Does it say anything meaningful about the boost program’s taste that so far presenting a boosted article in a feed is 50% more valuable than the best available non-boosted algorithmic recommendation? The success of boost seems obvious to us just with a eyeball test. But the numbers also support it.