Unsurprisingly, this is not at all what the article speaks to; it may well be true that, epistemically, some of the things you're worried about are going on. But just like this is not an article about whether trace BPA in the food supply has hormone regulation effects, or whether replacing your current car with an EV is carbon positive or negative, it's also not an article about what you just said it was about.

Rather, it introduces two basic concepts in how intelligence organizations qualify their intelligence gathering: (1) estimative language: a concept that maps numerical probabilities into specific language (e.g. in the UK, "highly likely" means 80-90%); (2) analytic confidence: a concept that speaks to the verifiability of an individual source or conclusion, typically based on the mode of intelligence gathering.

Having done this, it then suggests that most individual pieces of intelligence have low analytic confidence, but that depending on the opportunity cost (loss function) of (in)action, this may be sufficient to proceed; likewise agglomerations of low confidence intelligence may form a high confidence conclusion (as in an ensemble learner).

By way of examples, it discusses the following examples: (1) whether Russia was using T-62 tanks -> whether better sights would help night vision; (2) the DOE COVID WSJ reporting; (3) whether Iraq had WMD; (4) whether Germany sabotaged Nordstream; (5) whether Russia would invade Ukraine in 2022; (6) the NIE Havana Syndrome reporting. I assume the thing that made you write the fan fiction you wrote was #2.

Honestly, it's a 13 paragraph article and a link around the paywall was posted hours before you posted your thing, so maybe just read the article?