I recently answered a question which at the moment is titled
Hot to find the closest free points on grid
The typo is not the issue here. Now I wonder whether anyone is likely to find that question and its answers, using the search facilities SO provides. One problem is the high number of potential synonyms (or at least synonyms in this context):
- grid = lattice = integer vectors = pixel offset
- distance = length = euclidean norm = radius = radii
- increasing = ascending
- ordered = sorted = in order
- closest = nearest
- shape = object
Other answers here have suggested including these alternatives into the main body of the question, but that doesn't feel right here. Putting another name in parentheses after the main term will make the text harder to read, and add no value to the reader, as the synonyms are clear and obvious enough so they don't help towards understanding the text. Using different words in different places would cause me as a reader to wonder whether the change in terms implies a change in semantics as well, so that would be quite confusing.
You might also argue that as the alternatives are obvious, the searcher could simply try different forms. But keep in mind that there are multiple keywords with multiple alternatives, so the searcher would have to not only try all possible words, but all possible combinations of words. Given the fact that SO apparently uses the disjunction (“OR”) of its search terms, it might even work to specify all terms in a single query, but this aspect of the search is far from obvious, and results matching more words will still get sorted to the top, so you can't differentiate whether a post really matched more of the concepts you requested, or simply more alternatives of the same concept.
So I thought about adding a separate section. Perhaps not titled “Search keywords” but instead “Alternative titles” or similar, which would contain a bullet list of alternatively phrased question titles that among them contain all the relevant keywords. I wouldn't make the section artificially small, as I think that a reader might want to read these as well, and if not, he can simply skip the list after he figured out its nature. The list would probably look something like this:
- Enumerating integer vectors in order of increasing length
- Finding nearest lattice points matching some criterion
- Generating pixel offset coordinates sorted by ascending euclidean norm
- Placing objects on grid avoiding overlapping shapes and minimizing distance
But seeing this discussion here makes me wonder whether that approach would be acceptable. So I'll wait for a few more votes here, and see which directions they take, before employing that technique.