The very first input scheme the article talks about is Pinyin, with which you'd input biang by typing... "b" "i" "a" "n" "g" then selecting from a list (most likely a list of length 1). Not insane or even difficult.
(Except that biang is not encoded in Unicode yet so you can't type it anyway.)
Think about it, though. You type phonetically in one script, then select a character in another script that might be pronounced in a similar way. That's like entering Hangul syllables that sound similar to what you want, then choosing the right English character sequences.
What's there to think about? How else would you input a script with more than 50k characters?
> You type phonetically in one script, then select a character in another script that might be pronounced in a similar way.
Sure. Japanese works the same way, you input in kana or romaji, then select the suitable kanji (or kanji sequence).
Of course it only works when you have a regular phonology, that would be completely impossible for english since by and large orthography and pronunciation have no relation.
I've thought about it. I input Japanese every day, Chinese frequently, and Korean on occasion. The Latin alphabet is a first-class citizen in these languages; I don't see the issue.
In fact English is becoming the same way: when you input "apple" and choose the [U+1F34E RED APPLE; stripped from input on HN] suggestion you've done exactly the same thing.
(Except that biang is not encoded in Unicode yet so you can't type it anyway.)