If you were a government and you wanted to hide your true capabilities from your enemies, would you want them to underestimate those capabilities, or to overestimate them?
Most people will probably think that it would be best to lead your enemies to underestimate your capabilities, since you would then have more than what they expected.
But let's consider what situation would result if instead you led your enemies to overestimate your capabilities. For instance, if you put out 'leaked' reports that you had made technological advances in areas of science that were not already well developed or known about, it would lead your enemies to take steps to begin their own research and development in that area just to keep up, thereby wasting their efforts. At the same time, they would have a false impression of what to expect in the case of a potential attack, so that they would also waste efforts in taking steps to defend against it.
By leading your enemies to overestimate your capabilities, a definite advantage is created, where you will know exactly what you really have as well as what your enemies think you have, while your enemies will never be certain of what you might really have laying at the ready, and must assume that what they've been led to believe is true, even if they never see any evidence of it beyond hints and rumors.
On the other hand, if you were to lead your enemies to underestimate your capabilities, the moment you reveal your true capabilities by actually using them, you no longer have that advantage of uncertainty, and you must now worry about your enemies acquiring those same capabilities. You can keep your secret capabilities out of use so that your enemies don't ever discover them, but then there's no point in having them.
So, leading your enemies to overestimate your capabilities is far more advantageous, and this is done through the release of disinformation.
There is one other possible situation that should be considered here. If you have or are developing capabilities that would give you the advantage and thereby lead your enemies to underestimating you, and you want to lead them to overestimate you instead, you can create decoys that will lead them to think the capabilities are something other than what they really are. This situation can be dealt with in a manner that is explained further on in this article, where cover stories are discussed. In this case, the cover stories (are parts of them) are intentionally 'leaked', in order to lead your enemies into false perceptions so that they will take steps that will be wasted.
Let's consider another aspect of all this. What sort of disinformation would be better - that which sounds plausible, or that which sounds implausible?
Most people might think that the more implausible a piece of information is, the less likely it would be considered. However, in the world of secrecy and espionage, nothing is necessarily as it appears, and truths can be hidden in implausible stories as much as they might be found in plausible ones. But, just as we saw in the case of overestimations versus underestimations, creating a level of uncertainty would be in your favor, since an implausible story, although it might signify to the enemy that it's disinformation and will therefore contain hidden truths, investigating it will nevertheless waste a great deal of their time and efforts as they attempt to uncover those hidden truths.
Some real-life examples of this third situation include: using the UFO/alien abduction scenario to cover up mind-control research and technologies; using the events surrounding the purported Philadelphia Experiment to cover up the research into antigravity and stealth technologies*; and using the current exaggerated misconceptions about mind-control technologies to cover up more plausible ones.
Disinformation is usually created preemptively, meaning that a cover story is formulated before an idea for research and development into a prospective technological advancement even gets put down on paper. This occurs at the point when only a minimal number of people know anything about the idea, and it's at this point that cover stories are created that will satisfy the needs of anyone who will be brought in to handle the various aspects of the research and development.
Compartmentalization of information will also be incorporated at the very outset, and different cover stories can be given to the different people working separately on the different components of the overall project, and none of them will ever know that what they are working on might be related to anyone else's work, nor will any of them know what the overall project goal is. As each stage of the R&D is fulfilled, the next stages go through this same procedure, with cover stories being formulated preemptively and new people being brought in to work separately on each of the various components of that stage. Should any information leaks occur, only the cover stories will ever be revealed, since the real goals will remain tightly held by the few people who are in charge of the overall project.
Now, let's consider how this works with respect to someone who is brought in to work on some aspect of a classified project. They can be given a cover story and know that it's false, and this is fine as long as it doesn't hinder their ability to do their work. They're just there to do their job, and they know that the cover story is for their protection as much as it is to protect the security of the project. They accept the cover story and don't ask questions or speculate too much on what the real truth might be, because they know that they could be targeted by the enemy to gain information. This might be accomplished through a variety of social engineering techniques that could be unsuspectingly used on them in an attempt to draw out useful pieces of information, or it might take the form of more drastic methods that could involve their abduction and torture. So, for their own safety as much as for the safety of the project, they work under the pretense of a cover story and don't even speculate on what the truth might be.
Before they're even brought into the project, these people will have undergone extensive security checks, and as soon as they're brought in but before they're told anything about the project, they will be made to sign a security oath that carries extreme penalties if they breach it. Also, throughout the project and possibly for many years after (depending on what they know), they will be closely monitored to assure that the security is maintained.
Because the people working on a classified project can know that the cover story they're given is false, it's implausibility doesn't really matter, as long as it serves to explain the purpose of their work.
In the case where purposeful leaks are desired to throw off your enemies, certain people who are brought into a project can be selected because of their lack of ability to maintain certain levels of security, in which case they will only be given carefully selected information, cover stories, or weak security measures that will lead to possible 'leaks' of disinformation that appears to be valid. In these situations, these people will usually be selected because they're susceptible to believing that the cover story they're given is the truth.
_ _ _
* See Nick Cook's 'The Hunt for Zero Point: Inside the Classified World of Antigravity Technology' for an in depth look at how this implausible disinformation story was used to cover up more plausible R&D, which has effectively led to a gross overestimation (and misconception) of the US government's technological capabilities that continues today.