I think your paraphrase makes sense if Clarity was accidentally using the term effect instead of affect (in it’s “feeling or emotion” definition). But that doesn’t really fit with the last use of effect, so I would translate Clarity’s use of effect as “property” or “trait”. My paraphrase would be:
Suppose that our understanding of the properties of objects is entirely binary, they either have a property or not. Assume objects are categorized into groups based on their observed properties (this isn’t stated in the original, adding that for clarity). In general, should we assume that if we see a new object which shares some but not all properties with a group, that it shares other properties as yet unobserved?
I would say that if this is the intended meaning, you should assume that it shares other properties, even though it differs, strictly speaking. Mostly this is because we are coding all properties in strictly binary terms rather than as probabilities. If you had extensively studied dangerous trees, then someone showed you grass, and then they asked you binary questions about its properties, you’ll be correct in assuming the answer to the questions was identical for the tree and the grass. You’d get some of them wrong, but the vast majority you’d get right. Trees and grass are much more closely related than grass and petroleum, or grass and love, or grass and President Obama.
We do this in the real world. Most cleaning supplies are toxic, but none of them are carbonated. If someone handed you a bottle of some novel cleaning supply, and you saw it was carbonated and milky white, you’d be right to assume it was toxic even though it clearly has some properties different from all the other cleaning supplies you’ve seen. Of course it would be better to have a probability distribution about whether it’s toxic.
I think your paraphrase makes sense if Clarity was accidentally using the term effect instead of affect (in it’s “feeling or emotion” definition). But that doesn’t really fit with the last use of effect, so I would translate Clarity’s use of effect as “property” or “trait”. My paraphrase would be:
I would say that if this is the intended meaning, you should assume that it shares other properties, even though it differs, strictly speaking. Mostly this is because we are coding all properties in strictly binary terms rather than as probabilities. If you had extensively studied dangerous trees, then someone showed you grass, and then they asked you binary questions about its properties, you’ll be correct in assuming the answer to the questions was identical for the tree and the grass. You’d get some of them wrong, but the vast majority you’d get right. Trees and grass are much more closely related than grass and petroleum, or grass and love, or grass and President Obama.
We do this in the real world. Most cleaning supplies are toxic, but none of them are carbonated. If someone handed you a bottle of some novel cleaning supply, and you saw it was carbonated and milky white, you’d be right to assume it was toxic even though it clearly has some properties different from all the other cleaning supplies you’ve seen. Of course it would be better to have a probability distribution about whether it’s toxic.