Subjectivity is the phenomenon of the apparent truth of a predicate depending on a perspective of evaluation, such that one person may sincerely assert a proposition p while another may sincerely assert not-p. Among the numerous analyses of the semantics of subjective predicates (Lasersohn 2005, Stojanovic 2007, Stephenson 2007, MacFarlane 2014, Barker 2002, a.o.), few consider what makes them differ from objective ones: what makes delicious allow faultless disagreement while wooden or red do not? Assumptions that subjective and objective predicates differ In their semantics (do not have truth conditions as per expressivism, have another index or argument as per relativism or contextualism) ignore the fact that the same predicate may be subjective in a context where it is loosely defined and objective in a context where it is stringently defined. E.g. the truth of good figure skater is objective to trained figure skating judges but subjective to casual TV watchers.,I provide a relatively theory-neutral analysis of what makes subjective predicates what they are. I argue that objective predicates are precisely those for which there is a reliable consensus of what evidence matters (to distinguish from a reliable consensus as to whether propositions containing them are true: we do not know whether there is life on other planets, but we know what it would take to prove it). For subjective predicates, and propositions containing them, there is no reliable and socially enforced consensus as to what evidence matters, and how much, and what does not, and for some predicates, there cannot be. Thus, speakers are allowed, in a pragmatic context, to perceive the evidence differently (to have different taste perceptions due to genetic differences in smell receptors) or to classify it differently (looking at a painting, to judge whether it is excellent or poor based on differing prior expertise in painting). If we allow differing perceptions or different categorizations to be valid, we have a subjective predicate. ,As a follow-up, I explain the selection criteria of find (NP) (Predicate) - `I find the soup disgusting/wonderful’ - which is known (Saebo 2009 a.o.) to select for subjective constructions. I argue that find actually selects for direct experience of its object, as was proposed by Stephenson (2007), and I address subsequent criticisms of that analysis and extend it to modal expressions such as `I find the Cubs winning unlikely,’ which had not been previously considered in the literature. I conclude by showing how my analysis fits into different theories (expressivism, relativism, contextualism, metalinguistic negotiation) by providing them with clearer selection criteria for not only what a subjective predicate is, but why it is so.