In these situations one cannot just make an attempt and cease doing it if it doesn't work.
If you have no "handle" where to keep yourself your attempt to lean over the embankment to lift the other person can easily make you fall inside the trap. The man is grasping your hand and pulling you toward him with all his weight (as he's lifting himself) and you will not manage to lean backward. Gravity will work against you "big time".
A different case would be if, say, the man already had a knee above the embankment and you only have to help him for the last push.
For me it would be basically impossible to lift a man who is inside a hole up to his chest just by the sheer force of my arms/legs.
Where were the guard rails, the security, track and train avoidance equipment, and on and on. Statistics, that's what it comes down to. It was predetermined that xxx number of people would be injured and or killed in the subway. There are no absolutes in this world. Statistically next time or two maybe some citizen will save someone. And out of the saves stistically a few will be given the hero award as outstanding citizens which makes everything look really civilized.
If the people in this thread had their vote on whether to help or not help it would graph out as a nice bell curve.
Obviously the train driver now has to stop in exactly the right place!
Everything in this thread is pure speculation. What could have been done, what should have been done, what you or I could do, or would do, will always be speculation. We do not know.
I have on occasion saved people from greater or lesser peril at some risk to myself. On a few occasions I have chosen courses of action or inaction that were short of "heroic". I can't tell you the exact reasons for either, it just happens. The outcome of any of it, as anyone who has been in these kinds of situation will tell you, is pure fate. There is no conscious decision, you just do or do not, and you don't know which it was, or what you could or should have done, but didn't, until it is over.
What we do know is that a man pushed another man off the platform in front of an oncoming train. We know who was killed and by whom. Everything else is lost to the fog of war, even, most likely, to those present. Perhaps for someone who was present there is some miserable hindsight available. It may or may not be 20/20, and i feel for that person. But there is no hindsight whatsoever for armchair speculators. We like to talk big, but we weren't there. We don't know. I find it fascinating that so many think they do.
With the double doors the persons trying to catch the train no matter what will slow the closing of the doors much more I suppose. Some people might remain with bags, umbrellas etc. "in between".
Now that I think about it, one of the things that slow underground train operations is that people must first get off, then people can get on. People trying to get on, in crowded stations and situations, are an obstacle to people getting off.
With modern trains which have carriages communicating with each other (a long "snake" where you can walk freely) tube trains could be optimized so that let's say the first carriages are for getting on only, and last carriages are for getting off only. That would make the pavement traffic "one way". All the internal paths of the tube stations could be optimized so as to avoid double senses of circulation.
I know some cities (Toronto? Boston?) have subways where there are platforms on both sides of the train, so the doors can be opened on one side for passengers to exit before the doors on the other side of the train open to let passengers enter. The (relatively) new AirTrain to JFK airport in Manhattan has glassed-in platforms with sliding doors. New York's subways are old, and I could imagine any kind of retrofitting would be a challenge and a huge expense that most people would argue would be better applied to building more lines to reduce crowding.