Newly declassified info discusses an alarming discovery that led the CIA to credit claims by Afghan detainees that Russia offered bounties to spur attacks on US troops during peace talks -- and gaps in the available evidence that worried other analysts. https://www.nytimes.com/2021/05/07/us/politics/russian-bounties-nsc.html

The intelligence community assessed with "high confidence" that leaders of a Taliban-linked criminal network, where the bounty claims arose, had been working with and for a notorious Russian intelligence assassination squad. w/ @EricSchmittNYT & @mschwirtz

That the leaders of the detainees turned out to have been been interacting with Unit 29155 of the GRU - the group accused by various European governments of orchestrating a coup attempts, sabotage operations, and poisonings - was seen a strong circumstantial evidence.

It's not new that CIA+NCTC have "moderate" confidence in the Russian bounty assessment, while NSA+DIA are more worried about gaps so have "low." But the consensus about the most important circumstantial evidence - links between Unit 29155 and the Afghan criminal network - is new.

It's not new that the available evidence can't prove that any particular attack resulted from a bounty operation. But the other factual gap that worried analysts - they have no info on someone in the Kremlin ordering Unit 29155 to do this - is new.

The government apparently did not declassify everything. The statement does not include other things we have reported -- the names of two Afghan network leaders who interacted with Russians & fled to Russia, $ transfers from GRU to the network, the lack of surveillance intercept.

Confidence levels in an intelligence assessment accounts for complexity when trying to make sense of imperfect information. A "high confidence" assessment is not the same thing as a fact. And an analyst who assesses something with "low" confidence still believes it.

The world of political messaging is comparatively simplistic, which leads to complexity-flattening distortion as commentators talk about intelligence assessments as if they were simply true or false, depending on which is expedient for the agenda they are pursuing.