Paul Hartzer (and many others) have written--
I don't think it's necessary to have an understanding of linguistic
terminology to create a decent language. I do think it's important to have
an understanding of linguistic topics, by whatever terms one knows these, to
do so. That is, if someone understands that sounds make up words, words have
meanings which can change in context, words have parts of speech that
usually appear in a certain order, and so on, they can make up a perfectly
fine language without ever appealing to "phoneme" or "morphology" or "syntax."

That is true, but.....  Early on, I was fascinated by what little I learned from an encyclopedia article on Sanskrit. Consequently, both of my teen-age conlangs had "bh" and "th" (which I thought would be [t_T], which I still like) and one even had "mh". But I don't think either one had a full range of aspirated stops-- no dh, no jh, no gh, no vl.asp. series at all,  etc. 

I was unaware of the fact that the sounds (phonemes) of a language could be arranged, and described, systematically.  I was used to lists, as in my sister's Spanish grammar, where one ecountered things like "The letter _g_ is pronounced hard before a, o, u, and something like English h before e and i." (Presumably the teacher was supposed to clear up that "something like....") Needless to say, I was totally unaware of the IPA.

The second language, a clone of Latin, and of which I remember a few phrases, shows the following "letters"--  i, th, a, e, bh, l (and cluster bhl-), sh, u, e-macron (not sure how that differed from plain e, but it was always so written); k, m, r, z, n, f, g, d (and cluster dr). It had diphthongs ai and e:i-- that was the genitive ending, borrowed from Latin 5th declension, which I thought interesting.   I deliberately list these in random order, because that's how I thought them up, I'm sure.  There must have been others that I forget-- but the point is that there was no system. Most peculiar was the writing system-- some of the symbols were single C and V, others CV, perhaps some were CVC or VC, I don't remember; but one represented the 3-syl dative ending -ainigi. 

I think if I had presented this language on a 1950s version of Conlang-L, the verdict would have been: "interesting but unsystematic and lacking in sophistication."  That is certainly how I view it now. Pity all the documentation got lost years ago, it would be fun to revive ( though not the pseudo-religion associated with it :-(  ) 

I did not learn linguistic terminology until I was 35, in Ling 101 in grad school. Like almost everyone, I found the concept of "phoneme/allophone" difficult to grasp-- to those of us brought up on "how letters are pronounced" it can be difficult to separate phonemics from spelling. But I believe it's basic to understanding a language. "Morpheme/allomorph/morphophonemics" comes a little easier after you understand "phoneme".

My point is, that the more linguistics one can cram into one's brain, the better to create a reasonble, interesting, and sophisticated conlang/auxlang.  I don't expect everyone to grasp the ins and outs of Generative Phonology, Chomskyan TG or successor theories, Optimality Theory (which I personally don't quite get) etc. etc. But the basics, yes.