The Pankian Metaphor 3087

the other way of looking at it is that the "market" derives its efficiency from supposedly being a dynamic adaptive feedback system.

i had done a lot of dynamic adaptive feedback systems as an undergraduate in the 60s ... my implementations being picked up and shipped in products. some of the implementation was dropped in the morphing from 360s to 370s in the early 70s ... but a couple years later I was allowed to reduce it as separate Resource Manager product ... the 30th ann. of the "blue letter" announcement coming up on May 11th.

so a big part of a dynamic adaptive feedback systems is

1) the decisions are based on accurate information (requirement for good metrics and instrumentation)

2) effective control mechanisms.

The efficiency of the "market" feedback control system is, in part, based on what the "market" (consumers frequently, but not necessarily exclusively) are willing to pay (as an effective control merchanism, front #2 above). The problem is that subsidies can significantly distort the accuracy of the information (true, fully loaded infrastructure costs ... as opposed to some price that may have little or nothing to do with fully loaded infrastructure costs).

So not charging the fully loaded price ... can significantly skew the "market" behavior as well as the efficiency of the "market" paradigm.

A trivial example goes back to one of the opening examples about i.e. like in support of miniciple water purification plant. There becomes a gov. policy that the building of miniciple water purification plants should be encouraged for the good of the community and nation. Maing the bonds for building miniciple water purification plants mean that the local district doesn't have to pay as much to attrack investment (reducing the cost to build the plants).

For a little drift from dynamic adaptive feedback market forces ... you can have quiet a bit of confusion in legislative bodies when they have difficulty in reconciling

1) policies encouraging the building of water treatment plants through attract investment for the building of the miniciple water treatment plant)

2) policies where everybody should pay minimum (same or larger) percentage tax on earned income (regardless of the source of the income).

Back to dynamic adaptive feedback market forces ... the "market" as a dynamic adaptive feedback system can have a very short period (possibly because lots of people have difficulty looking ahead and-or they lack any long term memory ability).

Subsidies can adversely affect the efficiency of dynamic adaptive feedback "market" by distorting the information (aka prices) that decisions are based on (in effect, efficiency of the "market" is based on prices having something to do with the buttociated fully loaded infrastructure costs).

However, dynamic adaptive feedback "market" .... even with accurate data may result in less than optimal results if the window that it operates over is too small.

Part of calibrating the resource manager was a set of couple thousand automated benchmarks that took three months elapsed time to run.

After the initial thousand or so benchmarks (large variation in configuration, workload, and control functions), the remaining benchmark characteristics were under the control of a variation of the performance predictor (implemented in APL).

One of the things that the performance predictor implementation was programmed for was "hill climbing" algorithm that basically varied some number of benchmark characteristics looking for optimal solutions. A well recognized characteristic of such a methodology, is that if you take into account too small a scope ... you may find the top of a local hill (i.e. local optimal solution) ... but fail to find the top of the highest hill (global optimal).

sometimes, gov. polices can be characterized as totally distorting information (possibly via subsidies) used by the "market" to arrive at solution ... resulting in totally undesired outcomes. other times, gov. policies may skew the information used by the "market" to buttist it in arriving a more optimal solution ... by skewing the "market" information (subsidies, tax credits, etc) based on larger scope (time) calculations than the "market" normally operates with.

One might postulate that the example of the country insisting that all major companies invest at least X percent of their profits in IT innovation (or the gov. would take it as taxes) might be example at attempting the later.

misc. past posts mentioning the performance predictor

--