Diversity

Go back to the List of Patterns

Problem

Homogenous systems are brittle and prone to failure.

Occurs in systems that are under unpredictable stress. Scale ranges from large to microscopic, and long to very short timescales.

Drivers include:

• preference for standardization in an attempt to achieve efficiency
• desire for reduced uncertainty and control through simplification
• localized, short-term perspective that does not take into account the larger system or long-term consequences
• specialization?

Therefore

Resilient systems incorporate elements with diverse inputs/outputs/processing that respond differently to changes.

Examples include:

• long grass prairie based on four different plant types (extract information from Biomimicry and the Land Institute)
• John Todd's Living Machines seeded with diverse species and at least three interconnected ecosystems, allowing self-organization to occur (extract information from Ecological Design)
• Highly fault-tolerant computer systems using elements that perform similar functions but developed by different vendors
• sexual reproduction
• Interface: Entropy carpet design (new Flor design promotes 'mix and match' flooring)
• "Diversified, decentralized, and locally appropriate energy production." (Denise DeLuca)
• "Bacteria and fungii in the soil." (Denise DeLuca)

To implement a diverse solution <this section needs a lot more detail>

• determine the range of conditions that are likely to occur within the identified system
• if individual elements cannot handle all conditions, identify other elements that 'fill the gaps'
• link the elements such that they can back each other up

Diversity allows for co-operation by reducing direct competition for the same resources and encouraging 'waste as food'. Closed Loop systems depend on diversity.

But

It is not clear how to develop the range of conditions in an unpredictable system. A degree of 'over-engineering' may be required to handle unforeseen events. Conversely, adding too much diversity may increase complexity and instability. <Is there any research on natural systems that might suggest rules of thumb for 'optimal diversity'? See Discussion on cost of diversity in Information Technology.>

Identifying the boundaries of the system under study is key. Large systems become increasingly complex and hard to model. Stuart Kauffman (At Home in the Universe) suggests that systems need to be broken down into 'patches' that are interconnected, but with a limited number of links to avoid chaotic behaviour.

Describe how application of the pattern may change the context.

Monoculture agriculture is an example where growing a single crop (or even cultivar) appears to improve profitability through higher yields, ability to optimise machinery and other processing systems and greater predictability. Unfortunately, monocultures are prone to disease and insect infestations, which require increasing applications of pesticides. High crop yields deplete the soil, necessitating artificial fertilizers. Climate changes can cause sudden collapse, such as the 'Dust Bowl' of the 1930s. Economic forces can also cause havoc - loss of subsidies from the Soviet Union seriously affected Cuban agriculture Cuba: The Accidental Revolution, reducing individual caloric intake by 33-50% and resulting in a resurgence of interest in organic agriculture.

Standardised 'global architecture' has resulted in buildings that require large amounts of energy to maintain comfortable conditions. Hermetically sealing buildings has lead to 'sick building syndrome', and isolates the inhabitants from the natural cycle of diurnal and seasonal change.

See Also (Optional)

References to other patterns not mentioned elsewhere