Studies of languages have shown that any given language will compensate for complexity in one area with simplicity in another and that if you compare any two languages that they have roughly the same number of complex and simple areas.
Consider the following:
Imagine a language with only two vowels and two consonants, a rule that forbids consonant clusters and syllables ending in a consonant and without significant tonality at the morpheme level. We could characterise such a language as phonologically simple. However, discounting the possibilty of homophones, only four words of one syllable and sixteen words of two syllables would be available, which would mean that almost every word/ morpheme would have to be of at least three syllables; such a language would of necessity be morphologically complex.
Going to the opposite extreme, a language with a hundred vowels and three hundred consonants, even without allowing clusters and final consonants, has the potential to provide umpteen (I leave the mathematically gifted to work out how many) words/morphemes of one and two syllables and thousands more if clusters and final consonants are permitted. It is likely that such a language would be morphologically simple. It would of course be almost impossible to speak since the differences between the phonemes would be so subtle that both speaking and listening would require the utmost concentration at all times.
So we can see that both excessively simple and excessively complex phonologies would prevent effective communication. And the same applies to any language considered as a whole. An overly simple language would be of no use except to communicate at a very basic level indeed and an overly complex language could not be passed on to children.
There is a tendency to think of isolating or analytic languages as simpler than synthetic or agglutinating languages, but that is not the case.
Take these four sentences:
1. The bird ate the fish
2. The fish ate the bird
3. Avis piscem edit
4. Piscis avem edit
We know (if we know both Latin and English) that in 1 and 3 there is no doubt that it is the fish that had the bad luck, and that in 2 and 4 that it was the bird. English achieves certainty by word order and Latin ( where the order of words in 3 and 4 can be changed without altering the meaning) by changing the endings of words. From the perspective of an English schoolboy learning Latin, Latin is difficult because you have to change the endings, but from the perspective of an Ancient Roman schoolboy learning English, English is tricky because you have to get the words in the right order. English schoolboys wonder why Latin bothers with endings and Ancient Roman schoolboys may feel that English lacks flexibility because you cannot put the word you consider important first.
Agglutinating languages are often presented in a way that makes them look horribly complex. I have seen the Turkish verb presented in what appears to be a neverending set of tables. Further investigation though shows that the forms are made up applying quite simple principles and that the principles are applied universally, so the verbal system is not as complicated as you first thought.
A language can be neither too simple not too complex and all languages are simple/complex within a very narrow set of parameters.