Kinds of networks

The theory

The reader question I will try to address for today's post is: "How does one make a group outward-looking rather than inward-looking?"

This question takes us again to the theme of networks. Because what makes a group look inwards or outwards is fundamentally the kind of communications network it has, both for internal and external communications.

Let's have a look at different kinds of networks and how they work:

• The line or chain: This is the simplest kind of network there is, so simple that many people wouldn't even call it a network. When people talk about it, usually they are thinking about a part of a bigger network, as when they say "a food chain" or "a command chain". The most important thing you need to know about a chain is that anything that is passed along that chain (information, food, materials) has a risk of being lost or degraded in some way at each step, and the transmission from one end to the other will be slow. An animal at the top of a food chain has only access to a small percentage of the available calories, but at the same time, it has a higher risk of accumulating pollutants in its body. And we all know that information tends to get degraded or lost when it goes up or down a command chain.
• The ring or cycle: A cycle can compensate, to a certain extent, for the losses that happen along a chain. In ecology cycles are common, we have for example the water cycle, the carbon cycle, the oxygen cycle and the nitrogen cycle. What is lost at one end may be recovered at the other end. Good organizations have feedback mechanisms that will allow people to know if the supplies or information provided at one end arrived intact at the other end, and if not, some mechanism to recycle.
• The star: A star is a network with a central node, called hub, connected to several other nodes. It has the advantage that the connection between the hub and the other nodes is direct, with minimal losses. The hub has the best access to information, food, etc. On the other hand, links often need some kind of maintenance. The hub may spend a lot of time, energy, money, or other resources, just keeping those links working. A good example of a star network is a mother that provides for her young. The mother has often complete control over what she will give to her young, and they are completely dependant on her. But keeping her young will take out a lot of her energy and time.
• The bus: The idea of bus comes for computer networks, where a simple cable can do the function of a central hub. Sometimes, what is doing the function of a central hub is very different from the nodes, it's something big and passive. For example, a river connects in some way all the plants and animals living inside and at the banks of the river. Or a website could keep all the members of an organization in touch with each other. You could say that the river or the website are buses.
• The tree: A tree is a network that combines some of the best characteristics of stars and chains. On one hand, it keeps the distances between two nodes reasonably short, avoiding the problems of a long chain. On the other hand, each of the hubs has a reasonably small number of links, avoiding the problems of a big star. Trees are found in all sorts of situations, starting with... trees, with all their leaves connected in this way. Another well-known example is hierarchical trees in organizations.
• The fully connected network: A fully connected network is one where every node is connected to every other node. They are usually rather small networks, because maintaining a link is often costly or difficult. For example, in an embryo all the cells will be touching each other up until the point it reaches eight cells: from then on, they lose direct contact. An example from human communication networks is mailing lists: everybody receives all the messages. As soon as there are too many people sending too many emails, they lose popularity.
• The mesh: A mesh is a network where the links are easily done and undone, and are often changing. In some species, networks formed by family ties are very stable, while in others, they are a mesh. Human social networks are often a mesh, as well.

When it comes to communication networks, looking inwards or outwards depends on how many links a node has and how stable they are. Unstable links promote looking outwards, looking for new links to form. Also, the less links there are, the more time a node will have to look outwards. In the types of networks described above, the most inward-looking nodes are (in this order) those in a fully-connected network, the hub of a star, and the hubs of a tree - the closest to the root, the more inward-looking. Nodes in chains and cycles also tend to be quite inward-looking, partly because communication problems also increase the focus inward. The most outward-looking nodes (in this order) are the nodes of a mesh, the outer nodes of a star or a bus, and the outer nodes of a tree.

The practice

Experiment 1

This is a useful experiment to see how a network can look very different depending on which parts of it we are looking at. The network in this experiment will be a food web. To do this, you will need access to a patch of wildlife.

1. Pick a common plant you can see. If you don't know the name of the plant, ask or look in a plant guide to find out what it is.

2. Sit and look at it until you see an insect that will eat it. Identify the insect.

3. Find out a bird or other animal that is likely to eat that insect. You could do this by direct observation, if you are patient enough, or just find out from books or the Internet.

4. Continue up the food chain until you reach an animal that doesn't have natural predators. You have just identified a complete food chain.

5. This food chain, in fact, can also be seen as a cycle. Find out which microbes will decompose the final predator when it dies, converting it into fertilizer for the original plant.

6. Find out what other animals this final predator will eat. Now you can see this part of the food web is a star.

Experiment 2

Start with the experiment 1 in the post on networks. A few questions after you have drawn the red, black and green networks:

• Do you think that the red network is a mesh? Do you think that the green network is a mesh? Is this good or bad?
• Does any part of any network look like a long chain? Is there any way of making that chain shorter?
• Does any part of any network look like a cycle? Is this helping for feedback? Could it be used for feedback if it isn't working like that now?
• What parts of the network look like a star? Do you think the hubs may be overworked? Is there any way you could create something like a bus?
• Does any part of any network look like a tree? Are the people involved aware of this?
• Does any part of any network look like a fully connected network? Is this taking up too much time and resources? Should the groups merge?