Most of this I learned from this book:
http://www.amazon.com/Bicycling-Scie.../dp/0262731541
Which is probably the best source for the not fully understood topic of bicycle stability.
You balance a bike and it's load by steering so as to keep the wheels under the center of gravity. With good geometry the bike helps a lot with this. Stay with me, I'm getting to the topic of this thread.
If you ride through a puddle then swerve back and forth a bit, you will see that the tracks of the front and rear wheel take differing paths. To match my description, this should be done so that the handlebars never remain in one position. Observing the tracks, the front swings wider, and the tracks are offset. What you can't see from the tracks is that the phase of the two tracks in TIME (the tracks only show position) is shifted by 90 degrees...The rear wheel is discribing a cosine curve while the front wheel is discribing a sine curve. This is because the back wheel steers only after the front wheel has translated to one side...the steering of the rear wheel is delayed.
This means that when loads are located forward on the bike and you tip to one side, you quickly recover when you steer and bring the front wheel under the load. The front wheel moves more than the back, and there is no time delay between when you turn the bars and the front wheel starts moving back under the load.
When the load is at the rear of the bike, you have to steer the front wheel a long way in order to get the rear wheel of the bike moved back under the load. Not only is more steering required, but the timing is delayed, which creates a feeling of sluggishness.
The worst is if the load is behind the rear axle. In this case the initial response to steering makes the tipping worse, and it is only after the whole bike starts heading to one side that recovering of balance starts. Tail heaviness also negates some of the tendency of a bicycle to self recover due to the rake and trail of the steering.
Note I am not saying the handling is bad. Pickup trucks handle just fine mostly, but they are way different from sports cars. Humans are quite adaptable, and can quickly learn to adapt to most handling differences caused by load location, longer wheelbase, etc. Also, the load on the bike isn't just what you are carrying, but the bike itself, the rider, etc. etc. With the rider forward, and seated high, you can and do "over" balance the rider to compensate for not being able to balance the load on the back as effectively. This happens naturally without requiring any thought. It maybe "feels a little weird" at first then you adapt, forget, and wonder what the newbs are on about. Then the stoker gets off the tandem and it all feels weird again 'till you get used to THAT.
If it feels too weird, it may cause a green rider to compensate in ways that hurt rather than help. In aviation this is known as "Pilot induced oscillation".
Food for thought though: When you carry a load on the rear of the bike, it requires more steering the heavier the load. When you carry a load on the front, it requires less steering motion the heavier the load. EITHER can result in a green rider having balancing troubles. Rear loading make the handling more sluggish, front loading makes it more sensitive.
The above assumes the loads are well secured to the frame of the bike. Things can change a lot if the load can shift or it is secured to the forks. (It can help or hurt, depending on details) It also reflects what inputs are required of the rider to stay balanced. Bicycles will to one degree or another also act to balance themselves without any rider input, but loading will either improve or detract from this, and exactly how is best determined by experiment.
The main points are:
- loads are going to feel different depending on how they are distributed.
- in most cases, forward loading makes balance more responsive to steering inputs, and rearward loading makes the balancing more sluggish. Either change can cause trouble if too extreme, but mostly it is something a rider can easily adapt to.