I like the piece from Givewell, but they're doing things which are much easier to measure.

My impression is that SIAI is at a stage where most of what can be measured is inputs (money raised, hours worked) rather than outputs, and it's hard to tell whether an output (a new piece of theory, for example) is actually getting closer to one's goals.

I'm not saying that SIAI's work is unimportant, but evaluating it may be more a matter of logic than measurement.