What I found when I got this loco to my bench was that the motor contact strips were not making good contact with the board. This is a common problem and I always solder the contact strips to the board. What happens is that a current has to pass through the motor to complete programming. That is why you may notice a loco move or feel a vibration when sending it a program. On this type of mechanism the motor brush contact strips are held in place against the board by a plastic clip. This clip by itself sometimes does not stay tightly in place.
If soldering is not an option then I would suggest a thin piece of scotch tape be placed over the plastic clip and reaching the gray plastic motor saddle on both sides to help keep the clip from coming loose.
It is even possible that a loco could seem to run OK in DC or DCC and then show this same programming problem. That is because the contact is not good enough to pass enough current to make the programming function work but enough current could flow to turn the motor. I've also seen this happen a time or two on a wired decoder where the connection to one of the motor brush caps was just hanging by a few strands of the wire.
On this Kato E8 it is interesting to note that this particular mechanism was introduced by Kato in 1994 and was one of the first mechanisms designed to accept a board type decoder even though it would be a few years before such a decoder was available. Since then several other models have be been introduced that use this same design on similar mechanisms.
