Bitcoin, the distributed digital currency has just hit one of its pre-programmed epochs. Now that there are 10.5 million bitcoins in circulation, the production of new bitcoins is halved and this has implications both for inflation and for processing.

Bitcoin is based on a strange algorithm that is supposed to enforce security by using a proof of work principle. This is perhaps the most misunderstood part of the whole bitcoin algorithm, so let's take a closer look at it.

Each bitcoin transaction contains a problem that can be solved by brute force computation and the fact that the problem has been solved can be immediately verified by all. The actual problem being solved has nothing at all to do with the validity of the transaction or the security of any particular bitcoin - it simply serves to pick who validates the transaction.

You can think of proof of work as being like a distributed secure random number generator. For example, suppose you have a large group of potential transaction verifiers, then to avoid subversion you need to pick one at random each time. This makes it difficult to organize an attack on the verification system. However, you cannot simply nominate a central authority to pick the verifier at random because is creates a single point of failure and the system is not distributed but relies on the central control. 

Think for a moment about how you could pick a single person to verify the transaction in a way that was unpredictable but without having a central "choosing" server. This isn't easy.

One of the simplest ways of choosing a person using a non-central algorithm is to give the candidates a problem, contained within the transaction, that will take each of them a variable and unpredictable amount of time to solve. The problem has to be hard enough to take a long enough time to ensure that you don't get two or more people solving it at the same time. For example, if the computation only took a minute on average you would certainly get multiple solutions bouncing around the network and, depending on which you received first, you would claim different people were first. However, if the solution took an hour on average then the chances are that only one solution at a time would be transmitted across the network, so producing a clear first that everyone could agree on.

The one who solves it first is essentially the one picked at "random" to verify the transaction. All of the other potential verifiers can see that the problem has indeed been solved and this is proof that the work really has been done and the selection mechanism hasn't been fooled.

The big misunderstanding is that proof of work is somehow something to do with the intrinsic value of the bitcoin. The reason for this is that bitcoin verifiers are paid by being given some newly minted bitcoins for their efforts - hence the verifiers are often called bitcoin miners.

This mechanism was setup to make sure that there were people willing to verify bitcoin transactions. It is a way of paying for banking services that bootstrap the currency into existence. Many people think that the work done in verifying a bitcoin transaction is so that some electricity has to be burned up to give the currency a real value. This might be a psychological effect, but it isn't an essential part of the currency idea. A currency is worth what you and others believe it to be worth - the gold standard or commodity-based currencies are just a memory.

As part of the bitcoin algorithm, it was decided that when the number of bitcoins reached various levels of circulation the reward for mining would be cut. This serves two purposes. The first is that it limits the growth in bitcoins and hence it reduces inflation. The total number of bitcoins that can exist is fixed to be 21 million. You don't need to worry that 21 million isn't sufficient for world use of bitcoin because each one can be divided down into around 100 million smaller units called satoshis.

The second is that as the currency matures the verification process should start to contain a transaction fee. However, it is far from clear how this transition should come about. Close to the point where the full 21 million coins are in circulation there will be no bitcoin mining because there will be no reward - at this point the verifiers will need to be paid some other way for their efforts.

We have just reached one of the points, the 10.5 million coin mark, where the reward reduces - now bitcoin miners only receive 25 bitcoins per verification rather than 50. If you take into account the fact that the difficulty of the proof of work problem is adjusted so that time-to-verify stays the same no matter how much the hardware improves, then you can see that miners are getting less for more work. However, if bitcoin miners give up mining then the time to solve a problem will increase and the problem will automatically get easier and hence mining will again be profitable after a period of instability.

Will all this have any effect on the value of the bitcoin?

Could it be the start of an even more difficult phase of the currency's development?

Who knows, but it is an amazing experiment in both distrbuted security and trust and, of course, economics.