I suspect most people would say “good taste” is an ability to discern what other people in your social group (or the social group you aspire to) find attractive. Since most people cannot say much about why they like something, it seems as though good taste is just the ability to identify a shared preference, nothing more.
But looked at from the perspective of artists, musicians, designers, architects, chefs and winemakers, etc. this answer is inadequate. It doesn’t explain why creative people, even when they achieve some success, strive to do better. If people find pleasure in what you do and good taste is nothing more than an ability to identify what other people in your social group enjoy, then there is little point in artists trying to get better, since the idea of “better” doesn’t refer to any standard aside from “what people like”. So it seems like there must be more to good taste than that.
But philosophers have also struggled to say more about what good taste is. David Hume, the 18th Century British philosopher, argued that good taste involves “delicacy of sentiment” by which he meant the ability to detect what makes something pleasing or not. In his famous example of the two wine critics, one argued that a wine is good but for a taste of leather he detected; the other argued that the wine is good but for a slight taste of metal. Both were proven right when the container was emptied and a key with a leather thong attached was found at the bottom.
Thus, Hume seemed to think that good taste was roughly what excellent blind tasters have—the ability, acquired through practice and comparison, to taste subtle components of a wine that most non-experts would miss and pass summary judgment on them. The same could be said of the ability to detect subtle, good-making features of a painting or piece of music. The virtue of such analytic tasting of wines is that the detection of discreet components can at least in theory be verified by science and thus aspires to a degree of objectivity. Flavor notes such as “apricot” or “vanilla” are explained by detectable chemical compounds in the wine.
There is something right about Hume’s model of taste. Someone practiced at discerning elements that ordinary tasters would miss is an indicator that someone has good taste. But I don’t think this model is quite right.
Good taste involves evaluating quality, and the quality of a wine, painting or piece of music is seldom a function of the components of the work taken individually. A wine taster can identify a whole bowl of various fruits wafting from a wine, pronounce the acidity to be bracing and the tannins fine-grained but firm, and still have said little about wine quality. Wine quality is a function of structure, balance, complexity, and intensity supplemented by even less concrete features such as deliciousness, power, elegance, gracefulness, or refreshment. None of these features can be detected by analytically breaking down a wine because they are inherently relational. No single component can account for them; it is a matter of how the components are related. Even a prominent feature like acidity is not merely a function of Ph; perceived acidity differs substantially from objective measures of acidity and is influenced by the prominence of other components such as sugar and tannin levels. None of these relational properties seem amenable to scientific analysis. I doubt that gas chromatography can identify elegance; a wine’s balance cannot be appreciated by measuring PH and sugar levels.
Identifying these aesthetic features involves a holistic judgment, not an analytic one. The wine as a whole must be evaluated. (The same holds true of evaluating a painting or piece of music) But although these holistic features, in a wine, are a product of fruit, acidity, tannic structure, etc., no list of wine components will add up to a wine being balanced, elegant or delicious. Another British philosopher, from the 20th Century, Frank Sibley, argued that this is a general feature of aesthetic judgments. There are no rules that get us from facts about the object, regardless of how subtle, to these holistic aesthetic judgments.
Hence, the problem of good taste. What do you discern when you identify elegance, grace, or deliciousness in a wine? It’s not like picking out oak flavors. It’s a judgment about how everything comes together—a set of relations that emerge from facts about the wine but are not identical to any particular collection of facts. If it is not an analytic ability, what sort of ability is it?