You are right in that of course tenuously-related results are useless, but if I recall correctly a 0.5 rating should mean the results would be buried under dozens of pages of more relevant results. That’s why the sliding scale exists - if the rating was merely binary, 0.5 would be a 0. My point was that the test considered 0.5 to be just as wrong as a perfect 5 rating.
I’ve never used DuckDuckGo, but a colleague recommends it. Might have to give it a try, but since most of my work-related searches are highly specialised I suspect Google is the only realistic option for me.
