[Sorry for the wall of text.]

I did read the article. What I mean is that difficulty = complexity, when it comes to a script, has a lot of different aspects, and it’s practically impossible to weight them together. And to complicate (eh) it further, plenty of those “aspects” of difficulty = complexity will be language-specific.

Note that my issue is not that the claim would be discriminatory, like James assumes from Stevenson; it’s about being vague and useless unless we specify what we’re talking about.

I’ll dig a bit deeper into the text:

[James] rendering of foreign words

Why is James being so oddly specific with phonetic rendering, if semantics plays a huge role on Han characters?

If we play the exact same game from the other team, judging semantic rendering instead, here’s how it turns out:

• Latin: English ⟨water⟩ /ˈwɔːtə/, Czech ⟨voda⟩ /'voda/. Those two mean the exact same, they’re true cognates, they sound similar-ish… and yet they look nothing alike. Why is the Latin alphabet so hard?
• Han: Mandarin ⟨水⟩ /ʂueɪ˦˩˦/, Japanese ⟨水⟩ /mizu/. Those two aren’t even related, they sound nothing alike, but they look the same since they mean the same. Han characters are easy.

…in both cases we’re fooling ourselves, as if we reached some meaningful conclusion. We didn’t - it’s simply a poor argument based on a straw man over what each script is doing.

‘Tom Stevenson’ is far simpler and more phonetically precise than 汤姆•史帝⽂森，‘Tangmu Shidiwénsen’, which adds two syllables, six tones and six individual character meanings.

Except that a lot of this imprecision doesn’t come from changing the script, but rather changing the spoken language being used with that script. Show that same written name to a (for example) monolingual Spanish speaker and watch as they butcher its pronunciation, and vice versa; same deal if you show Spanish names to monolingual English speakers. Both languages use the Latin alphabet, but where’s the phonetic precision?

And how he re-transliterates the name into the Latin script, with ⟨ng⟩ and ⟨sh⟩, exemplifies a hidden complexity of alphabets - they assume a certain set of sounds or phonemes. You have no letter for /ŋ/ or /ʃ/ because Latin doesn’t use either, got to hack it with digraphs, now you got another layer of complexity. And you do need to relearn them depending on language; ⟨sh⟩ for English and Romanised Mandarin, ⟨ch⟩ for French, ⟨sci⟩ for Italian, ⟨sz⟩ for Polish, ⟨sch⟩ for German…

Han characters as used by Sinitic languages do this, a bit; after all they do have a phonetic component. But it’s way less of an issue than in a purely phonetic system, like an alphabet. (In a purely logographic system this would be a non-issue.)

The Committee for Language Reform in China acknowledged the relative simplicity of the Latin script as one of the factors behind its abandonment in 1956 of the attempt to develop a phonetic script based on Chinese characters.

If anything this argument can be used against James’ claim - the Committee saw no pressure to introduce a phonetic system for Mandarin, because Han characters were working fine. (There’s one by the way, bopomofo.)

then he might prefer to consider the example of the long-vanished Tangut people […]

Note how now the criterion shifted - from phonetic precision to number of characters. Sure, if you have more characters they’ll be more similar, and you’ll need to learn more of them to use the system… but it has its benefits, that James is ignoring.

One of them is that homophones don’t need to be spelled the same, by design. So, what if ⟨時⟩ “time”, ⟨石⟩ “stone”, ⟨十⟩ “ten” are all pronounced /ʂʐ̩˧˥/? If you see them in a text, you have an easier time telling them apart.

And this is so advantageous that you see something similar popping up in phonetic scripts - by violating the phonetic principle. Cue to English ⟨two to too⟩, German capitalisation rules, Portuguese ⟨há a à⟩…

Another is that, the way that Han characters were built (and Tangut by inspiration), even if you don’t know what a specific word means, you can look at the component radical to throw some guess that, alongside the context, might be enough to understand its meaning. For example, if ⟨木⟩ is “tree”, what do ⟨林⟩ and ⟨森⟩ mean? (Grove, forest)

Now, specifically about Tangut. Odds are that it had such a large number of characters (~7k) and radicals (~750) because it was short-lived, not the opposite. If the system had any chance to mature, odds are that those numbers would drop, as speakers ditch the “fluff”.

[From the blog author, not Benjamin James] To accomplish this arduous task, one of the first things I had Nikita do was go off to Kathmandu to study Classical Tibetan for a summer.

The Tibetan script is also phonetic (an abugida). The phoneme/grapheme correspondence is extremely opaque for the same reason as English - the spelling for both was made for an older version of the language.

If you only have one year to learn a new script, don’t try Tangut or Chinese.

If you have only one year, odds are that you won’t learn the written form of a language, no matter if it uses a phonetic or phonetic/semantic system.

The former was even created by someone who was himself illiterate!

People often highlight the genius of Sequoyah behind the syllabary, but rarely talk about how much hard labour it involved. This sort of “one symbol per word is too much, I’ll need to represent word fragments” insight is not something that comes up naturally, it’s clearly the result of trial and error.