Sorry, I know I'm a bit late to the party on this, but commenting, because it's an interesting subject.
If we were to do an apples to apples comparison of the frequency bands themselves (27MHz vs 40 MHz vs 2.4GHz), using the same equipment, and technology sets, 2.4GHz would lose hands down, and 27MHz would do slightly better than 40MHz. To make 2.4GHz useable, is entirely dependent on the technologies used.
The first problem with 2.4G is penetration. Apart from penetration (or lack of), through water, it has much less penetration through solid objects. Obstacles, including obstacles on the water, parts of the boat etc, will attenuate the 2.4G signal much more than 27MHz or 40MHz. This is one of the reasons for dual antennae. Personally, I would never use any 2.4GHz system without dual antennae, but that's not to say that any dual antenna system is necessarily good.
Secondly, multipath effects, as mentioned by Ian. This exists at all radio frequencies, but presents a specific problem here. Imagine that the radio signal has a sine format (hopefully most people remember what a sine curve looks like), with peaks and troughs. A reflected signal is delayed, compared to the main signal, which means that the peak of the reflected signal will arrive later than the peak of the direct signal. If this delay is such that the peak of the direct signal, arrives at the same time as the trough of the reflected signal, you have cancellation. If the peak and trough are exactly aligned, and of the same strength, you have full signal cancellation. On the other hand, you could receive the previous peak of the reflected signal, at the same time as the peak of the directly received signal, they reinforce each other in terms of strength, but the reflected signal is slightly delayed (going back to radio / TV analogies, in the world of analogue TV, the differently timed reflected signals, were the main cause of the often seen "ghosting" effect on the picture, but this is in the world of longer distance transmission, where the reflection could be travelling much further than the main signal. The delay in itself is not likely to be problematic in our environment, for several reasons). Between these extremes, are all the different potential degrees of signal alignment (direct vs reflected again). To put some of the above into context, at 40MHz, the wavelength is over 7M, which means that the distance between peak and trough is about 3.5M. At 2.4G, the wavelength is about 13cm, so the distance between peak and trough is only 6.5cm. This gives much more potential for problems (the relationship between direct and reflected signals will constantly change significantly as the model moves).
The second antenna can help a lot, both with the blocked signal scenario, and issues with signal cancellation from the reflected signal. Because of the nature of the signal, if one antenna is blocked, it's very possible that another, not too far away, will receive a useable signal. The same goes for cancellation. If the signal to one antenna is cancelled by a reflection, the signal to the other one may well be good. If the antennae are very close together, and in the same plane, these advantages are negated. In an ideal world, if we are trying to pick up a direct signal (rather than a reflected signal), the Tx and Rx antenna should be in the same plane, which is usually vertical. You can have both antennae vertical, but if you do, don't run them up the same tube, or have them right slap bang next to each other. Futaba show a setup where each antenna is 45 degrees to the vertical, in opposition, so that the antennae are at 90 degrees to each other. I believe this is intended primarily for aeroplanes, so that one is roughly vertical when climbing, and the other is roughly vertical when descending. In boats, some have one antenna vertical, and one horizontal. The horizontal antenna will never receive the best direct signal from a vertical TX antenna, and horizontal antennae are also somewhat directional (if a horizontal antenna is oriented along the boat, it won't work very well when it's coming towards, or going away from you, only going across), but reflections of radio signals also tend to scatter their polarisation, so one vertical and one horizontal often works ok in practise (also bear in mind that we are nowhere near to the distance limitations of the equipment, it is only the physical environment which we are trying to compensate for.
A by-product of frequency hopping, is that when the frequency hops, the wavelength also changes slightly, which shifts the cancellation modes / nodes a bit, so if a signal is fully cancelled before the frequency hops, it won't quite be afterwards. In practise, our models are moving quite fast, and movement is constantly shifting the relationships of direct and reflected signals.
I personally like the Futaba FASST equipment too, and have also found the FrSky FASST compatibles to be very good. I initially tried these, because I'd previously used the native FrSky system - after reading up on it, and used the FrSky hack module, to convert an old Futaba 35MHz Skysport set I had lying around, which worked very well.
Ian S