siacoins behind the scenes. The siacoins spent on storage go directly to the hosts that hold the file, with the exception of a fee. The fee is distributed among the owners of Sianote. Currently, just under 0.5% of all host income has been distributed/sold/circulated as Sianotes on the Nxt Asset Exchange. The devs are no longer selling sianotes directly, they must be bought on an exchange. More can be found at www.siacoin.com/sianotes.html

Use Cases:
Since files are spread across hundreds to thousands of machines on Sia, the throughput for downloading a file is expected to be on the order of gigabits per second. Optimized files can enjoy latencies below 100 milliseconds from nearly anywhere in the world (or just on a single continent if the goal is to save money), and low redundancy files can be maintained at slow but highly reliable speeds for much cheaper than existing cloud storage solutions.

Any static data that already gets transferred over the internet would be more efficient through Sia. Every image on the internet could be embedded through Sia. This reduces the workload and bandwidth costs of webservers and makes websites as a whole cheaper to run. The same also applies to videos, which would substantially lower the barrier to entry for video websites. We plan on making Sia javascript and HTML5 friendly, meaning that pretty much any website will be able to integrate with Sia and pull content from Sia.

Sia is also friendly to applications. Imagine a music library that does all streaming directly from Sia. Right now, services such as Spotify and Pandora stream compressed music, because bandwidth costs are high when moving media around. If they streamed the music from Sia however, the costs would be reduced such that streaming FLAC from Sia would be cheaper than streaming mp3 or ogg from their own servers. Imagine video games that manage save files from Sia. Imagine streaming movies from Sia. Imagine downloading applications from Sia instead of sourceforge, a package-manger mirror, or Steam.

I'll have to count on you guys to build a better coin service that the public can understand more easily then.

Oh this could be huge. I will this thread.

There's likely to be some giveaway after we launch. One of our fears is that there will be a bunch of people trying to mine and nobody storing data. If there is nobody storing data, people can't mine (it's part of the security of the network, you can only mine once you have something to store). So we might do some giveaways so people can start storing things.

The other bump (looking for a work-around) is that you need to make a down-payment in siacoin before you can start mining. Before Sia is on any exchanges, the only way to start mining will be if someone donates you coins. We'll be passing coins around to help people out. I'm pretty sure though that we'll think of a better way to kick-start the network.

We might also just nix the required down-payment for the first 3 months. It opens up a bunch of attack vectors but if nobody is abusive initially then it won't be a problem.

This one looks intresting. I would also like to know the total cap coins.

actually it sound interesting (sound like a bit like torrent...)
But what about the network traffic generated by this ?
What kind of file gets into the storage ?
How do you protect the rest of the computer from being infected by crappy files downloaded there ?

random post ?

That sounds really great. But  it isn't appropriate   to advertise here .

In this system, I wonder how the blockchain consensus will be reached.

"network will mine coins such that the annual inflation rate is kept permanently at 5%"

No maximum. Every year the coin volume will go up by 5%. We do it this way because we want the coins to be about the storage, not about the speculation. There will be explosive growth in the first 4 years however, so there's still a ton of opportunity to jump on the coin early.



It's a lot like bittorrent, but instead of each file being entirely stored on each seeder, each file is only partially redundant (you get to pick the redundancy). It's hard to guess at exactly how much network traffic there will be, but probably a lot for people who are hosts.  There will be support for capping your bandwidth, though this will affect your income. (The model for charging for bandwidth isn't worked out yet).

Any file can get into storage. The clients will encrypt them first, but the network doesn't care. If you want to upload an unencrypted file (so it's available to the public), you will be able to do so. I'm not sure what you mean by "crappy" file, but it's just storage. If someone uploads a virus, it can't hurt your computer because the client doesn't open or read files, only stores them. If you try opening it with some program however, it might infect you.


1. Maidsafe has been working on this since before Bitcoin. A lot of solutions that are now available and have been thought about in greater depth were not available when maidsafe started. I'm pretty sure that if they started now, they would get a lot farther in the same 7 years.
2. The maidsafe project is considerably more ambitious than Sia. They are doing proof-of-bandwidth, proof-of-storage, proof-of-computation, and something related to uptime monitoring. It also seems like they intend files to be mobile (?). Regardless, from the best that I can tell, Maidsafe is much more ambitious than we are, and takes a more complex approach to many of the problems. This will increase the development time.
3. This project is still very ambitious, and is a few months away from a prototype. We are definitely trying to do some very difficult things. We are not asking for money at this point, nor are we giving away coins at this point. We plan to do both in the future, but will wait until we have something more substantial built.

I know that a lot of you are cautious about scams. If you look, you'll see that nowhere in this post or on our github page does a way to pay us appear (no bitcoin addresses, etc.). This project is very real and is very much not a scam; we aren't asking for your money.

Do we have a proof of concept yet?:

Well, we have a working demo that does not a whole lot. This currency is based on quorums, and we just finished building the base for quorums. A bunch of hosts can get online, engage in communication (following Leslie Lamport's solution to the byzantine generals problem) and kick dishonest hosts off the network. At the moment, there is no support for joining with a wallet or storing files. That's what we're working on now. Stay tuned.


Each set of 8GB that a miner offers to the network becomes a 'participant'. Each participant downloads and watches a portion of the network, and is paired with 128 other participants (chosen randomly). The 128 operate together reaching consensus using the Byzantine Generals solution provided by Leslie Lamport in his paper "The Byzantine Generals Problem." The pdf can be be found in the doc/ folder. We used the signed solution, meaning every honest host in a quorum is guaranteed to reach the same block as every other honest host within the quorum.

Quorums are arranged into a tree. Each quorum has a set of sibling quorums that all share the same parent.

When a block is found, it's announced to the sibling quorums, and to the parent quorum. The exact algorithm is not pinned down, but each participant will send the block to 2 or 3 other random participants. This keeps communication to a minimum. A group of dishonest hosts could announce a dishonest block, so there is protect against this; upon receiving a new block, each host will fire off a confirmation to the other participants in the announcing quorum. With very high probability, at least one honest host will see the dishonest block, and will be able to alert the receiving quorum about the dishonesty. After we pick the exact numbers, we will write a formal algorithms with proofs demonstrating what attacks are possible and with what probabilities the attacks will succeed. (Our target is to make the attacks as unlikely to succeed as double-spends on the Bitcoin network when an equivalent volume of hosts are dishonest).

Does that make sense? It's very different from any existing model, using a bunch of communicating quorums instead of using a single large blockchain. The primary reason for doing this is because Sia will have a lot more information in the blockchain than Bitcoin. For example, the header data of every single file on the network is stored in the blockchain. If every miner needed all of that data, you'd be looking at hundreds of gigabytes even before figuring out which miners had which files. A tree was absolutely needed.

We believe the algorithm to secure, but that needs to be confirmed by others. At this point, the algorithm is not formally written down, which makes it hard. Expect to have something more concrete regarding quorum consensus in the next 2 weeks.