Performance of currently untested alternatives

Previous  Top  Next

Presently unobserved alternatives can be judged regarding their likely success. That is because variants that differ from known acceptable alternatives only in unimportant dimensions, will be most likely successful as well. And variants, that differ from all known successes in important factors most likely will be failures.
Using all known examples and a relevance hypothesis the success of hypothetical variants can be guessed. Those inferences should not be based on variants that are just acceptable and close to failure. Even minor further deviations from the ideal values lead to failure. That is why variants with supposedly ideal values should be preferred as a base for evaluations of currently unobserved examples.
Variants that differ from successful examples only subcritically will be more likely successful than a ramdomly created example. Combining new variants guided by former successes is advantageous.

A tv company hires a famous entertainer who had considerable success with his 8-o-clock-show. If the timing of the show would be a subcritical factor, the late night show with the same entertainer plausibly would be a success as well.

Variants that differ more than critical from all known successes most likely are failures. This implies that this new variant does not coincide with a formerly unknown ideal values. In this case, despite the critical variance this new example would be a success. This would put the relevance hypothesis in question as the assumption would be violated that from each ideal value at least one example would be known.
Of little interest is the relationship that can be established between failures. Alternatives that differ subcritically from other unacceptable variants likely will be unacceptable themselves. Infering relevance from comparisons of failures is a too weak justification for valid conclusions. As factors are assumed to be important the number of possible failures is much larger than the number of the possible successes. Consequently, inferences are restricted to only positive examples.
One interesting question is whether it would be possible to find acceptable variants just by differing enough from all known failures. This seems not to be sufficient justification to expect successful variants. This follows from the assumption of important dimensions and a much larger proportion of unsuccessful alternatives.
The search for successes would be easy if avoidance of non-ideal values necessaryly would lead to successful variants.

A tv station assumes high relevance for the factor "on air time". If avoiding inacceptable variants would lead to successes, the company could expect a success when a showmaster fails with his late night show and later is assigned to a breakfast show.

A brief conclusion of the previous line of thougths: The relevance is inferred from successful variants. From knowing observed examples and hereby inferring the relevance of the contributing factors the analyst can create hypothesis concerning the success of new and unobserved variants.
Candidates are plausible as a success, if they differ from at least one known success only in factors that are less than critically important. Variations from all known successes that surpass the critical value of 1.0 lead to expecting a variant with inacceptable success. Still, it is possible that the differing factors are less important than expected and do not endanger reaching acceptable success.
As well, it is plausible that variants fail if they differ more than critically from all known successes. Even if different in many important factors from all known successful examples, a variant may be successful as by accident a new ideal of variants is discovered.
Successful variants should be expected if the values of the important dimensions are the same as in acceptable other examples. The same effect is reached by avoiding the values of inacceptable examples in all the important dimensions. The strive to discover new ideals encourages to neglect the conservative tendency to maintain successful values in highly relevant dimensions.