You're correct for ordinary sailboats. But once you admit you can go downwind faster than the wind at any angle, you're almost there. In principle it doesn't cost any energy to jibe (it's actually called jibing, not tacking, when you turn with the wind rather than into it). So you can jibe back and forth and go as straight downwind as you like, with an average speed much faster than the wind. In principle you could attach two sail or ice boats together, each jibing back and forth in a crossing pattern, so the whole contraption moves straight downwind.
My experience doesn't support this conclusion. My apologies for skimming the thread but as a sailor, what I can offer here is that a boat can often travel faster on a reach (downwind version of a tack - a vector) than a boat on a dead run (straight down wind) but neither is actually travelling faster downwind than absolute windspeed. That would entail extracting more energy from the wind than exists.
You can travel faster than windspeed on a vector. But you can't exceed direct downwind windspeed. Not without another source of energy, that is.
What does happen is that reaching boats sometimes get downwind faster because their sails are working more efficiently. Lamilar flow (a reaching sail) uses energy better than turbulent flow (a running sail)
I think what you are seeing with the iceboats is stored momentum. They have little friction in good conditions, and so when jibing they exceed the windspeed downwind for a short while, but averaged over time they won't exceed it in an absolute sense. and when reaching, their sails are working optimally, like wings, keeping pressure and flow close to perfect. Running downwind spills air in turbulent vortices and wastes energy.
Meaning if wind is due South at an average of 20 mph for one hour, a boat won't go more than 20 miles due South in one hour (unless energy is stored in some way). It may travel 60 miles on other compass points, but not due South.
The boat reaching and jibing may be moving faster through the water but must cover a greater distance than the a boat running downwind, and actually may travel closer to the directional windspeed because sails are more efficient when operating in laminar flow (as an airfoil) than when running straight downwind, which creates a great deal of turbulent flow and spillage around the sail.
Visualize a parachute vs., a airfoil wing and you get the idea.
But the performance difference between reaching downwind vs. running can be subtle. In a race like the Transpac, where the predominance of conditions favor running, boats are designed with huge sail area and flat hulls, basically surfboards with giant spinnakers (the sailing equivalent of the parachute) to plane as close to windspeed as possible. But it is never exceeded.
To add to the complexity, a lot of what goes on in sailing strategies invloves trade offs of hull design and rigging. Without going into details, but proving the subtleties of the described dynamics, one can often reach on downwind vectors, crossing back and forth close to a boat running straight downwind, and see advantage wax and wane as hull speeds are exceeded, sails are changed, windspeed changes, etc. The point is the fine line of all this tuning suggest that the crafts are using energy well, and often win or lose based on very small variables.
