Towards a more useful distinction
Yeah that concept creep around "conspiracy theorists" gets applied as a thought terminating cliche, or just a way to indicate a person shouldn't be listened to, without the need for an all encompassing conspiracy to actually take place.
Like, yes, it is a consipracy theory to claim the head of the NIH organized a campaign to discredit the lab leak theory and it's proponents, (not to mention the likely successful conspiracy by the people actually operating the lab to keep the evidence out of reach). But that conspiracy doesn't require the silence and pre-planned deception of hundreds of people, it is not "too big to fail". Reminds me very much of your original benign smoking example.
You see it all over the place now, "Racism" basically used to mean "hatred for another race" and all the emotion and disgust you would have towards someone who was that way might get transferred to someone who was against Afirmative Action. Similar for someone who is labeled a "predator" could be anywhere from a forward sexual person, to a creep, to an unwanted touch, to a full on rapist.
One dynamic I've observed, and use as a fuzzy "indicator" as you say, is that conspiracy theories tend to have a strong propensity to backfill any gaps of seeming contradictions, by just deepening the conspiracy. (eg: "You ask why none of X noticed? That's because all X were in on it!").
That is, rather than looking into every allegation, see how the person handles some question for which they don't have a stock answer to regurgitate. If they don't have "yeah, that part doesn't fit yet, it needs more investigation" in their toolkit, and instead quickly grab the 'expand the conspiracy tool', that's a clue. Or likewise, making a new but obviously bogus assertion, which even they might not want to be held to, in order to fend off disconfirmation.
Of course, this is more about the mindset of people who follow conspiracy theories, than about the theories themselves, but it gives one some measure of how seriously to take the allegations (eg: how much effort to make to check an assertion).
Haven’t yet finished the piece as I’m working atm, will do soon. Can’t resist the temptation though:
“Shit, Yassine figured out the coffee conspiracy.”
“Yassine, have you heard of kale? It tastes great! 😀”
"The classic counterexample could be General John DeWitt citing the absence of sabotage by Japanese-Americans during WWII as evidence of future sabotage plans. There is indeed a trend of conspiracy theorists digging into their belief in belief, and dismissing contrary evidence as either fabricated, or (worse) evidence of the conspiracy itself."
Most of these are variants on assuming as true that which one seeks to prove. The classic example is that if there is no actual evidence that Donald Trump is in fact Mickey Mouse, that just means we need to look harder. In the meantime, we can safely assume that Trump has big black ears and a tail.
"The impossibly stringent standards appear to be in response to the fact that the offense carries the death penalty, and shows it’s possible to raise the bar so high that falsifiability is intentionally rendered out of reach."
There is a famous monograph comparing the plea bargaining system afoot in current US criminal law with confession through torture in medieval Europe. To summarize - medieval rules of evidence required at least two unimpeachable direct eyewitness testimonies - OR - for the accused to confess to the crime. Circumstantial evidence was not permitted.
This made convictions nearly impossible without a confession. If X and Y are alone in a room, and X is seen leaving the room with a bloody knife and Y found warm and dead, under medieval rules of evidence, that would not be sufficient to convict X of murder.
Hence the rack.
Analogize to the high standards for conviction in the current American criminal justice system, the expense of trial, and the draconian sentences for those found guilty, all of which incentivize a confession.