Given that there had been wrongly installed letters continuously since at least 1956 and we have no proof that an entirely "correct" version ever existed, I'd consider the inverted H historically accurate and I hope it won't ever get fixed and especially not as an overreaction to the article.
Given that the original drawings did show all letters in the orientation that's obviously correct for the font, I'd be hesitant to say the upside-down installations were ever historically accurate. It most certainly wasn't FLW's vision.
And this where we disagree. The original drawings show the artists original supposed intention but what counts for historical accuracy is how the work was perceived for the majority of its existence.
Which, as the article and other comments here make clear, has been changing back and forth over the decades.
In a situation like that, I think you kind of have to acknowledge the original drawings as the preferred state for things. Especially when just about anyone with training in this area would readily agree that that design aligns with what they'd all consider usual or correct.
Sure, it has been changing back and forth which makes the pristine version the least historically correct because few, if any at all, ever experienced that.
