In order to orient the inquiry of the Metaphysics, Aristotle begins with the traditional opinions about the causes of things as a whole, and draws out their inherent difficulties. This, in part, follows from his general dialectical method: he does not begin with first principles and deduce from them a universal philosophy, but begins with the traditional beliefs that he has inherited. In one sense, this method takes into account the "thrownness" of the philosopher; that is, the fact that the philosopher is always historically situated and does not have immediate access to objective truths from which he can begin his philosophy. To start in any other way covertly imports one's inescapable intellectual inheritance into the inquiry, and allows this inheritance to be acknowledged and addressed up front. This gives Aristotle's dialectic an advantage over any deductive metaphysics, in that his starting points need not be incontrovertible.
Aristotle examines the philosophy of those that went before him by drawing out their inherent tensions and contradictions. From these tensions, he establishes the problematic from which the Metaphysics will work. Aristotle must get some idea of what sort of thing metaphysics reveals; that is, of the nature of the metaphysical question. The immediate difficulty lies in the fact that one cannot know what metaphysics asks about without knowing the object of the metaphysical question. One must know the end before the beginning.
As an initial matter, the metaphysical question concerns the source of things. But is this source one or many? If there are irreducibly many sources of things (e.g., the four elements, or the four causes), then there is no metaphysical knowledge, but different kinds of knowledge for each kind of source. The sources can be irreducibly many either in kind or in number. If the sources are irreducibly many in kind, then thinghood is impossible, for thinghood implies a kind of unity which is grasped in thought when one grasps its cause. However, if the causes of the thing are multiple in kind, then no unity exists by which one might grasp the thing. Yet things present themselves to us in a kind of unity which we immediately and pre-reflectively grasp without trouble. Positing a multiplicity of sources different in kind is simply insufficient to explain everyday experience.
On the other hand, if the source of things are irreducibly many and differ in number, but not in kind (Aristotle calls these elements), then there will be nothing other than the elements. The sources would differ by virtue of their particularity alone (being this atom and not that one, for example), and if there were no causes higher than these elements, nothing could exist other than these elements. Syllabic sounds, for example, in order to come together and form words, have to take on the reality of a whole above and beyond the parts. This whole necessarily takes the form of an unified cause incompatible with an ontology that posits irreducibly many sources.
A multiplicity of causes precludes the unity that things possess, and fails as an explanation of ordinary experience. If, however, the cause of things is one (this would be called "oneness" or "being", and applies univocally to all things, the Parmenidean problem arises. To understand being in this way would be to understand being as a universal genus or category. "But", Aristotle says, "it is not possible for either oneness or being to be a single genus of things," for then there would be only one Being. Being, understood as a universal category, would rule out individual beings as illusions, because a species is differentiated within a genus by differentia outside the genus. "[I]t is not possible either to predicate the species within a genus of their own differentia, or to predicate the genus without its species of the differentia."
In the footnote to his translation, Joe Sachs explains it this way:
If we define doves as wild pigeons, the species is doves, the genus pigeons, and the differentia is being wild. If this is a sound definition, it cannot be true that (all) wild things are doves, or, the more important point here, that (all) wild things are pigeons. The reason is that all characteristics by which a genus is differentiated into the species are outside the genus.The characteristics that differentiate genus into species must be outside the genus, for if the differentia were within the genus, then those characteristics would belong solely to the species or to the genus as a whole. In the first case, if only pigeons were wild things, then the terms "wild things" would have no meaning or extension other than "pigeon", and differentiating doves by the characteristic of wildness is simply to differentiate doves by the character of being doves. If the differentia existed only within the species, the only way the species could be differentiated from the genus would be by tautology (essentially saying a dove is different from other pigeons because it is a dove). In the second case, no differentia would separate doves from other pigeons, collapsing the species into the genus.
The basic problem: being, it seems, cannot be one because it abolishes all difference, and it cannot be manifold because it abolishes all unity. It is from this problematic, the apparent tension in being between the one and the many, that Aristotle's metaphysics takes its direction.