An approximate answer to the right question or an exact answer to the wrong question

Question

There is a quote from John Tukey in one of his papers on data analysis

Far better an approximate answer to the right question, which is often vague, than an exact answer to the wrong question, which can always be made precise.

Does something similar hold in the OR community as well, in general?

As an example, suppose a complex problem which is hard to model and optimize, is it better to try and get an approximate solution of the exact problem, or is it better to try and simplify/approximate the problem such that it can be solved exactly?

Answer 1

Yes.

I believe the same analysis can be made for OR. In the industry, it is far more important to solve the actual problem as well as you can than to obtain the perfect result on a different problem.

The business objective is often vaguely defined, and the data is not perfectly accurate. Stated otherwise, it is important to make sure that you are solving the right problem with the right data, rather than solve it perfectly.

However, there are a few caveats:

• you do not need an exact model of reality either, just something close enough for the business needs

• solving a different problem exactly may be the best way to obtain good candidate solutions (or yield a starting point for a heuristic)

• giving a lower bound on your solution's quality, or an optimality proof, even on a slightly different problem, may be the best way to convince some stakeholders

• this does not apply to most academic publications, as the problem to be solved is usually well defined from the start