Archive for January 30th, 2008

focus and context, part IIIp: evaluation and the consciencement

Wednesday, January 30th, 2008

This is going to be a longer post in which I will talk about an additional possible task for the previously proposed internet platform consciencement.org – a parliament for science. (please see also here). Short retrospection: such an internet platform could operate as a global instrument in order to provide scientifically judged answers to complicated questions.

Some possible problems which could be adressed on such platform were previously mentioned, today I would like to draw attention to a very special agenda of problems, namely the problem of evaluation, i.e. the assignments of value to a thing, with an emphasis on evaluation in terms of economics, which is usually called “pricing”. In other words: The platform could serve as a tool to provide indications for pricing.

In order to explain what I mean by that and what could be the motivation for setting up such a thing, I will need to make a little excursion on what is involved when a value – a price – is assigned to a good. In particular it will be necessary to discuss main forms of values in order to understand the main mechanisms of pricing and thus serve as an explanation on why the current pricing mechanisms in finance are insufficient (at least in my point of view) and how they could be enhanced by a platform like “the consciencement”.

I will do this in a rather simplistic approach. Simplifying often helps to shed light on the main constituents of a system. However simplifying to much may rather blurr the involved main mechanisms.

short remark: In the series focus and context I am collecting and sorting various issues which are related to principles of structure formation in a cognitive sense. Evaluation is part of this, so this post will also be part of the focus and context series.

The first part of the series focus & context is mainly dedicated to questions related to the issue of “attention”. Here a lot of references were made in relation to the socalled advertisment business, in particular questions like evolutionary needs and branding were discussed. The later parts are more devoted to “choice” or “filtering”.

A main method of “filtering” is that you are given certain values, which allow for some kind of order. The question of assigning values is somewhat settled between “attention” and “choice”, since by raising attention (like for a good) I may implicitly raise the “value” of something in order to make it a preferred choice. However in between attention and choice is the process of acknowledging this value.

In order to study the process of evaluation or pricing one should discuss the notion of a “value”.
For brevity I shall call both -the name of a value and its actual quantity – a value. Its should be clear from the context what is meant.

In the following I will distinguish between three different types of values. I will call them measurable values, which are in opposition to rather “emotional values” such as what will be called subjective values and rational subjective values .

If you have a sieve and a bowl of sand then the large sand grains stay in the sieve and the small ones drop through, i.e. one can order the grains according to their size and hence the here involved “value” is called “size”. Moreover there was no need to “invent” this “value” as it was physically already immanent. So a measurement (in the example the measurement of the grain size) is a process where a value is assigned to something, i.e. it is a kind of evaluation. In the case of the measurement this evaluation is done via a physical process (sieving). In the following I would like to call also processes, which can be mathematically quantified, but which depend on a set of values measurements and the corresponding values measured values relative to something. So e.g. if I have two grains of different weights then I call their compound weight a measured value relative to the given individual weights. (Likewise on could see the grain size as a measured value in relation to the size of the sieve holes). However I do not want to fix this notion too much. The important point here is that for the determination of the value we make use of scientific procedure which is more or less exact repeatable and describable. Nevertheless it should be pointed out that the choice of what and how is also related to “attention”. I.e. if we stick to the example of the sandgrain then the “different sides of the sand particle” caught our attention and made us “prepare” the experiment with our sand and sieve in this particular way. This is no bad thing per se but it is important when talking about objectivity in scientific reasoning. Good scientific reasoning tries to reduce the attention factor.

Another method of assigning values is by a subjective and mostly emotional judgement, like e.g. to assert that one likes one person more than the other or both equal etc. means to assign a subjective value. Here it is even harder to seperate the choice of assigning a value from the issue of “attention” especially if these emotions are linked to evolutionary needs. However it is possible to a certain degree. In particular values which were agreed upon within a community/collective are in some sense subjective, however they are moderated by the collective knowledge. For that reason I would like to call these kind of moderated subjective values rational subjective values . Among others they are intended for detaching the subjective emotional evaluation of an individual from highly fluctuating mood changes. An example: If someone stomps on your feet, your reflex could be to stomp back, however usually you would decide to maybe turn away grudgingly because you dont want to behave badly.

A constitution consists to a great part of or is based on “rational subjective values”. Assertions like wether death penalty is acceptable or not are rational subjective values. There is no scientific method to measure wether killing someone should have the value “good” or “bad”. Some communities state that such values were assigned by deities (like e.g. “you shall not kill” in the bible) however they are still something which had been agreed on by a community, so I’ll keep calling them “rational subjective values”. Thus an important feature of rational subjective values is that they are often written down somewhere.

The motivation of a collective to establish rational subjective values is – as already indicated – in order to moderate between the sometimes rather immediate values of the individuum and values which concern the collective, like e.g. how to make a collective survive or to work more efficiently, which often includes nonemotional measurable values such as size of water ressources etc. (on the efficiency of a collective please see also this randfrom post about trust.)

So the term rational subjective value is always in relation to a collective. However a collective is usually part of a bigger collective. If the size of the smaller collective is small in comparision to the bigger collective (like a family in a nation) then the values of the smaller collective shall for simplicity also be regarded as “individual subjective” with respect to the bigger collective. So families or companies may be regarded as “individua” in relation to a nation, i.e. they are likely to adopt the “nations values”. If the groups are rather bigger (like e.g. migrants) than their values usually have to be regarded seperately, but this can also be done with respect to a nations values, i.e. they may add different values or reevaluate a nations “rational subjective values”. Rational subjective values can thus be seen as a kind of average value, averaged in terms of population and time, but still belonging to the individuum (like in terms of a nation these could be called cultural values, in terms of a religious group, religious values, in terms of consumers, consumer values etc.)

Under generic conditions one can assume that a collective behaves mostly according to the given physical circumstances (measurable) and their collective desicions (based on the rational subjective values, laws, rules). However as pointed out above this behaviour is “blurred, disturbed” by the subjective individual values, depending on the size and nature of societal/collective control.

A collective chooses their values often in order to achieve a set of common collective goals. E.g. the rational subjective value of some christians “to explain christianity, to evangelize is good may lead to the common collective christian goal to make everybody turn into a christian.

An important point here is that it is somewhat possible to determine to what extend a choice of rational subjective values meets a common collective goal or in other words: a collective rational subjective value. This can be e.g. done by observation. Another example: if a society holds the rational subjective value: killing is not good (you shall not kill) and controls this rational subjective value e.g. via law enforcement then one can observe wether this reduces the overall killing rate (a collective rational subjective value) in that society.

Often collective rational subjective values compete with each other so it is usually never possible to reach one common goal. (like if a religious group decides to advertize their religion in a nonforceful obtrusive manner, as their common goal is “not to force people into their believe” then this could mean that they may gain less followers). So in short: rational subjective values are somewhat intended to meet collective rational subjective values in an optimized way.

Interestingly two important physical quantities enter here, namely time and space. space enters via the questions of how the individua are connected and having their values compete with each other (the connectivity and the amount of individua has a geographical/spatial component to it), whereas time enters in that it is needed for observing the collective process, which includes the formation of “rational subjective values” and likewise the observation wether “rational subjective values” are in accordance with e.g. a collective rational subjective value.

An important fact is that the duration and size of the involved time and space can be “abstractly reduced” by mathematical computations which model the involved system including its values and the procedural connections (dependencies) between these values. But of course the mathematical model can only approximate the real outcome.

For the accuracy of a mathematical prediction the correct assessment of all individual values as well as all measurable values and their interplay is essential

For large collections of values this usually turns out to be feasible if one restricts the sample size and identifies the main modes of interaction. Lets call this a choice of sample.

So the accuracy of the mathematical prediction relies to a great part on this choice of sample and the stability of that choice over time.

Roughly one can say that this mathematical choice of sample is easier and more likely to be accurate if the involved values and their interplay can be measured or when they are fixed in laws/rules etc. i.e. if they are rational subjective values. So this is why comlicated physical predictions (like about climate change) can be quite accurate (as they include mostly measurable values), whereas predictions which involve subjective values, like this is often the case for economic values are usually not so accurate. I.e. the more an evaluation is less “rational” (in the above sense) like in panic buyouts at stock markets or like in the recently observed desaster at the societe generale, the harder it is to determine the main involved values and their interdependencies.

In economy the evaluation (the assignment of values) is usually done via pricing (I will call incentives such as interest rates etc. and other economic judgements also “pricing” in order to make things simpler). So for example the price of a resource depends on the amount of the physically accessible part of that ressource (measureable), the mining regulations in a country (rational subjective values) and e.g. the political situation (these are subjective values as they imply that no collective agreement on values had been achieved).

Lets look at another example, which – due to recent events (the moving of a Nokia plant from Germany to Romania) – may be worthwhile to investigate, namely the value of labour. In a collective, like a nation the value of labour depends among others on

-skill (time and quality of education needed for that skill, amount of people which are in principle capable of acquiring that skill) [more or less measurable value]

-educational resources (amount of capable people who can be trained) [measurable value]

-basic living costs (relative to given prices, like to a basket of available commodities (Warenkorb)) [measurable values in relation to the basket]

-geographical flexibility (this applies to employer as well as to employee: a farmer can hardly move, if his soil is under drought, whereas an agent in a Call Center is not bound to location, however this flexibility may depend also on rational subjective values such as language (journals, journalists, lawyers etc.) and immigration/principal deployment laws. The geographical flexibility also determines how much labour costs are in dependencies to labour costs in other countries. [depending on rational subjective and subjective values]

-negotiation (the prices are under negotiation between the employer and the employee. These depend on cultural (like gender and ethnicity biases) and political powers (trade unions, exclusivity of skill etc.)) [mostly depending on rational subjective values]

-political (national and international politics may subsidize certain sectors, overall conditions may be unstable (which secretely enhances costs) etc. [dependent on rational subjective and subjective values]

The above example illustrates again how mostly subjective values make the pricing into a complicated issue.

Consequently failures in economic predictions are most likely less due to overworked quants (mathematicians/physicists in finance) (although one should definitely think about how to make also these people happier) but due to the fact that economy involves non-rational judgements/values.

(As a side remark: It is illustrative to discuss the NOKIA example based on these criteria, since it may display that romanian workers are not necessarily working harder than germans for the money, but that mostly the living conditions are partially cheaper and that the social costs/investments are lower. In particular it is probable that the working conditions (work hours, vacation time) are not as good as in Germany (part of negotiation), however this point most likely doesn’t account for the vast differences in wage. Moreover it displays that people with the skills required for working in that NOKIA facility are relatively easy to find. Moreover due to high moving costs, immigration, language, cultural and social conditions (and social problems) and to the probably a little smaller Warenkorb (less goods can be purchased for the wage) presumably not too many germans will move to Romania in order to keep working for NOKIA.)

Usually the above kinds of analysis’ are done by e.g. traders/economical analysts. Economical decisions are made upon their judgement and ability to assign prices. This is also why they are usually rather well paid. This fact by the way holds true since the beginning of trade. A good trader was able to (more or less) correctly determine prices (usually assigned to a grographical location) in order to know where to buy and where to sell. Of course part of the job of a sales/trade person is also negotiation, but this changes a price usually only partially – the main point is the ability to correctly determine and analyse the prices (last not least the negotiation itself depends on this ability).

The correct assignment of prices can get arbitrarily complicated and so nifty mathematicians use for that problem awkward sounding tools like risk calculus of von Neumann – Morgenstern or Arrow-Pratt risk aversion.

Moreover various platforms for financial analysis’ like e.g. Bloomberg L.P. provide background information (including political) and software tools.

So why should a scientific platform like consciencement provide a similar service?

As I already pointed out before a lot of the pricing mechanism depends on the ability of traders to analyse and judge about prices. Here a funny feedback happens. In particular the accuracy of pricing depends also on the property of being tradeable. Or in other words if there is no trading reason for assigning a price to something then this missing price will destort the actual price. This explains why such subjective values, which are often even rational subjective values for nations (!) like “a healthy environment for future generations” or “human social conditions” are not necessarily rational subjective values for traders. This implies that the pitfalls of capitalism are also due to that “organisatorial problem” of an inaccurate and/or missing pricing of values beyond the trading scope. It makes the thinking about capitalism less into an ideological question.

However it is of course possible to assign values/prices beyond the trading scope. This has e.g. been demonstrated in the socalled Stern review, where the future costs of climate change were estimated. However these prices are only sparsely reflected in nowadays economy and trade.

Similar estimations could be made for nuclear energy. Nuclear energy is currently (especially with respect to regenerative energies) relatively cheap, however judged by environmental costs and future environmental costs (like over the next 20 000 years) it is very expensive [this is a measurable value]. So the actual traded price for nuclear energy is simply one thing: wrong.

Concluding: There is a need to analyse the price of goods also with respect to apriori nontradable values, like pollution, social conditions, future living conditions etc. As already explained an economic platform will do this only to a limited extend. Future institutes have usually only very limited resources to provide such a service. A global platform, where economy/financial mathematics departments of universities could bring in their expertise, where agriculture experts, climate scientists, cultural experts etc work together would have a different quality.

As a side effect such a service could even have an economic value, as e.g. insurance companies often need to find substitutes for these pricing informations in things like “future risks”. This may encourage on the other hand the financial world to donate financial information to a platform, like consciencement.

Last not least a more detailled and true analysis of prices would allow for political measurements similar to the emission tradings. One could e.g. introduce a similar thing for the future costs of nuclear waste, for deforestation etc. One could introduce certificates for maximum work hours, health benefits to employees etc. A more correct and scientifically sound determination of a price is a necessary condition for that.

Remark: Please feel especially encouraged to comment on the above proposition, let it be via comments in the comments section or via emails: nad ^^@^^ ^daytar.de. If there is enough interest then I may bundle the whole proposition about consciencement which had been now spread out over three blog posts into one document.