In order for the plane to sustain a given turn, say 6g, a certain download on the tail is required. This download is spread over the entire surface of the complete horizontal tail. Now, take away half the horizontal tail and that 1/2 surface area must now generate all of the download if the pilot's going to sustain a 6g turn. That means that 1/2 the surface must generate twice it's original load in order to make up for the loss of 1/2 of the total surface area. The sailboat idea is looking at the same thing but from a different perspective. Take away half the surface and 1/2 of the power is produced, i.e., the boat will slow down. To sustain the same speed the remaining sail must be retrimmed to generate twice it's original load. Of course there are big limits to this. Can the remaining horizontal tail surface have enough elevator travel and stall margin to generate twice it's original load? Probably not. Can the remaining sail be trimmed to produce twice the load it was originally? Probably not. This is grossly oversimplified since jibs and mailsails don't generate the same power but I think you get my point.
Also, the distribution of the load is very different. Think of a complete horizontal tail as a teeter totter board with the fuselage/horizontal tail joint represented by the pivot. Lift up equally on both ends of the board and board will not pivot and the load at the pivot point is directly vertical. Now, cutoff one end of the board and lift the other end and it'll pivot because there's no counterbalancing load on the missing end. That's the torque I'm talking about that is transmitted through the fuselage/tail joint. That joint isn't designed to counteract torque, it's primarily designed to carry normal vertical loads. Subject a joint to stresses it's not designed to handle and it'll fail.
Now, HiTech is hinting that there is not an increasing load on the tail for greater G, it could actually be a decreasing load (less upforce). I don't see this but then, like I said before, I don't have a way to model it. From a pilot's perspective there is a measurement that looks at this called stick-force-per-G. In other words, it's how much rearward force the pilot must place on the stick in order to generate greater than 1G. That force must increase for every additional G. Say 2lbs for 2g, 4 lbs for 3g, 7 lbs for 4g, 10 lbs for 5g and 15 lbs for 6g. That's called a positive gradient and that's what you want because it means the pilot must pull harder for every additional G and it gives the pilot a positive sense of what the airplane is doing. A negative gradient where stick-force-per-G declines for greater G is a very bad thing and any fighter that exhibited would probably not be considered safe and would be assigned a PartI deficiency and rejected because it would be very easy to inadvertantly stall or overstress. Note: I'm talking about subsonic planes here, strange things can happen in the trans-sonic region including reduction in stick-force-per-G.
