"It's actually self-evident for anyone who knows both languages that Chinese is semantically more dense than English and by that effect alone it's much faster to read"
By the same token, it is self-evident that English text, gzipped, then base-64 encoded, being denser, will be much faster to read than bare English. Because of that, I do not think that argument has much value.
It's not flawed in its context. It would be flawed if your point was valid. For your point to be valid, the obfuscation you imply should either be true, or unknown. It's not unknown to whoever knows the language. It's simply false.
In other words, you cannot assume a variable to be unknown when one of the parties does know it.
By the same token, it is self-evident that English text, gzipped, then base-64 encoded, being denser, will be much faster to read than bare English. Because of that, I do not think that argument has much value.
On the other hand, contracted Braille is more complex than uncontracted Braille, but reading speed _in_cells_per_second_ seems to be about equal for both (http://faculty.sfasu.edu/mercerdixie/spe520/uncon_vs_cont_br...). That makes reading contracted Braille about 30% faster than reading uncontracted Braille (http://vision.psych.umn.edu/groups/gellab/Legge99.pdf)
So, I do not rule out that something similar applies to Chinese vs English.