2.2 KiB
aliases, title, tags
| aliases | title | tags | ||
|---|---|---|---|---|
|
Heuristics |
|
Heuristics
A heuristic, or "rule of thumb", is a solution that neglects some of a problem's complexity, while still producing a satisfactory result.
The namesake rule of the rule of thumb is that the width of one's thumb is about an inch. With that heuristic, one can closely approximate small dimensions, even without the proper tools.
It is tempting to believe that heuristics are not ideal, that the most accurate method ought to be preferred, but the most accurate method is rarely practical. Afterall, even using a ruler is a heuristic, subject to the tool's tolerance and reliant on its proper use.
The suitability of a model is dependent on the purpose of the measurement. To determine the width of dimensional lumber, for example, to determine if a board is a 2x4 or a 2x6, the rule of thumb is perfectly suitable, and a micrometer would be silly.
[!quote] George E. P. Box, British statistician All models are wrong, but some are useful.1
Realism vs. Instrumentalism
Realism holds that the purpose of scientific research is to describe the world as accurately as possible. Instrumentalism (anti-realism) argues that the purpose is to forecast as accurately as possible.
[!quote] Opinionated History of Mathematics, "Did Copernicus steal ideas from Islamic astronomers?" (pp.) Ptolemy's lunar model is "flawed" in that it inaccurately describes Luna's distance from Earth, however Ptolemy only ever used the model to predict eclipses, which the model does very well. The discrepancy is only a flaw from a realist perspective.
Realism is noble, granted, but instrumentalism is far more practical.
See Also
- Map--territory relation - Wikipedia
- "Mistaking the map for the territory" (logical fallacy)
-
↩︎[!quote] box-et-al_1987_empirical-model-building#The Use of Approximating Functions The fact that the polynomial is an approximation does not necessarily detract from its usefulness because all models are approximations. Essentially, all models are wrong, but some are useful.