SDHC accepting only improving moves performed better than SDHC accepting non-worsening moves in some cases because SDHC accepting improving or equal moves could choose a different neighbouring solution to SDHC accepting only improving moves when there are multiple solutions of the same quality.

SDHC accepting only improving moves performed better than SDHC accepting non-worsening moves in some cases because of the presence of plateau/shoulder regions in the search landscapes of some instances.

SDHC accepting only improving moves performed better than SDHC accepting non-worsening moves in some cases because it didn’t waste time exploring plateau/shoulder regions.

SDHC accepting only improving moves performed better than SDHC accepting non-worsening moves in some cases because the nature of the search landscape of those instances meant that accepting solutions of equal cost moved the search into a sub-optimal region.

SDHC accepting only improving moves performed better than SDHC accepting non-worsening moves in some cases because accepting solutions of equal cost limits the ability of SDHC to exploit the local search landscape.