> I always wonder whether people advocating phonetic spelling have ever encountered someone who speaks a different dialect.

For just this reason, English spelling reform was probably doomed from the start. Any new standard seems just as arbitrary as the old one to a speaker of a non-standard dialect. The chosen set of vowels—whatever it might be—would ring particularly untrue to the ears of millions of English speakers, who might use any 12 to 20 vowels of a pool of a few dozen in their own speech.

My own example: I grew up in a part of the U.S. that makes no distinction whatsoever between the vowels in 'thought,' 'lot,' and 'father'—yet these are three separate sounds in Received Pronunciation, and might be divided into two elsewhere. Not only can I not make three different sounds for these vowels—I do not know how they would differ!—I cannot tell the difference between them when listening to someone who can. (Perhaps the 'a' of 'father' I could note in contrast to the other two with some conscious effort.) If I were to use a phonological spelling system that split this vowel group into three, I would have to memorize by rote—again!—the spelling of many words. What was supposed to be an easy system that did away with rote memorization, turns out to be more of the same. Perhaps this second learning curve is not so steep—no 'ough'-es—but I've already gotten past the first one! Why bother again?

If, on the other hand, a phonemic alphabet were provided—something like a more elegant IPA—making our spelling as idiosyncratic as our speech, we would just have a new set of problems. First, second language learners would find no refuge from the dizzying amount of dialects and accents while they were still learning fundamentals. This would be a sore spot for any language, but especially for a common lingua franca like English is today. And then, the language of the law, the academy, and business might recede still further from students unprivileged in class, birthplace, and schooling, if they first learned to write in their own dialect, not that of the ruling classes.

There is an advantage to an orthography having some remove from the spoken language, as it provides a common ground for speakers of different dialects to communicate. Chinese ideograms do particularly well here—to draw us back toward the article—as entire articles might be written which could be understood equally well by two speakers of mutually unintelligible dialects, dialects so far removed from one another that they could be called separate languages.

I don't think I'm letting you in on anything new here, but as it is a topic I've put some thought into, I took the opportunity to turn into a windbag.

Sorry about the Twain link. Here's Gutenberg[0], if you're interested in the text all the same.

[0] http://www.gutenberg.org/files/70/70-h/70-h.htm#link2H_4_001...