Filecoin: Understanding the Complex Crypto System Meant to Rival AWS

Filecoin might be the most complex thing the blockchain industry has ever brought to market.

The Web 3.0 data-storage project, funded by a $257 million initial coin offering that closed in September 2017, has been building out its technology ever since. 

While Filecoin has been quieter than most over that time, Filecoin is currently wrapping up a very active incentivized testnet called Space Race. There are 1.5 million FIL tokens allocated to testnet participants, though it's hard to assess the value of that offer when the company has refused to discuss details that might inform FIL's underlying price. Nonetheless, this testnet has been quite popular. 

"Miners are learning a ton," Filecoin's Ian Darrow told CoinDesk. More than $100 million worth of hardware has come online to serve the testnet, he said. More than 200 pebibytes of data capacity has been proven. "Overall, it's exceeded expectations," Darrow said.

There are futures markets for the forthcoming token, too. Gate.io is currently trading futures at $18.50 in USDT as of this writing. In a recent investor letter, Pantera Capital said that Filecoin futures were "trading well above our cost basis." 

Filecoin's creator Juan Benet has a bright vision for the future of the web, and thus the Filecoin network aims to be a next-generation marketplace for data storage and retrieval, potentially competing with the web giants that control the data storage space – Amazon, Microsoft and Google – and content delivery networks like Cloudflare.

While it's true the cost of storage has only trended down and selling storage has effectively made it a commodity product, the Filecoin thesis is two-fold: 

The last Filecoin roadmap update that addressed mainnet launch said it would happen in mid- to late September. When we spoke to Darrow he said this was still on track. 

"Unlike centralized cloud storage services, which back up data in ways clients can't change or verify, Filecoin allows clients to easily express their own preferences for reliability and cost," a new document released ahead of the network launch, Engineering Filecoin's Economy, explains. It goes into several key elements of the new system.

Here are some of the main takeaways.

"One thing that I don't think was super-well understood before this came out is the mechanics and the rationale around how miner collateral works in Filecoin," Darrow said.

Mining on Filecoin is largely about providing storage space using traditional storage systems. This is commodity hardware. Almost anyone with an internet connection can participate. A quality one-terabyte hard drive can be purchased for around $50, though most entities participating will put up vastly more than that.

The point being that Filecoin does not enjoy the expensiveness of specialized hardware (such as Bitcoin or Ethereum ASICs) as a way to discourage unserious participants, so it needs to require a slashable stake in order to prevent malicious actors from accepting deals for terms they don't honor.

The Filecoin system requires upfront collateral. Block rewards are also subject to vesting, in order to help ensure participants stick around long enough to honor commitments.

Filecoin wants the storage capacity of the network to grow, but without inducing an ugly "Day 1 gold rush," said Darrow.

"There's this really nasty incentive for people to pile in early," Darrow said of other projects. "Mining rewards are the highest early, total mining power is the lowest, so you attract a lot of people who are bouncing from project to project."

Filecoin has a pre-set growth schedule for storage capacity and 70% of the mining rewards are tied to that baseline. If the network hasn't hit that target, then block rewards are reduced to the proportion it has achieved.

The idea, Darrow explained, is to lessen the first-mover advantage. "Because the network reward grows as mining power grows, it matters less when a miner enters the network," he said via email.

"The overall result is that Filecoin rewards to miners more closely match the utility they, and the network as a whole, provide to clients," the "Economy" document states.

Filecoin increases the baseline 200% annually, only giving the community an option to slow its growth down once the network reaches over 1% of global storage.

Based on the experience with the Space Race, Darrow said, "It seems like we won't have trouble hitting pretty gaudy numbers in the first year."

One of the features that might seem strange about Filecoin at the outset is that participants can earn rewards for storing nothing.

This is referred to as committed capacity. It's effectively empty space that has nevertheless received the Filecoin cryptographic treatment and is getting logged and rewarded on the system.

At first blush, this might not make any sense, but it's a matter of having some room in the system.

"What you don't want is a network where there's no extra capacity available or you don't know extra capacity is available," Darrow said. Further, because all nodes are also participating in securing the Filecoin blockchain, they are contributing to consensus.

That said, Filecoin doesn't want that many participants to commit empty blocks, so an incentive system has been created to push folks to find real clients.

Serving verified clients earns significantly higher block rewards. These are real companies with real needs making real deals for data. The idea is it won't be that hard to become a verified client, but its business should be significantly more attractive to those on the network.

"Verified clients are certified by a decentralized network of verifiers," the new document explains. 

This will be a set of large, recognizable entities in the ecosystem, Darrow said, though the list is not public. It will be groups such as: nonprofits that do a lot of data storage, academic entities and the major foundations of the blockchain world.

Storing verified client data will receive higher block rewards than unverified data or empty blocks (since it is encrypted, the Filecoin network cannot tell the difference between the two). In fact, Filecoin has already been seeding the world with verified data. 

Because verified data is so attractive, Darrow said the team expects miners will do business-development work to try to secure deals with verified clients, helping to improve overall competitiveness across the system.

Filecoin is usually described as entering the space of Amazon Web Services or Dropbox, but it's a bit more disruptive than that. 

Filecoin also seeks to account for content delivery network (CDN) services as well.

The most famous CDN is probably Cloudflare, and these networks fulfill a surprising role on the internet.

Data moves fast but the world is big, and it's still true that a file closer to a user gets there noticeably faster than one that's further away. Most stored data never gets touched and so will never need these services, but some stored data becomes very popular.

For such popular data, companies can provide retrieval services, meaning they would get paid for serving the data to customers. An example Darrow gave was when a certain video goes viral. A CDN could cache copies of that video all over the world and bid for the chance to serve it to viewers.

These additional copies would not count toward the copies the client had paid to keep securely on Filecoin, but they could be used in a retrieval market.

In many if not most cases, Darrow said, the storage providers will also provide the retrieval services, but there are likely to be cases in which it makes sense for CDN-like entities to step in for certain files.

Clients pay for storage in FIL, which is volatile. The Filecoin system requires storage clients to commit their payments upfront over the life of a deal, though miners will only get paid if they meet their commitment. The "Economy" document illuminates the advantages of this arrangement:

On the other hand, many storage clients may be accustomed to paying as they go rather than paying a lump sum at the beginning. Darrow said the trade-off here is they only pay for what they need.

As money on the internet becomes more complex, this will also likely present an opportunity for third parties to finance upfront payments for companies that want to use Filecoin but pay at flexible regular intervals.

One of the most important facets of Filecoin to understand is that it relies on a concept called content addressing.

Most addressing on the internet relies on locations. Go to a URL and see the thing you want there. As the internet ages, more and more of these links are dead. Called "link rot," there's a growing push among bloggers to link to archives rather than the original sources.

Filecoin points to content, not location, though, which means it might be found in any number of places.

Filecoin takes this a step further and also cryptographically proves the uniqueness of each copy. So if a client wanted to know there were 101 copies of file all around the world, they could use proofs to verify both that each copy was unique and that they were well distributed around the globe.

This is Filecoin's $257 million bet: Storage customers will make considerably more nuanced demands of storage providers as the internet becomes vastly more important than it already is.

Update (Sept. 15, 13:47 UTC): Added clarifying information about the intent of Filecoin's incentive structure in the second section.