Hi Andrew - you make some good points. I have also been looking at OO-SF for nearly a year now. I have a slightly different set of dot points to you. In the spirit of avoiding the usual taking points, I'm happy to accept that different people can get different things out of a particular subject. So while some of my talking point might mirror your own, others might not - but that doesn't make me right and you wrong, it's just a different look at the same subject.
And I hate this dancing around, so I'll just deal with the facts - at least as I see it.
OO-SF is a track standard, not a wheel standard. It can be for anyone to use for any scale or wheels they can make it work, if they desire.
Some, perhaps many, wheelsets can operate without modification. Anecdotally, many people have found that their wheelsets work fine, but others having heard this have tried for themselves and found the opposite. I am one of those people, for example. So such an experience can leave them confused and frustrated and perhaps questioning. The publishing of some numbers really helps modellers decide if their wheels would work or not work - but the success or failure of the wheels has nothing to do with OO-SF, which is a track standard.
I think the strongest reason for adopting OO-SF is to be able to run code 88 or semi-scale wheels, which do not run well on traditional HO standards. That does not mean that OO-SF runs better than those other track standards, because those track standards were never designed to run those wheels in the first place.
Gauge narrowing through turnouts must result in a greater minimum radius for the same wheels used, especially if the turnouts are curved. The same principal applies to gauge widening on normal track - widen the gauge, you can run smaller radii; narrow the gauge, greater radii. This may be of no importance to some, or critically important to others. But it is a fact. (edited in italics)
There are a few options for running wheels with a smaller width:
have wider flangeways with controlled depths so that narrower wheels run on their flanges. The European NEM standards have this written into their track and wheel standards, for instance. It has been mentioned many times on many forums that a good way to get reliable running of smaller width wheels through traditional wider flangeways is to fill in the bottom of the frogs with small strips of thin metal or plastic to allow for flange running- and some manufacturers make frogs this way so you don't have to modify anything. This isn't new. (edited: additions in italics)
use a finer track standard designed for those wheels such as Fine:HO or the UK equivalent DOGA-Fine. Again, not a new idea.
use the current HO standard and just have narrower flangeways, which are still within the published standards. I don't know why this isn't discussed much, but I have since found some older threads on the subject, so I am looking into it further.
narrow both the flangeway and the track gauge - OO-SF.
That's four options, each with their plusses and minuses. What you choose to use is up to you. Is OO-SF better than the other three? I guess it depends upon what is important to the individual concerned.
What do I think? My only gripe - my one and only negative - is the inability to find out factual information to help me decide for myself. Too many claims made that, when I go to the trouble of building a turnout and testing, turn out to be false. I feel I have wasted considerable time. And believe me, it is not helpful for some poster to airily claim that I "should" have known such and such, or that this-and-that does not apply, when such claims for OO-SF are made multiple times. "Try it for yourself, you can't go wrong" - well, I did, and it didn't work - for me. Trying to work out why and nail down some actual numbers uncovered the culprit - not every wheelset or wheel standard will work. That is not a drawback for OO-SF, which is a track standard - but it is not helpful to have some wild claims made to the contrary that turn out to be false. (edited: emotive language struck-through, my apologies)
I'm glad Martin helped out, he put up some numbers on a OO-SF web page. Now everyone can look and decide for themselves if their wheels will work.
My last point - OO-SF will work or not work depending upon engineering, not on your beliefs or motives. Beelzebub can use OO-SF if he wants to. Questioning someone's motives or other such clap trap belongs in a cult, not a hobby.