When we define discrimination as (act or instance of) making a nonrelevant distinction or as practise of making nonrelevant distinctions, we will be talking about something else than someone who fights racism (or ethnocentrism), but who calls it "racism and discrimination". The statement of such a person saying that 'e opposes 'racism and discrimination' is terminologically as bizarre as that of someone saying that 'e loves 'women and (also) human beings' or, for that matter, 'men and (also) human beings'. If racism denotes the belief, attitude, practise or act in which one or more nonrelevant distinctions are made between human beings on the basis of race, ethnicity or skin color, then racial discrimination is simply part of it. Perhaps some confine the meaning of racism strictly to the belief or attitude, but then the corresponding form of discrimination is racial discrimination and not discrimination in general. Conversely, what corresponds to discrimination is not racism but exclusivism. (This term, however, we will use ourselves to refer to discriminatory practises and acts as well.) By speaking of "racism and discrimination" it is suggested that discrimination is nothing else than racial discrimination (and that exclusivism is nothing else than racism), and that it would be wrong because it has something to do with race. This is about the narrowest conception of discrimination possible. The practical implicature of this (seemingly exclusive) antiracism, when recognized and mentioned as an ideal in isolation, is that people might be allowed to make nonrelevant distinctions on all other grounds than race. (This also applies to the exclusive emphasis on a right to freedom from racial discrimination.)

Altho we all denounce discrimination, and are talking about discrimination in a condemnatory sense, it is not race or any other particular factor which is at issue here but the principle of relevance itself which is violated regardless of the kind of factor concerned. Our own definition of discrimination is therefore etymologically entirely justifiable. Yet, we must admit that etymology is never a sufficient criterion in itself. (Thus, (to) live derives from caelebs which means unmarried, but no etymologist, celibate or not, can seriously maintain that 'to live is to be unmarried'.) Another fact --and a more important one-- which favors our definition is that it does seem to agree with what many ethical theorists have always understood by discrimination in general, or by racism, sexism and so on, in particular. Sexism, for instance, has been described as counting sex as relevant in contexts where it is not. (See I.5.1.2.) As regards the grounds for discrimination between human beings of different races which racists sometimes offer, it has been said that they are 'not relevant to the question of the capacity for bearing acute pain' and that they 'therefore should be disregarded'. It has been pointed out, in turn, that if this argument is valid against discrimination on the basis of race, it is valid in an analogous way for discrimination on the basis of species.

Taken literally our definition might still be considered too broad compared with the traditional usage of the word (not so much by those who speak of "racism and discrimination", but by many others). For lots of people 'discrimination' seems to be confined to irrelevant (or nonrelevant) distinctions which are disadvantageous to people (sometimes sentient beings) in a social environment. This typically moral (rather than generally normative) use of discrimination also explains why quite a few theorists tend to speak of "a discriminatory action or practise based on a morally irrelevant property" or "factor". Hence, they do not only qualify the fundament of the relevance relation (a distinction) but also its terminus (a moral goal or other moral entity) and even the effects of something that is not part of the terminus proper. Not only would making an irrelevant distinction when alone on a deserted island not count as 'discrimination', making an irrelevant distinction to someone's advantage would not either. Yet, if the happiness, well-being or interests of personal, human or sentient beings are to be regarded as ends in themselves, this requires an extra principle of happiness or another maxim of that nature (besides, possibly, the right to personhood). All these additional terminological and normative considerations do not concern the principle of relevance as such, and from the point of view of this principle they do not justify a narrower sense of discrimination than we have started out with.

Sometimes theorists do not explicitly refer to the nonrelevance of the distinction drawn, or to the 'making of distinctions' (leaving irrelevant as superfluous). Instead, they speak of a difference in treatment or favor on a basis other than individual merit. In this formulation it is implied that such a difference made is never relevant. Also when the reference is just to favor (instead of difference in favor), it carries implicitly with it that the favor itself is not justified, as it is founded upon --again-- a nonrelevant distinction. (If the distinction actually is relevant, the term favor would not be employed to imply partiality.) Finally, discrimination is sometimes also defined as making an unjustified distinction, but the only thing this definition is good for, is that it tells us that there are 'just' and 'unjust' distinctions. Just and unjust themselves are normative or evaluative concepts which do not describe any factual(-modal) state of affairs. According to the relevance principle, a distinction is unjustified because it is not relevant, and not the other way around. It is only with this principle in mind that unjustified can be substituted for nonrelevant.

All the above definitions are 'objective' in that discrimination can even occur when the group or person concerned does not notice it, or feel it that way. The normative question on the performatory level is what is discrimination? and not what do people discrimination believe to be? or when do people feel discriminated against?. The latter, doxastic notion is primarily interesting from the empirical point of view of the social sciences. Thus, one phenomenological sociologist attempted to define discrimination by means of a discrepancy between the (doxastic) relevances prevailing in different groups. This discrepancy was said to be one between an 'objective' and a 'subjective' definition of a concrete group situation. The 'subjective definition' is the one of the afflicted member of the group and the 'objective definition' the one of those 'imposing a typification' on this individual and the group 'e belongs to. At the same time the sociologist in question wrote that others create the 'group' and invest it 'with a fictitious scheme of relevances which can be manipulated at will'. Somehow, 'e had to intuitively admit that discrimination involves more than a mere difference in relevancy judgments, and that some people are indeed discriminated against by others 'imposing fictitious schemes' on them.