Just stepped back in and caught up. Lots of stuff going on in the few hours I was away.
Regarding min fee, I think Passion_ltc's estimate is good: 0.25
That's assuming there is a large jump in Nxt price. $1 USD per NXT seems a bit unrealistic to me at this time. Then again, as I like to tell my friends, if there's one thing I know
I do understand about cryptocurrency markets it's that
I don't understand cryptocurrency markets.

On Visa speeds and marketing: If saying we have Visa speeds or 1000 tps is worrying people that it may lead to unfulfilled expectations then I think
we don't have to compare Nxt to Visa in marketing. Instead perhaps we should say near-real-time or near-instant transactions. Or just
the fastest transaction times in cryptocurrency.
On Turing-completeness and marketing: we shouldn't use the T-word in marketing, or perhaps only in the fine print. Instead, we should push "smart transactions" or "complex transactions automatically" or "programmable transactions" and have aside-style blurbs to explain further for those interested.
On Turing-completeness and other Nxt work: some people expressed concern that we would be working on this instead of other features. They might have a point. However, only speaking for myself, the area of virtual machines is something I have experience in and is a place I can contribute. Other features of Nxt, that's all new territory for me so I can't just jump in an be productive in those areas.
I took a look at zero-knowlege proofs and that might as well be black-mathemagic to me.
But VMs I learned how to do those back in the '90s.
Forging on low-cost computers with limited storage: those days may be numbered. Leasing forging power will be where we want to shift marketing once it is implemented.
On 0.5.12: oh shit. Upgraded.