Biological systems are resilient against certain kinds of damage and defects. Animals and insects continue to function when losing limbs or even certain organs. Organisms constantly deal with novel and unexpected stimilu in their environments.
But for all this robustness, biological systems are also fragile. Minor damage to the heart or the lack of a certian gene are enough to cause death. A virus - a tiny biological program built from RNA or DNA - subverts the machinery of the cell to replicate itself and can do enough damage to kill its host organism.
There is no agreed upon definition of what robustness is. But I think that everyone has a vague idea of what it is and I am sure that we all agree on the examples given in the opening paragraph.
The reader would be forgiven for thinking that this is a dry, academic point that holds no interest for anyone outside of my narrow field of Systems Biology. I hope to convince you otherwise.
There is much interest in taking engineering lessons from Nature because biological systems are robust, self-healing and yet seem to be built using simple principles.
Telecommunications networks (of which the Internet is the largest example) have attracted researchers interested in robustness, because of:
Nature seems to achieve simplity, robustness and low cost all at the same time. Engineers are lucky if they manage two out of these three.
A telecommunications network is maximally robust against failure if every party is connected to every other party. We lack the economic and physical resources to achieve this: hundreds of millions (if not billions) of cables would be needed to connected every telephone, computer or Blackberry to every other device.
The situation for biological organisms is similar: in principle, an organism could have much more backup capacity to deal with failures (two hearts instead of one or four arms instead of two). But every additional organ needs energy, both in food and in maintenance. And every additional bit of biomass makes escaping from danger more difficult.
Because all of the earth's resources are limited, we should forget about thinking of robustness in terms of unlimited resources. It means that we have to avoid thinking of robustness as implying guaranteed failure recovery. Failure is always an option.
Nature has selected for making organisms resilient against common failures without providing protection against fatal faults that occur with low probability.
This makes sense in the wild - all biological species exist as populations of many individuals. The death of a single individual does not destroy the entire population. What organisms lack in robustness, Nature makes up for in numbers. Replication, from the point of view of a population, is a simple and effective form of robustness. It becomes even more effective when coupled with mutation, since not all organisms are susceptible to the same diseases.
But still, simple diseases can wipe out entire species. Nature can never fully hide its soft underbelly.
Let us learn how biological systems fail. In this way, if we apply ideas from Nature, we'll be prepared for failure.
Finally, we should avoid the temptation to view the everything as an organism, as we do the Internet. No living organism exists as the only member of its species (at least not for an appreciable amount of time).
If we succeed in applying Nature's lessons well, but at the wrong level (for example by treating the Internet as an organism rather than a population of organisms), we're sure to be disappointed by the results.
Maybe it can be argued that a population is just an organism itself. Perhaps; but there is no guarantee that it is goverened by the same systems that govern the biology of the individuals in the population.