My instincts tell me that the door is good on the cost-benefit calculation. We have had multiple instances of intrusion into the cockpit over the years, in some cases with fatal results, in other cases with the planes being diverted. This type of events are getting increasingly rare, and I think we can attribute that to improved security. The last decade there has been less than 6 hijackings a year, while in the 1993 alone there was 36 hijackings.
Yet we have had about 10 cases of pilots deliberately downing traffic planes in murder-suicides in the last few decades. It would be good if we could include both threats and build a security system to prevent them. The two-man-in-the-cockpit concept is one easy way, but it's expensive and not obvious that it will always work. Federal Express Flight 705 is one example where it barely worked.