Your first claim is, as I said before, not necessarily true. Your second is simply wrong. This can be easily demonstrated despite your cites.
The most preferred pronunciation of February is Feb-yoo-ary, and yet both the American Heritage and Random House list the Feb-roo-uary pronunciation first. They do so, of course, because that is the older pronunciation, which either directly or indirectly makes the lexicographers hesitant to change the order despite the fact that the majority of people no longer say it that way. In other words, they are setting aside the reality of English as it is used by most people and instead are relying on their own judgment. I don’t begrudge them that, but I’m also not going to pretend that their judgment is the final arbiter of truth with respect to preferred pronunciation, because in this sense, “preferred” means only “preferred by a minority of people”. In this situation, the term has lost any useful objective meaning.
The word “interest” provides another type of problem. The two dictionaries list different pronunciations in the first position. In this case, I don’t know which pronunciation is actually preferred in reality, which isn’t surprising given the fact that both pronunciations are fully standard. I do know, however, that the blatant wordbook disagreement is direct evidence against the notion that the first listed pronunciation is always preferred. If that were really true, then it would be one or the other. They couldn’t both be right.
Now, I am not denying that in cases where there is a clearly preferred pronunciation (e.g. nuclear, realtor, ask), it will be listed first when a dictionary entry is originally written. Even the Merriam-Webster folks doubtless try to follow that guideline, even if they don’t worry overmuch about it when they fail.
But those guidelines are just that: guidelines. After the book is printed, it starts to carry with it an aura of authority, which makes the lexicographers a bit reluctant to switch the listed order even after the facts of usage have changed. Most seem satisfied if they manage to keep up with all the common pronunciations, which is a hard enough job by itself.
And age will always be important. Merriam-Webster has been around longer than the others, so it has more institutional baggage to work with, which means it’s less surprising that they list “fort” as the first pronunciation of forte, a choice entirely contrary to modern usage. The American Heritage is younger (spawned in a fit of lingual pique as a response to perceived excesses in Webster’s Third), which might give it a bit of flexibility. It’s not going to have older pronunciations already enshrined onto the pages of previous versions, and the people on the usage panel change their opinions as well. Despite the silly circumstances of its birth, it’s normally quite a good dictionary, and it manages to list “for-tay” first.
Even so, it’s not smart to rely just on that listing and not on any other evidence. To add another example of obvious error to the two I listed above, it’s well worth pointing out that Random House takes the old fogey route and has “fort” first. Despite your cite of its explicit claim that it lists “preferred” usages first, it fails to do so even in a case when the overwhelming majority of speakers (including professional wordsmiths) agree on the new pronunciation. I’ve never been a huge fan of Random House, and it’s no surprise to me that they fail so utterly at their claimed objective of ranking order of preference. I didn’t even have to look hard for it. All the examples I’ve used in this thread came off the top of my head. I needed to double-check them to make sure, but I had no need to put in any actual effort to find discrepancies. And there are many many more out there to be found.
The conclusion from all this is plain: It is just not true that you can rely on the first pronunciation being generally preferred, even if the dictionary itself claims otherwise.
