We know where it started, the impression of Celine Cawley as a bossy driven woman who had her husband completely under her thumb.
The impression started with the questions asked by garda detectives directly after her death. Questions that suggested Eamonn Lillis was a poor, underpaid, put-upon weakling who, when she said "Jump!" had no choice but to ask "How high?"
Read out in court after many months had intervened, the questions sounded overly sympathetic to the man the gardai must have known had killed his wife.
They started to paint a bad picture of the dead woman -- a picture which will never be fully erased.
And yet the Cawley family went out of their way, after sentence was passed, to praise the police officers involved.
That's because the Cawley family are smart, and understood the process the interrogators had to go through, like it or hate it.
When I train detectives in forensic questioning, the first element of the training is the establishment of trust.
A detective, sitting down with a suspect for the first time, may instinctively know the suspect's confected story is full of holes -- as was Eamonn Lillis's improbable saga of a balaclava-wearing thug wandering around Howth at nine o'clock in the morning.
But attacking the confected story is not the way to go because it may back the suspect into desperate, irrational defence of the story or -- worse still -- close them down into obdurate silence.
The objective is to open them up, and that requires relationship-building. The interrogator has to link arms, metaphorically, with the suspect. Even in cases of child murder, they may need to suggest to the man they're questioning that he's been a wonderful partner and father, that he's given his life to his partner and her child, that he's been working all the hours that God sends and that the child constantly crying at night would have tried the patience of a saint.
They may feel queasy inside while they're doing it, but if they've been well trained, they're physically at ease, their gestures are unthreatening and -- at this stage of the interrogation -- they know they don't have to prove anything.
All they have to do is relax the guy and establish trust. Once they've done that, he gets into "Relinquishing Mode". He decides he can safely abandon some element of his story. When that happens, the rest of the unravelling is a matter of time.
Training detectives (and HR professionals and investigative broadcasters) is all about helping them to spot and unravel lies. Lies are the bullet-proof vest every serious criminal wears.
Because of TV programmes, untrained people are convinced, these days, that they can spot a liar a mile off by their behaviour.
The suspect ran their finger round their collar? Ah. That shows stress. They're definitely lying. Or they put their hand over their mouth? Giveaway. Liar, liar...
Except it doesn't work like that. Most people are remarkably poor at identifying when someone else is lying.
They're not bad at working out when someone being interrogated is stressed, but being stressed doesn't mean you're lying.
Innocent people get at least as stressed as guilty people do when interrogated.
Research in the United States has established that even experienced detectives are only marginally better at spotting lie-behaviour than members of the public.
That's because untrained observers watch for signs in a vacuum: a dry mouth, or a stammer, or an apologetic way of answering a question. They then decide those signs indicate deception.
Big leap. Big mistake.
In training, we establish a simple but essential basic understanding: Until you know how the suspect acts and talks in normal situations, you have no baseline against which to judge their behaviour under harsh interrogation.
One of my colleagues constantly puts his hand over his mouth when he talks. It doesn't mean he's a liar. In fact, he's more truthful than most people.
But -- perhaps because he had bad teeth as a teenager or for some other reason -- he has the mouth-covering habit.
An untrained detective who goes "Eureka!" when he spots the hand going over the lips is going to build a criminal case on sand.
The final area in training a questioner depends on them being great listeners, as well as great note-takers.
Great listeners spot discrepancies. They'll notice when a detail in the description of an event doesn't match another detail. In the beginning, they find it difficult to conceal their excitement at their own discovery.
But they learn to let the contradiction pass, apparently unnoticed, to find ways to get the suspect to repeat it several times. Then -- and only then -- will they, in a puzzled sympathetic way, ask the suspect to make sense of what clearly is a tapestry of lies.