Information is Physical (?)

If you know anything about computers, then you probably have some idea of the way it stores and processes information. It is all about the 1’s and 0’s right? You write the 1’s and 0’s somewhere, the computer somehow reads it and BAM! It does what you want. Well, that is perhaps oversimplifying things just a little, but life is too short to care for the details. In any case, you have to store those 1’s and 0’s somewhere, perhaps on a microSD card or a hard disk drive or a compact disc or any one of the whole multitude of devices meant to store your digital information. In this sense, you probably already have an intuitive idea that information is physical because you always have to write them down on physical objects.

In 1961, Rolf Landauer took this (admittedly obvious) observation to another level by demonstrating a physical principle using it. It essentially states the following: the erasure of a single bit of information has to cost a minimum amount of energy called the Landauer limit (it has a numerical value of KTln2, see  Details aside, the essential argument revolves around a theoretical engine, named the Szilard Engine (Figure 1) after the scientist who studied it.  The engine appears to be able to do physical work without actually requiring any input of energy. This violates the 2nd law of thermodynamics, which says that this engine is giving us too good a deal and that a free lunch is impossible. The Szilard engine, however, does require you to make a measurement that gives you a single bit of information about the system in order to do work. Since the information has to be written down somewhere, and the actual writing on information appears to be doable without costing any energy, it should cost energy to erase this bit of information. In this way, there is no free lunch since you will eventually run out of space to write down the measurement and erasure of information has to be performed.

Figure 1: A rough Schematic of the Szilard Engine. You initially have a box containing a single particle in a heat bath. At first, you don’t know where the particle is inside the box. You then insert a movable partition inside the box and perform a measurement to find out if the particle is on the left or right of the partition. After this measurement, you know which direction the partition will move when the gas expands, therefore performing useful work.

These arguments are fairly convincing, and Landauer’s principle is accepted by many in the scientific community today. However, the issue of its validity is far from completely settled. One may for instance argue that the assumptions underlying the Szilard engine cannot be valid. Using the same set of assumptions, you can actually design a separate engine that does work without even performing a measurement (See Figure 2), also a clear violation of the 2nd Law of Thermodynamics. Is there a flaw in such a design

Figure 2: An engine design similar to the Szilard engine that performs work without measurement. For details see Mind, Matter and Methods by Paul K Feyerabend

So this casts some doubt as to whether Landauer’s principle is actually valid or not. Needless to say, this is an issue that is still being debated today, decades after the principle was proposed. Whatever your stand is on the issue, information remains physical, so keep that hard disk drive handy.