It's a long article and I didn't read all of it, but the changes they made seem more fundamental than that:
A year after the change, the results of this “extreme makeover” were striking: Not only had congestion decreased in the intersection—buses spent less time waiting to get through, for example—but there were half as many accidents, even though total car traffic was up by a third. Students from a local engineering college who studied the intersection reported that both drivers and, unusually, cyclists were using signals—of the electronic or hand variety—more often. They also found, in surveys, that residents, despite the measurable increase in safety, perceived the place to be more dangerous. This was music to Monderman’s ears. If they had not felt less secure, he said, he “would have changed it immediately.”
That the locals who know the roads perceived the intersection as more dangerous indicates to me there's something more than "people familiar with roads don't need signs."
So if one would believe in a universal "fair trade" principle, the cost you pay for security and fluidity is heightened attention from the participants in traffic. One would wonder how well would this scale: if drivers had to pay the same attention to every intersection, how long before they would start making mistakes?
So, you mean that the locals can ignore traffic signs because they know the place? Like in "oh, yes, there's a speed limit, but it's useless, see, the street is safe enough, screw that". That could be actual quote, I've heard it so many times. And I guess it's much better to remove such sign and make the street actually look like it's not safe so you have to pay attention.
And of course, routine and automatization are a grave danger for any driver. Especially at the roads where the conditions are constantly changing. Traffic signs would just provide a false sense of security.
Signs and indicators are for people who are not from the area.