I tend to think that there are a number of general problems around these Indus Symbols.
A. The first one is a number of completely oxymoronic concepts, such as the nondescript "logo-syllabic" word, which means nothing. Writing systems are either sound-based or meaning-based.That "logo-syllabic" word which tries to invent a hybrid concept is nonsense. The first obvious step is to categorize existing systems with words that make sense...
B. Another problem is the black-or-white conception of writing. There are lots of shades of gray between drawing and writing.
1. Pure drawing
2. Graphic narratives "2-D prevails over 1-D". For example: the map Lean Wolf drew of his raid for Sioux horses from Fort Berthold to Fort Buford, Dakota, along the Missouri River
3. Mnemonical scriblings "isolated symbols"; Road signs, Early cuneiform, pot-marks, possibly Vinca signs, Pre-alphabetic Touareg signs, etc. Pictish glyphs as well.
4. Defective writing "segmentally 1-D but many items (especially grammatical) missing". For example: Linear A, Late pictographic cuneiform. It cannot be read directly even though it is "worded" and means something.
5. Full-blossomed writing "not ambiguous for a competent speaker (or reader)"
- approximative: Linear B, Mayan, Cuneiform
- precise: Latin alphabet among others, Chinese
Before we try to "decipher" the Indus Symbols, we first have to understand where they stand.
Defective writing is well-attested in Mesopotamia. For example the first inscription in Hurrian lacks all grammatical morphemes. It's not even clear how it should be cut in sentences.
Before that stage, we also have plenty of Mnemonical scriblings: most (pre-)Sumerian inscriptions are of that kind. Quite surprisingly Egyptian Hieroglyphs pop out of nowhere directly to stage 5, which suggests that the whole prehistory of that system remains to be
If we turn to Indus Symbols, the linear 1-D linear strings suggest that this system is at least level 3. There is a kind of symbolic encoding in that system. It seems hard to believe that they are just drawing for the sake of drawing.
What the frequency calculations show is that this system cannot be a *precise and full-blossomed* writing for sure. Indus Symbols have the profile of sound-based systems and not that of meaning-based systems. Cf. Farmer-Sproat-Witzel graph. The consequence of that is that these symbols would encode grammatical morphemes if they were a precise script, which is absurd.
Now it is possible that we can further refine the statistical analyses. A first point is that a full-blossomed writing can be approximative: it may not write vowels for example or it may lack a number of features (syllable codas, etc).The impact of these approximations on the statistical profile of scripts is unknown. What is the profile of English when written in the std spelling, when written without the vowels, when written in Linear B? This is really the basickest thing we first need to know.
At this stage, it cannot be excluded that the Indus Symbols could be a writing system with a high level of approximation. This sounds a bit improbable considering the fact that we have zero long corpora in that system.
In my opinion, the issue is to understand if the Indus Symbols are just Mnemonical scriblings or if it is a kind of Defective writing. It would be nice in fact if Statisticians would concentrate on trying to find objective criteria that enable to differentiate 3 from 4 (and approximative 5). If these symbols are a defective writing it's going to be awfully complex to tell what kind of language it encodes as the clearest indications will probably be missing. Considering all the features of the system, no long corpora, attestations on very specific artefacts, my personal point of view is that these symbols are most probably Mnemonical scriblings.
It can be noted that it took more than one thousand years in Mesopotamia to move from Level 3 to Level 5.
Moreover full-blossomed systems can also be represented in Mnemonical scriblings: Linear B is a good example of that situation.