In my spelling reform proposal for Spanish, I floated the idea of allowing for equally acceptable spelling variants in some cases, to allow for greater ease for those who have variant pronunciations of some words. I believe this shouldn’t cause many problems when reading for those who don’t have those pronunciations and don’t go for those spelling options, provided the options don’t affect too many words or cause collisions with other words. If the reader is also familiar with the variant pronunciation, there will be even less chance of a reading problem.
I believe a certain amount of this is also necessary in a reformed spelling for English. The fact is that for many English words, there is more than one acceptable pronunciation. And sometimes, these variant pronunciations don’t have to do with systematic accent differences between different varieties of English, but are equally acceptable even among people who have what appears to be the same exact accent.
Two examples: data, where the quality of the first vowel can be TRAP or FACE (I say it with FACE), and amphitheater, where some people (like me) pronounce an F sound for the PH, and others a P sound.
Similar cases are herb and vehicle, the first of which has a silent H in American English but a pronounced one in British English; and the second of which does not have a silent H in some accents of American English.
In cases like these, I think the best solution is to allow variant reformed spellings: datta/data, amfitheeater/ampitheeater, urb/hurb, veeikl/vehikl.
A different case is where there is a systematic accent difference. If the difference does not result in any kind of phonemic split or merger, then it does not need to be reflected in the orthography. For example, there are many different ways to pronounce the GOAT vowel depending on accent. In my accent, as with many in America, the first element of the diphthong is a back unrounded vowel ɤ; in Contemporary RP, it’s a central unrounded vowel ə or schwa; in Minnesota and Idaho you can find an upper mid back rounded vowel o. In some accents, this vowel is not even realized as a diphthong. These differences do not need to be reflected in spelling, because they do not result in the GOAT vowel intruding upon any of the other lexical sets, or in a part of the set splitting off into a new set. In other words, these differences do not introduce any new minimal pairs into the sound system of English.
In other cases, an accent difference does introduce minimal pairs. In some accents, the so-called wine-whine merger never took place, and because of it word pairs like witch and which do not sound the same; they start with different sounds.
What should we do if a minority of speakers of English exhibits a feature in their accent like that one? Should a reformed spelling reflect it? (The very conservative traditional spelling does.) If we eliminate the WH spelling whole cloth, it will help speakers without the merger spell words with it; but it will mean people without the merger will have to spell some words the same that they pronounce differently. On the other hand, if we preserve it, it will mean people who don’t pronounce the difference (which is the great majority of people) will have to learn those spellings by rote. However, they will have no problem reading those words.
So, it is pretty clear that a “show all possible contrasts in all accents” approach to spelling reform will not necessarily cause reading problems, but it will surely cause spelling problems for many speakers. In a case like WH, I think a viable compromise solution is to allow variant spellings. People without the wine-whine merger could spell why, which, witch like whi, which, wich and people with it could spell them wi, wich. Context will help speakers without the merger read texts written by those with it, and this is a case where the set of words like which has a low functional load.
On the other hand, nobody, with the possible exception of time travelers, would have any reason to object a reformed spelling of hoo for who.