At what point does it make sense to model the activity? Certain molecular properties will effect the behavior of a system in ways a different elemental combination would not.
The question, for example; 'does morality come from the benzine molecule?' is a failure in this type of analysis. The description of fluid dynamics vis-a-vis the propagation of wavefronts in a large body of water will likely only gain small benefit from discussions of hydrogen and oxygen reactions (at best). Instead things that require a statistical description will after a certain threshold only be burdened by details.
The question then becomes one of picking the right degree of reduction. Below which the information is lost to granularity and above which the involved factors become either irreconcilably large or in other ways of retraction to the system.
This will require a degree of sensitivity to the data set as well as the goal of the inquiry; lacking this information only the grossest of descriptions will be made available.
Again it becomes one of defining a question, allowing for a recursive development to treat the data set will at some point lead to diminishing returns. But should always be considered if there is time and processing power to spare. Otherwise it may be more useful to open multiple conflicting models and look for best fits within each constraint base.
If data is ill-defined, then some form of self referential learning system will be required.
What do you mean by a self referential learning system?
ReplyDeleteGood question. Sorry that wasn't clear. A self referential learning system: Anything capable of being conditioned in the Pavlovian sense would likely be a minimum requirement. A more tangible definition involves the ability of the system to learn about it's own structure and adapt it accordingly. I will spend some time developing this idea in other posts.
ReplyDelete