Propositions might fall into these categories:

• that something is (∃, existence)
• what something is (quiddity, essence, substance)

I can confidently claim that the Sun exists. Whatever exactly it may be, it can be shown and felt. Answering what the Sun is, becomes more involving. An exhaustive answer may not be attainable, just some (sufficient) characteristica. In this case the former (that) is implicit in the latter (what).

When something is (only) defined, the roles can be reversed. In that case it becomes questionable that something exists, as per the definition of what it supposedly is. Other than the definitions themselves, things cannot be defined into existence, as it were. The ancient astronaut hypothesis will have it that extraterrestrials visited Earth thousands of years ago, and meddled in human affairs. So, these beings are defined sufficiently for the hypothesis — what they supposedly were — but that they were, is more questionable.

• showing that something is → now let’s find out what it is (and/or what it is not)
• defining what something is → now let’s found out that it is (or not)

Presumably anything existing is something in particular (at the very least self-identical), and anything that’s something in particular would have to exist, so the two kinds of propositions are tightly related. (And, there are both ontological and epistemological elements involved.)

Yet, that and what are not identical. We often express what by predication (e.g. “the apple is round”). Following contemporary mathematics/logic, Kant (“existence is not a predicate”), Quine (“to exist is to be the value of a bound variable”), and others, existential quantification is properly formalized as

p = ∃x ∈ S [ φx ]

where p is a proposition (that may be true or false), φ is a predicate (without unrestricted comprehension), x is a (bound) variable, and S is a set. Exemplified in ordinary language, the proposition states that an apple exists in the world, that’s round. The ∃ and φ symbols are not interchangeable (besides, if they were, then apparently we could have something already, except it does not exist, which seems like nonsense). In general, anything that exists is self-identical, and existence is that which can have no complement (∁ existence = ∅). Long story short, that ≠ what.

Dictionaries and encyclopedias are inherently circular. Look up entries for “elephant” and “run”, take each of the words used, and look them up as well, etc. Eventually you’ll visit entries already visited. There are no running elephants in dictionaries and encyclopedias, though they are intended to give (sufficient) accounts of what the entries are (including fictions). We invariably strand on showing running elephants, or perhaps evidence of a stampede, or whatever, if we’re to conclude that they are. (Subsequently we could turn to abstracts, categories, self-reference, words themselves, Platonism and such, if we must, but that’ll have to wait for another day.)

Certainty (error-free knowledge), to the extent there is any, seems resigned to the former category (that):

• something exists (whatever it may be, since the contrary is self-contradictory)
• I’m not omniscient (since otherwise I’d know that I were)