Being able to redefine the problem to get at a "quick and dirty" solution is absolutely a useful skill. What I think gets glossed over in the fine article and in some of the other comments, is that this is really only a good approach when one can also see the full solution as well and understands how to evolve past the "mechanical finger" if it becomes necessary to refine the system.
Where people get into trouble is when they have a short problem space horizon and don't have a decent feel for the trade-offs being made and a reasonable understanding of how the "wrong solution" is related to the "right solution". That can lead to the creation of a hack-y, fragile and unchangeable system that can really limit progress in the future.
If you know you're cutting corners, the ability to change in the future should be the priority. It usually means that you need to consider how the "perfect" solution would look like, so you know where the possible refinements will come to play.
Where people get into trouble is when they have a short problem space horizon and don't have a decent feel for the trade-offs being made and a reasonable understanding of how the "wrong solution" is related to the "right solution". That can lead to the creation of a hack-y, fragile and unchangeable system that can really limit progress in the future.