I love what I do, now let’s do it in 1/4 of the time

I started my PhD…

Eight months ago. And I love it. It’s everything I hoped it would be and more. It makes me happy. Really.

Although, the fact that I haven’t written anything for eight months might give you some idea about how busy I’ve been. Yeah, I’ve had lots of work. But I’ve also put a lot of effort in. I’ve become a bit of a workaholic.

It’s strange, how much of a complete turn around it’s been: from disinterested free-rider to dedicated recluse. I can see how kids who find their niche from an early age could miss out learning about how to socially interact. Anyway, it’s almost over. And I’m going to try and do things a bit differently. Everything I get set, I’m going to give myself 1/4 of the alloted time to do it. The last 75% of the time is spent perfecting the last 20% of the result, and I don’t have time for that.

The one thing I need is a way of setting a real deadline. I’m going to try and set it internally – i.e. just by self discipline. If that doesn’t work, I’ll try something else. Ideally, I need to create motivational leverage by signing up to 3rd party deadlines with real consequences. Let’s see how it goes.

Decentralise your mind

I’ve spent some time thinking about centralised and decentralised systems over the last few months, and have come to realise that the understanding of decentralised systems is the main aim of complexity science. In a nutshell, we want to know how simple rules make system level behavior – whether that’s so we can understand how a flock of birds fly, or build a software (or social) system that uses agent level rules to produce system level behavior.

There are obvious advantages to systems where the information processing and rules are distributed, and it might even be that we can get more ‘out’ of such systems than we put in, as intelligent behavior just drops out due to the interaction structure of the agents. Even if not, we can still save massively due to the parallel-processing nature of the system, the reduced information transfer time and increased resilience.

Some of these ideas are discussed below in an essay I wrote a month or two ago on the subject.

What is the centralised mindset, and how might complexity science free us from it?

1. Abstract

Increased interest in the study of complex and decentralised systems has prompted some to propose that an understanding of such systems is one of the major issues facing science. Resnick (1991, 1994, 1996, 1997) suggests that the human predisposition of a ‘centralised mindset’ – in which complex systems are by default viewed as centralised – may be overcome by education via the application of complex systems modelling. Arguments for and against the case are presented, with a brief technical description of the relevant terms, and the merits of each discussed.

2. System Types

Decentralised systems can be described as those in which system-level behaviour is determined by low-level interactions of individuals in the system acting according to their own, internally defined rule-sets, without external direction from a ‘leader’ or controlling body (e.g. Resnick, 1996). Such systems are common in nature, with examples ranging from bird flocking (e.g. Heppner & Grenander, 1990), spontaneous traffic jams (Nagel & Schreckenberg, 1992) and path-formation across public parks (Ball 2001), to the market economy (Smith 1776, Farmer 2001). As eloqunelty described by Resnick (1997), the patterns created by flocking birds are not due to a single ‘leading’ bird dictating actions to the rest of the flock, but instead by each bird independently matching its velocity to, and maintaining a precise distance from, neighbouring birds. It is this simple rule-set, applied in parallel across all birds in a flock, that leads to the intricate and graceful behaviour we observe.

Centralised systems, on the other hand, do have some form of centralised controller (e.g. Resnick 1991), or at least an external rule set that is applied across all individuals within a system, to some extent affecting their behaviour. For example, a computer is a system comprised of many components, however, (traditionally) all of the information processing and decision making occurs within the CPU, which then passes information on to other parts of the system, precisely dictating their actions to direct system-level behaviour.

3. The Spectrum of Centralisation

Before proceeding further, it is necessary to accurately describe the ‘landscape’ of centralised and decentralised systems, and separate simplistic rhetoric from discernable fact. Resnick often poses a rift, or ‘simple dichotomy’ between centralised and decentralised systems (e.g. Resnick 1991) to highlight the historical neglect of decentralised viewpoints; however, he admits that this dichotomy is to a large extent manufactured (Resnick 1996).

Looking at other common examples of centralised systems, the fallacy of this dichotomy becomes apparent. For example, a traditional, hierarchical army is an often-quoted example of such a system (Resnick 1997). Traditional armies are controlled by a central authority that uses the hierarchical ‘chain of command’ to direct the actions of individuals, in an attempt to dictate the overall behaviour of the system.  However, unlike a computer, the individual soldiers in an army are not purposefully designed computer components; they have their own predefined rule-sets and decision-making ability, and left to their own devices, would create emergent system-level behaviour that would likely differ from the behaviour desired by the central controller. Hence it is incorrect to classify an army (and all other centralised systems in which humans are the individual components) as a totally centralised system, as the individuals within the system have decision making ability and can choose whether or not to follow rules dictated by the central controller. Furthermore, because each human in the system consists of some 100 trillion independent cells, comprising an extraordinarily complex, partly-centralised system in its own right, a ‘centralised’ system of humans is in fact at best a partly-centralised collection of independent, partly-centralised systems.

Taking this viewpoint, it is clear that the picture is much more complex than suggested by Resnick: As opposed to a discrete binary distinction between centralised and decentralise systems, there exists a continuous spectrum system centralisation. This is an important point, and has a deep impact on the question of what constitutes centralised thinking, and how, whether and why we might want to be ‘saved’ from it: Portraying two viewpoints as extreme, mutually exclusive cases (in this case the centralised and decentralised approach) is a common rhetorical device, and implies exaggeration of the positive and negative impacts of each.

3. The Centralised Mindset

Humans have a predisposition to assume the behaviour of systems that they do not understand is due to the direction of a central controller (Resnick 1991). This ‘centralised mindset’ (Resnick 1996, 1997), is apparent in many areas of society, and appears to be, to some extent at least, hard-wired into the operation of our brains. For example, in the vast majority of cases, when asked, young children invoke the will of a single controlling entity when creating explanations for systems that they do not understand, such as the weather (Resnick 1996).

Origins of the Centralised Mind

Deism is a strong expression of this phenomenon: throughout history, the behaviour of many, poorly understood complex systems has been attributed to ‘god’ – a central controller, who interestingly, in the vast majority of cases, takes on the form or a human or an animal (Hume & Root 1957). One can intuitively speculate as to the origins of such an adaption. During the ancestral evolution of the human brain, the interactions which had the largest impact on reproduction and survival where most likely those with other humans and animals (for example, when hunting, forming social groups, avoiding predation and mating). Because such organisms act in an approximately centralised way, evolving an innate understanding of centralised systems, and having a default assumption that all complex systems are centralised, may well have been the optimal fitness strategy over the course of recent human evolution. Under this hypothesis, humans assume complex systems are centralised by default because it has been advantageous to do so over evolutionary time.

Centralisation and Complexity

This leads on to the question of whether decentralised systems are a) more complex, and or b) inherently more difficult for humans to understand than centralised systems. Regarding objective complexity, as we have discussed in section 2, many systems labelled as ‘centralised’ in fact comprise multiple levels of partly-centralised systems, which, without a central controller, would be highly complex decentralised systems. Hence, from an objective standpoint (for example, in terms of Kolmgorov complexity – Li & Vitanyi 1997), it may be that such centralised systems are more complex than they would be without a central controller, because the inclusion of a central controller and hierarchical information flows just adds to the length of description of the system.

However, in terms of human understanding, and perhaps objective predictability, it may be that centralised systems are less complex than decentralised systems. By definition, the actions of centralised systems are directed by a central controller. Assuming that the individuals within a system follow the instructions of the central controller and have no decision-making ability of their own, to understand and predict the actions of the system, one needs only to understand the ‘motives’ or mechanism of the central controller, rather than having to predict the parallel interactions of many decision making agents. Even with partly-centralised systems such as multi-cellular organisms or societies, it may be that the system behaviour can be fairly accurately approximated by understanding the ‘motives’ of an implied (or real) central controller, rather than having to calculate the cumulative effects of massive parallel agent interactions of the system.

This ‘shortcut’ method of predicting systems which are (or can be approximated as) part-centralised, may have been a central strategy in human evolution. This hypothesis implies that predicting systems with the assumption of centralisation is (more often than not) computationally less intensive than predicting a decentralised system. If true, it may be that the human brain is simply not equipped to intuitively understand decentralised systems. Resnick (1994) remains somewhat agnostic on this point, but argues that exposing people to simulations of decentralised systems can at least dispel their initial, incorrect centralised predictions, if not always allowing them to intuitively create accurate predictions based on decentralised reasoning (Resnick 1996, 1997).

4. The Cost of Centralised Systems

Having defined the centralised and decentralised approach to classifying systems, we might wonder why it is important to consider the implications of one or other approach (e.g. Ball 2001). One reason behind the interest in obtaining an understanding of such systems is to capitalise on their benefits, and avoid potentially costly mistakes through misunderstanding. History is littered with the application of centralised control on decentralised systems, with varying results. Democracy, national healthcare, and standardised education, to name but a few, are all arguably positive results of the application of central control, or centralisation, on a decentralised society. Rules are applied by a central body (government) in the form of laws or regulation that influence the actions of individual agents in the system (citizens), with the aim of producing a desired system-level behaviour.

Parallel processing

Some examples, however, have not been so successful. Communism represents the application of extreme centralisation to the decentralised system of an economic market, and its failure highlights one of the costs of centralisation. A largely decentralised economic market, as in the capitalist western world, is one in which supply and demand are determined by the aggregate trading behaviour of all agents (people) within the system. With communism, the supply and demand decision-making is removed from the people, and is instead undertaken by a central authority, which dictates the results back to the agents. Communism failed simply because the government was not able to solve this resource allocation problem on its own (Beinhocker 2006). The result in the USSR was massive overproduction of some goods and underproduction of others, resulting in chronic shortages: the economy was too inefficient to support its population.

Hence, a major benefit of decentralised systems, and perhaps why they seem difficult to predict, is that they fully harness the parallel processing power of all of the agents in the system. Centralisation negates this processing power by moving all decision making to a central authority. As an aside, regulation, on the other hand, as present in capitalist markets, can be seen as a weak form of centralisation that influences the parallel processing power of the individual agents, rather than removing it altogether, allowing the economy to maintain the majority of its efficiency.

Latency, Corruption and Resilience

In addition to a potential loss of parallel processing power, centralised, hierarchical systems also have a time lag, or latency, which is defined by the time taken for information to pass from the hierarchical controller to the agents. Furthermore, at each stage of the hierarchy, there is a potential for this information to become corrupted. Both of these factors can lead to inefficiencies that are not present in decentralised systems. Moreover, centralised systems are inherently susceptible to failure if the central controller, or information channels are compromised. Again, this is not the case with decentralised systems.

Why Centralise?

With the number of associated problems, one might wonder why we create centralised systems at all. The answer may be that we, as a species, have an innate understanding of centralised systems, and when creating systems we do so in a manner than we understand. By doing so, we may be missing a wealth of efficiency offered by decentralisation that Resnick’s arguments and suggestions (for a deeper understanding of such systems) could provide some real, tangible progress towards.  However, manipulating systems in a decentralised manner involves creating innate, rule-set changes at the agent level – and in the case of societies, agents already have innate rule-sets. Despite that, we attempt to achieve this feat by training, culture and education; however, changing sociological systems at the agent level is a slow process, and history reflects that some form of centralised control is often necessary.

5. Escaping the Centralised Mindset

The desire to understand the centralisation spectrum is not limited to the application of centralisation and decentralisation to sociological problems. Recently, perhaps due to the availability of the processing power capable of modelling such systems, there has been increased interest in understanding decentralised systems from a scientific perspective (e.g. Ball 2001), in an area recently referred to as ‘complexity science’. Resnick (1997) argues that obtaining an intuitive understanding of such systems by modelling, and specifically developing educational models, is important. While that may be true, interest in, and progress with such systems is not novel (e.g. Smith 1776, Darwin 1859), and the current interest could be attributed, as discussed, the recent availability of applicable tools.

Many areas of society and technology have been in fact undergoing a gradual progression away from centralisation for hundreds of years. For example, armies that once lined up to fire muskets in synchronised batteries are barely recognisable from the highly autonomous agents of the modern army (Keegan 1994), and in extreme forms act as a largely decentralised guerrilla force; Monarchies have fallen in favour of more decentralised democracies and business hierarchies have flattened (Beinhocker 2006).

6. Conclusion

It seems that a gradual progression towards decentralisation is apparent, is occurring, and has been occurring for some time, and the present is no exception. Presenting a simplified version of the facts (Resnick 1996), while useful for providing contrast to some potentially important issues, may in fact be more damaging than good. Furthermore, while it is true that the scientific study and understanding of decentralised systems (in the form of complexity science) has increased in the last two decades (Ball 2001), to suggest that decentralised paradigms are on the verge of overthrowing Newtonian notions of cause and effect (Resnick 1994) is overly sensationalist. Likewise, linking the development of complexity science to postmodern theories of cultural relativism and the disproval of objective truth (Resnick 1994), is again, sensationalist, irrelevant, poorly argued and potentially damaging.

Despite these issues, it is clear that some of the major sociological and technological challenges facing the human race at present are intimately linked with the understanding of decentralised systems, and that a deeper comprehension of such systems will likely be important to maintain the current rate of technological progress. Inappropriate application of centralist thinking stands as a hurdle to this progress – however, it is a hurdle that the human race has been gradually addressing for some considerable time, and will in all likelihood continue to do so. Complexity science will no doubt be an aid to this process, but whether it will be a saviour, remains to be seen.

Bibliography

Ball, P. (2005). Critical Mass: How One Thing Leads to Another. Arrow Books Ltd.

Beinhocker, E. D. (2006). The origin of wealth: Evolution, complexity, and the radical remaking of economics. Harvard Business School Pr.

Darwin, C. (1859). On the origin of species by means of natural selection, or the preservation of favoured races in the struggle for life. New York: D. Appleton .

Dennett, D., & Kinsbourne, M. (1997). Time and the observer: The where and when of consciousness in the brain. The Nature of Consciousness: Philosophical Debates , 141–174.

Farmer, J. D. (2001). Toward agent-based models for investment. Developments in Quantitative Investment Models .

Heppner, F., & Grenander, U. (1990). A stochastic nonlinear model for coordinated bird flocks. American Association for the Advancement of Science. Washington D.C. USA.

Hume, D., & Root, H. E. (1957). The natural history of religion. Stanford Univ Pr.

Keegan, J. (1994). A history of warfare. Pimlico.

Li, M., & Vitanyi, P. B. (1997). An introduction to Kolmogorov complexity and its applications. Springer Verlag.

Nagel, K., & Schreckenberg, M. (1992). A cellular automaton model for freeway traffic. J. Phys. I France , 2, 2221–2229.

Resnick, M. (1996). Beyond the centralized mindset. Journal of the Learning Sciences , 5, 1–22.

Resnick, M. (1994). Changing the centralized mind. Technology Review, Manchester NH , 97, 32–32.

Resnick, M. (1991). Overcoming the centralized mindset: Towards an understanding of emergent phenomena. Constructionism. Norwood, NJ Ablex Publishing .

Resnick, M. (1997). Turtles, Termites and Traffic Jams: Explorations in Massively Parallel Microworlds (New edition ed.). MIT Press.

Resnick, M., & Wilensky, U. (1993). Beyond the deterministic, centralized mindsets: New thinking for new sciences. American Educational Research Association .

Smith, A. (1776). An Inquiry into the Nature and Causes of the Wealth of Nations. Adam and Charles Black.

Selfish Academics

Are, quite often, extremely selfish. They become so entrenched defending a closely held viewpoint, the truth of which holds a great deal of real personal benefit, that they:

a. Become extremely close minded about accepting other viewpoints – about accepting that they might be wrong. This often persists to the point that it greatly hinders human academic progress.

b. Massively neglect their teaching responsibilities. This can, in the small scale, involve teaching students their particular point of view with more preference, and not presenting hypotheses in context of global opinion. However, in some cases, this may involve completely ignoring the students learning needs, and essentially totally changing the course content and syllabus just to be able to preach about a particular closely held belief, which will most likely leave the students without the required knowledge for their chosen career paths. This approach is insidious, and a real abuse of power. At its worst, it could be viewed on the same level as teachers forcing their religious beliefs on students. It’s just plain wrong.

Event 20/20 BAS V2′s

Aren’t worth the money. They do look cool though. How do I know? Because I’ve owned a pair for 5 years. The truth is that the KRK rp5′s that I bought in New Zealand (at a fraction of the cost of the 20/20′s) are much, much better.

Yes, they don’t have the bottom end, and only reach lows of about 60 hz. But the simple fact is that tracks mixed on the KRK’s sound good everywhere, not just on the KRK’s. In fact they sound great everywhere. Tracks mixed on the 20/20′s, on the other hand, sound quite good on the 20/20′s, and pretty damn avaerage everywhere else.

Anyone want to swap?

I just

bought some new headphones. They’re AKG 271 Mk II’s. I know, I’m trying to avoid the trap of the capitalist economy and devote my time to more fundamentally rewarding pursuits, but man, they RULE.

All of the other headphones I’ve ever owned, I now realise, made sound out of cotten wool, candy floss and twigs. AKG 271′s, on the other hand, forge sound from molten titanium, diamond and marble. Bullet proof.

New Cars

are ridiculously expensive. And why? What do they do that your old car didn’t do? Is a £1,000,000 Mercedes SLR McLaren really five hundred times better than a £2,000 second-hand Subaru Outback?

It does come with red seatbelts.

Fair enough.

I can’t quite believe

how much my emotional viewpoint changes from one hour to the next. It makes it next to impossible to get an objective picture of the world. What I’d reallt like is a real-time, intelligent monitor system that measures my blood glucose, dopamine, hormones (etc.) and translates that data into a snapshot of my current mental and physical state. It would be helpful to also compare the values against historical / mean data to put the snapshot in context, and possibly, by mapping periodic (daily/monthly/yearly) cycles, predict my baseline mental / physical state at specific points in the future.

On a fundamental level

I don’t actually believe in right and wrong. It’s all a matter of perspective.

I do have a sense of right and wrong; I have a moral compass – I just don’t believe that my (or and other organism’s) moral comapass orignates from anything more than arbitraty fluctuations in a highly complex evolutionary landscape.

Think about this: Is it morally wrong to kill potntially deadly bacteria by cooking your food? No. Unless, of course, you happen to be that bacteria. Then you might find it quite inconvenient being boiled alive and you’d probably think – if you could think, of course – that being boiled alive as a result of doing nothing than minding your own business was, well, quite wrong.

Somebody once told me

they wouldn’t say I had rapier wit; more like a samurai sword.

You could say you caught me

in a transitional period.