Researchers from the University of Maryland and Microsoft Research have proposed a modification to Bitcoin that would repurpose all the computation power used in mining for data archival. 

In other words, they want to bolt a big, decentralized file storage system onto Bitcoin mining. 
You can read the team’s entire proposal here [PDF]. I’ll highlight a few of Permacoin’s points below. 
 

Mining waste 

The research team first refutes the Bitcoin Wiki’s assertion that Bitcoin mining isn’t actually wasteful. 
“We show that Bitcoin resources can be repurposed for other, more broadly useful tasks, thereby refuting the widespread belief reflected in the Bitcoin FAQ,” they write.  The key idea in our scheme is to make Bitcoin mining depend upon storage resources, rather than computation. Permacoin then utilizes storage resources in the Bitcoin network.” 

Local storage 

Permacoin relies on users storing data locally, as the scheme would require “local, random access to a copy of a file,” unlike MaidSafe, for example. 

“A standard POR [proof of retrievability] involves a single prover holding a single file F. In our setting, however, multiple clients collectively store a large dataset F (too large for a single client) in a distributed manner. Of these an adversarially selected fraction may act maliciously. The first challenge in creating our SOP [“scratch-off puzzle,” their term for Bitcoin’s underlying mining mechanism] is to construct an adversarial model for this new setting, and then present a distributed POR protocol that is secure in this model.” 
“Additionally, we must ensure that clients indeed maintain independent storage devices. If, for instance, clients pooled their storage in the cloud to reduce their resource investment, the benefits of dataset-recovery robustness through distribution would be lost. Thus a second challenge in our SOP construction is to ensure that clients must make use of local storage to solve it.” 

Security 

“We construct our SOP such that efficiently solving it requires continuous use of a miner’s private key. Given our assumption of local key storage, therefore, we are able to show that clients in Permacoin perform mining locally and thus that fragments of F are distributed across distinct clients and enjoy the full physical distribution and robustness of a true peer-to-peer network.” 

Opinion 

I don’t consider myself informed enough to offer any publishable opinions on Permacoin, but I do know that some of Cointelegraph’s readers are. If anyone has read the full paper, I would love to hear your thoughts on whether Permacoin could be useful or what weaknesses you might have spotted.