We are naturally far better at identifying what's plausible instead of sorting out what is likely, and we're better at giving kudos and recognition for the former than the latter even though it's easier. That makes some people hyperdevelop "theorizing" without working on "validation."
We are dangerously good at making up stories to explain data. Talk to anyone about a spurious correlation[0] and they can provide you a just-so story that explains it as a real correlation in under a minute.
Provide anyone with two competing theories that each account for some but not all of the available data, then ask them to rank the two based on likelihood. You'll make people visibly upset. (Our ability to invent new theories to fit any circumstance might partly explain why, for a while, explorers forgot how to cure scurvy.[1])
And it's easier to "look smart" the first way. No one can immediately determine whether you're good at the second unless you make a long series of predictions evaluated over time while constantly gathering more data. Everyone can see if you've connected a few dots in a clever and unexpected way. Talking heads are regarded as experts for going on shows and making vague predictions based on a few recent events, but are never called back to scrutinize their predictive track record.[2]
Conspiracy theorists are generally just arguing something is plausible, and ignoring the messy step about what's more likely, with the implication you should just leap from plausible immediately to unwavering conviction.
Some of this is just, well, one of these things is hard and the other is easy. But I'm confident both can be learned to some extent, based on personal experience of getting a better sense for these over time.
So to the extent this is "fixable" at all,[3] maybe greater education about meta-cognition (esp. cognitive biases), the history of analytic philosophy, a humility about what one knows, and/or Bayesian reasoning earlier in education might help.
Sidenote, this itself is arguably just another just-so theory too. Maybe I'm wrong about it all. :)
[0] e.g., http://tylervigen.com/spurious-correlations
[1] https://idlewords.com/2010/03/scott_and_scurvy.htm
[2] This is Phillip Tetlock's whole area of research.
[3] To be fair, maybe we've already "fixed" this pretty well, we tend to attribute far fewer things to witchcraft than a few hundred years ago, maybe normal everyday education and expansion of literacy is enough and there will always be a handful of outliers and trolls. See also: https://slatestarcodex.com/2013/04/12/noisy-poll-results-and...