Comment by nkurz

Sure, and thus the default should be to assume the user is correct and not change anything. But how much chance is there that the user actually typed Revlon with a capital R? I think practically nil. Instead, something was typed that wasn't in the dictionary and the autocorrect decided that "Revlon" was best fit. Depending on what was typed this might be true by Levenshtein distance (https://en.wikipedia.org/wiki/Levenshtein_distance), but I'm really doubtful it could possibly be the best fit in this sentence by any metric of likelihood that considered context. Hence my question of why context isn't used by most autocorrect systems.