I’ve spent a bit of time digging into Bitcoin lately – heck, I even read Nakamoto’s self-published paper. To me, there’s something fundamentally fascinating about a system that has been thought out so well, regardless of its purpose.

In addition to being intrigued by the way Bitcoin works, I also find it interesting due to its boldness of attempting to create a currency system that is fairly hard to control by conventional means, such as a single government’s monetary policy. To a certain degree, Bitcoin implements an anarcho-capitalist’s dream of an independent monetary system, where capital cannot be created at will.

But there are also drawbacks inherent to the way Bitcoin works. From a technical perspective, one valid criticism aims at the fact that it consumes enormous amounts of computational resources (and thereby, electric energy): In order to prevent attackers from introducing false information into the global history of transactions (which would allow them to regain the money they have already spent), transactions are only considered valid after they have been confirmed using the combined computational effort of many participating systems. Confirming a set of transactions requires solving a “proof of work” problem that is scaled in difficulty such that, on average, the time it takes to find a solution remains constant regardless of the performance of all systems combined. In order to encourage people to dedicate their systems to solving these challenges, finding a solution is awarded with a certain amount of newly generated Bitcoins. In an analogy to mining gold, dedicating processing power to creating Bicoins is called “Bitcoin mining”.

Unfortunately, mining Bitcoins means dedicating processing power to solving a problem that – other than ensuring no one tampers with the transactions – serves no real-world purpose. It’s all about finding a signature that is considered legitimate only because it’s incredibly hard to create. While it would be great if Bitcoin’s "proof of work" would map to a task that actually serves a real-world purpose, it appears to be difficult to actually find such a problem without compromising Bitcoin’s needs. Such a computational task would not only have to incorporate the transactions it is meant to confirm, it would also have to be arbitrary scalable, highly parallelizable without relying on intermediate results and still provide useful solutions at predictable intervals.

Even when knowing why Bitcoin works the way it does, it is still hard to accept the incredible waste of processing power. At the time of writing, the entire network of Bitcoin miners computes roughly 12∙1012 SHA-256 block hashes per second. Even when computed on the most efficient GPUs that are currently used for mining, achieving this hash rate requires a constant power output in the Megawatt range just for the graphics cards. The thought alone that several petaFLOPS worth of graphics hardware are constantly working on a problem whose must fundamental purpose is to be hard makes my heart bleed.