Reductionism is the attempt to “reduce” everything to physics. Should it not be possible to describe everything as a physical process? Chemistry is clearly a branch of physics. Biological organisms work in terms of chemical processes. Psychological processes work in terms of brain processes, which are biological, culture is a product of those psychological processes… So why do we have biologists, psychologists and scholars studying different aspects of culture? Why are there not just physicists in the universities?
In a way, the reductionist project should work; everything might really just be physical processes. But reducing everything to physical descriptions seems to be impractical. In the last three articles, I have tried to think this through for a relatively simple technological example, the WordPress blogging platform. Let’s analyze this a little bit further.
Scientists are trying to arrive at exact descriptions of systems by separating them into a set of laws on one hand and some “content” on the other. In physical models, the content often appears as the “initial conditions” of the system or as a “stream of inputs”.
But as we have seen in the previous article, things are becoming complicated if a system is programmable. For a system to be programmable, two conditions must be fulfilled:
- The system has a memory and can store information entering it as input. In order to have a memory, the system needs to be able to reconfigure its internal state in some way under an influence from the outside (i.e. an input).
- Information stored in that memory can influence what is happening with later inputs, i.e. part of that stored information can, in some way, be “executed”.
In such a system, a clear separation of laws and content is no longer possible. Condition 2 means that information entering the system (as content) can influence the way it is working i.e. it can modify (or become part of) its “laws”.
Take your mobile phone as another example. You can download a new app into it. That app is a piece of information stored in the device. If you start (“run” or “execute”) it, it influences the way further information entering the device is processed. The mobile phone might thus “morph” into a notebook, a calculator, a toy, and so on. And the memory is extensible since some of the apps you are using might store information and do some processing on remote servers, in the “cloud”.
In general, the new information entering the system cannot always be derived from the information contained in the system before, i.e. it is possible that the system encounters new or surprising information and that this new information is then incorporated into the system. If this is the case, any theory about how the system will develop in the future is incomplete because the system can be changed by incorporating new information not derivable in the existing theory.
So we can add a third condition to the two above:
- A system that is programmable (in the sense of conditions 1 and 2) can develop in unpredictable ways if its environment contains novel information (i.e. the environment contains surprises).
“Novel” information here means that there is information that is new to the system in the sense of not being derivable from the information that was there before. Such information can then be stored inside the system and influence the way it operates. It looks like the real world we are living in fulfills this requirement. It is rich enough to always provide surprises that can inspire new developments. As a result, a complete scientific theory is not possible for such a system.
Reductionism states that everything can be explained in terms of physics. In a sense, this is possible, but it is still not feasible to describe any system completely in terms of a physical theory. We can distinguish here between two types of “reductions”. Let me suggest names for them:
- “Implementational” reductions are always possible: at any given time, any existing system is completely implemented in terms of physical objects and physical processes. Describe the system at that particular time and describe the current input and you can describe what is going to happen next (although in reality, this might be impractical due to the system’s complexity and the difficulty to retrieve all the relevant information).
- However, “complete explicit” (or “formal”) reductions are not always possible. There are systems for which it is not possible to describe every possible development and state of such a system and thus give a complete, explicit formal description of such systems because they are programmable and thus what is implemented and the way it is implemented can be changed by incorporating new information into the system.
“Science” in the narrower sense of the word is concerned only with systems for which complete explicit reduction is possible. But there are many real systems for which this is not the case. These include many technological systems, especially systems of information technology (like blogging platforms, mobile phones and the internet), complex organisms and ecosystems, human beings, groups of human beings, like human cultures and institutions and their sub-systems.
These include, among other things, even the process called “science” itself.
(The picture, showing a view of “science” that I find rather problematic, is from https://commons.wikimedia.org/wiki/File:The_Scientific_Universe.png)