Abd Ul, Peter, this is a very enlightening discussion.
let me comment on some issues, where I hopefully can contribute something of value. Let me concentrate on one. 'reliability' Abd Ul says. >>>>>>>>>>>>> A complete theory will include explanation of the variability (which you call "the reliability problem"). Once we know how and why something is happening, the possibility of control can open up. Not necessarily, mind you. But it's probable. You don't have to imagine the specifics, just recognize the possibility. I hope I have shown that the variability does not make investigation impossible -- and it can even facilitate it in some ways, as long as some minimal level of "success" is obtainable, through the power of correlation. >>>>>>>>>>>>> I am very much with Abd Ul on this one. As far as memory serves, the Pons/Fleischmann Pd-material has been delivered by an italian manufacturer, who had a very peculiar way of processing the material. Because P/F were not aware of that, they did not disclose it as relevant. So also did not the replicators. So they failed. A 'reliability' problem? Not so much. Nowadays there are multiple reports concerning the materials composition. (Btw, anyone arguing with secret ingredients/catalysts/secret sauce like Rossi does, does a dissservice to the scientific method and be rightfully excluded from the community!) If only one of six experiments is giving a 'positive' result does not invalidate the single one, as long as this single one can be reproduced with >50% probability. To term this 'unreliable' would be a misnomer. It is just that we do not know where in the parameter-space (in a very wide sense, which could even include the experimenter himself) the core of the effect is located. One possible solution to situations like that would be a) to use some variant of the 'morphological box' ala Zwicky (->wikipedia) to identify the dimensions, along which the effect could be scaled. b) use the evolutionary strategy, as developed by Ingo Rechenberg, which dates back to 1971 ( http://en.wikipedia.org/wiki/Evolution_strategy ), still wildly underappreciated. As we know, 'evolution' is a very 'unreliable' process in the first place, based on trial and error, based on the triad variation-adaptation-reproduction, which is not very much appreciated in the scientific community outside biology. One problem with this approach is, that it is sort of incompliant with theory-based approaches, which are basically deductive. (Hard) Scientists like to deduce real-world instatiations from last principles. Engineers are more pragmatic. Anyone remember 'fuzzy logic' or 'neural nets' ? Or some hundred Neurons extracted from a frog brain, trained to control a microsoft flight simulator en par to real pilots. Embarrassing. Why? It works, but we do not understand. We ultimetely want to understand, or mentally anticipate why we have a safe flight. Right? As an engineer I have to acknowledge this urge of the physicist and the user, who want to have mental models, different as they are. Guenther
