That's not what the stat says. It says only 25% of findings selected specifically for sounding important replicated for that particular company, not that 25% of

That's not what the stat says. It says only 25% of findings selected specifically for sounding important replicated for that particular company, not that 25% of output overall does. And replicable doesn't mean correct or useful, it's just the first step. There are lots of ways for a study to replicate yet be wrong. 25% is crazy low, but 11% is even lower. If someone told you things and 90% of the time they were wrong, you'd stop listening. Solar panels having low effectiveness would actually be a good argument against them, because although the solar industry needs solar panels the correct comparison is with the energy industry overall, which doesn't.