Your question naturally breaks down into two parts:
- How valid is the detection of water on the moon's surface?
- If present, how has it lasted so long?
I'll start with the second. In *equilibrium, water in a vacuum will evaporate away as a gas. We call this sublimation if it occurs from a solid to the gas (as occurs in freezer burn inside your refrigerator), and evaporation if it takes place from the liquid.
But a solid sitting in a vacuum is not in equilibrium. If it were, it would already be gone. So a process of sublimation is ongoing. The question is how long it takes. This is directly related to temperature.
The insides of dark craters in the lunar polar regions are exceedingly cold, -250F (-160C). At these cryogenic temperatures, sublimation proceeds at an exceedingly slow pace. A good rule of thumb in gauging the speed of chemical processes versus temperature, dating to the 19th century chemist Arrhenius, is that things take place about 3 times slower for every 10C drop in temperature. So one could see that at a quick approximation, sublimation will occur 50 million times slower than at freezer temperatures (3 to the 16th power). This is also why comets, which are also leftover pieces of ice from the early days of the solar system, are also still around.
But your question #1 is still with us. Have we really detected water? The evidence for the detection reported by the Clementine satellite in its flyby of the lunar polar region several years ago was based on a characteristic radar signature, which, however, is not definitive, and which has been criticized widely in the scientific community. This is a major reason why the recent Lunar Prospector satellite was designed and launched. But its spectrometer can only detect *hydrogen*, not water. Hydrogen is present in a number of minerals as well as water, and its detection, while supportive of the presence of water, is not definitive. In addition, the detection is only of the very crust of the surface, and says nothing about the depth to which a permafrost may be present. In addition, we know only a little about the past history of the lunar motion, billions of years in the past. A slight wobble in the still-ongoing oscillation of the moon's face (called the lunar libration) could easily have exposed these craters to the sun's rays at some point in the past, rapidly evaporating any water present.
So many in the scientific community, while excited about one possible interpretation of this data, remain naturally skeptical until we can be more sure. It is unfortunate that NASA exerted their well-oiled PR machine to overblow this find. This is what happens when billions of dollars are at stake.