Here's my "Universal senescence" essay. It can be found on the web - at: --> http://alife.co.uk/misc/universal_senescence/ <-- To offer a brief synopsis: I suggest that the tendency of many sorts of object to fall to bits represents a general rule of complex systems. This rule is distinct from the second law of thermodynamics, and - unlike the second law - can be applied to open systems. The aging of biological organisms represents a particular instance of this rule. I think the idea is a pretty obvious one - and I'd be suprised if it originated with me. Any references to prior work would be gratefully received. Universal senescence ==================== An increasing tendency to fall to bits -------------------------------------- A naive interpretation of the second law of thermodynamics describes it as an increasing tendency of things to fall to bits as time passes. This isn't what the second law /actually/ says - but the idea /does/ have a sort of intuitive appeal - since it /does/ seem that many common objects /do/ eventually fall to bits. Could it be that this tendency deserves to be described as a universal aspect of complex systems? Occasional disintegration ------------------------- Complex adaptive systems follow lawful behaviour just like other objects do - though in many cases the rules they follow may not yet be entirely clear. These laws are best regarded as /additional/ rules of thermodynamics designed to deal with complex adaptive systems. By far the most obvious rule is the statment of how such systems affect entropy - which I have previously attempted to describe [snip reference to my "bright light" essay]. Here I would like to propose that the phenomenon of senescence is a common feature of a range of complex adaptive systems - and can be characterised by the occasional disintegration of such systems. The suggested rule is different from the second law of thermodynamics. That also suggests that complex systems tend to fall to bits - but it can only be used to make that prediction in closed environments. The proposed rule does not have that restriction - the intention is that it can also be applied to self-organising systems in open environments. Subjects of senescence ---------------------- Probably the most familiar senescing objects are biological organisms. Senescence is common in biology, and - among more complex organisms - it is / nearly/ universal. However, other sorts of system also exhibit senescence. In particular, complex pieces of machinery also exhibit limited lifespans - and often have lifespan curves suggesting that progressive degradation followed by catastrophic collapse is occurring - the signature of senescence. To gives some concrete examples, I claim this is true of cars, houses, computers, stereos, clocks, motors, hyraulics and companies. Causes of senescence -------------------- Why do complex things tend to fall to bits? It appears that there are many reasons - and that different ones can apply to different sorts of system: - *Environmental damage* - complex systems are exposed to damage from their environment. Often no individual bit of damage it worth repairing - but the cumulative long-term effects of the damage is that the system degrades over time - and eventually stops functioning. - *High repair costs* - complex systems are often difficult to repair. They can need specialist knowledge to locate the problem and specialised tools to fix it. This can mean that some problems are more cheaply fixed with a new model. - *High cost of durable components* - one simple strategy aimed at preventing things from falling to bits is to build them out of strong, durable materials. Unfortunately, these are often expensive and difficult to machine. - *Planned senescense* - some things fall to bits simply because they were / designed/ to fall to bits. A broken piece of machinery often creates demand for a replacement. Sometimes, there is little incentive for manufacturers to build long-lived components - since doing so destroys their future market. - *Disposable soma* - reproductive and maintenance processes can compete for resources. Reproducing early has many advantages - and is consequently somatic tissue maintenance programs do not receive sufficient investment to support indefinite survival. - *Antagonistic pleiotropy* - this is the idea that genes that /delay/ the expression of other deleterious genes are favoured. More generally, it suggests that genes may be favoured if they have beneficial early effects but deleterious later effects. - *Cumulative parasite load* - systems accumulate parasites faster than they succeed in ridding themselves of them. The result is progressive loss of function - followed by catastrophic collapse. - *Obsolescence* - in a rapidly changing environment systems may be discarded because they are out of date. This effect is commonly seen in computer systems - but also arises among organisms who are in combat with parasites which evolve to adapt to common host genotypes - where older genotypes are more likely to encounter parasites that have evolved the capability of exploiting their resources. The role of complexity ---------------------- Complex systems are often more expensive to replace than corresponding simpler ones. This makes replacing them more expensive. However, they also have more different bits to go wrong. It is harder to identify the cause of problems - and it typically require an engineer with more spare parts and specialised knowledge to fix. The case of companies --------------------- The theory of "universal senescence" suggests that companies exhibit senescence. However - as far as I am aware - this theory has never been tested. It should not be too difficult to test - and the results may be of some interest. In theory, a number of mechanisms of senescence ought to apply to companies. In particular, companies: - ...are vulnerable to environmental damage; - ...can grow large, become unable to manoeuvre properly - and then crash; - can accumulate garbage in their components in the form of well-disguised dud employees or poorly-designed systems - in a way that makes the problem difficult to repair; - ...can reach a size where they become attractive to predators - and vulnerable to being swallowed up and destructively digested - in a merger with a larger company; - ...that have been around longer have increased chances of attracting parasites - either agents inside the company who do not have its best interests at heart or externally- run protection rackets - and the like. Companies are an interesting case - since most companies are still very young - and have not directly descended from other senescing companies. Thus - unlike most complex organisms with substantial genomes - they are not "built to senesce". However the theory suggests that they will - nontheless - exhibit senescence - and that the effect will /probably/ be big enough and visible enough to be evident from their lifespan curves. Unfortunately, companies /also/ exhibit the equivalent of "large infant mortality" - and the effect is a substantial one. This may act to obscure the effect of senescence over the lifespan of the majority of companies. This may make company senescence more difficult to detect. Modularity ---------- Can senescence be defended against using modularity? It seems that - by using a modular design and "unit tests" that failures in individual modules can be identified, isolated and repaired - without large-scale failures necessarily being involved. Modularity is - without doubt - a major weapon aganist senescence. However there are a few drawbacks: - Modularity is a design constraint that conflicts with efficiency. This is probably why (for example) we don't have twelve hearts - one heart simply works better; - Unit tests - and the time spent executing them - also represent something of a burden; Also, modularity doesn't /completely/ fix the problem. There is still the issue of connections between the modules. These themselves senesce - and may not be so easy to diagnose problems in. There are also the possibilty of unplanned-for failure scenarios within modules - failures that cause unit tests to mis-report, failures that knock out whole modules - and so on. Exceptions ---------- There are a few complex organisms that don't senesce. How do those fit into the idea of "universal senescence"? Not very well, it must be said. However, they are not very common - and the theory doesn't suggest that /all/ long- lived organisms must senesce - it merely provides reasons why most of them might. One way to ensure you have good repair mechanisms is to make sure finding a mate and reproducing is very, very difficult - so that it can take a very, very long time. Under such circumstances, selection will favour very long lifespans. I am happy to agree that selection is powerful enough to produce organisms with extremely long lifespans. However I am inclined to doubt whether that sort of selection will ever be acting very frequently. -- __________ |im |yler http://timtyler.org/ [email protected] Remove lock to reply.