"that all writing systems, properly so called, are systems for writing out speech."
This is true by definition, at least by the definition of a "writing system" commonly used by DeFrancis and others. For the Chinese system, it is true that the vast majority of written characters are based on sound, but there are also characters that are primarily semantic in construction (e.g. pictographic, at least in origins).
The more complicated (and somewhat philosophical) question is whether or not it make sense to call something a language (or writing system) that is not based on speech, or cannot be represented as speech. Symbolic logic, for example, includes a formally defined set of symbols that represent meaning in a consistent and intelligible way. How we discuss these sets of symbols is a very interesting topic (albeit rather distinct from the topic at hand).
In the case of Chinese, the pictographic properties are often retained in the construction. For example, a common radical is that of the roof, which appears at the top of this character: 安 an1 meaning peace The perception (correctly or incorrectly) that the meaning of the word is to some degree derived from its parts affects the consumption of the language in a way that differs substantially from other languages with the roman alphabet. In the particular example, all sorts of people talk about how "peace" is a woman under a roof.
I think the point is that there is nothing phonetic about an emoticon or similar items. Doesn't mean you can not "read" it. Similarly, consider the reading of a map. If you are like I am, you convert the map you are looking at into the words you'd use to describe the path you want to take. Does not mean the map is at all based on those words.