I also prefer tops I can spin on a table. Keep in mind, however, that if you could extend a Simonelli's tip just enough to give it a good hard twirl on a table without scraping, you'd still have a nearly ideal design for maximizing spin time.
How do you think stem diameter is a important parameter? Is it its relation to the maximum rpm one can achieve?
Exactly. For any top, the greater the release speed, the longer the spin, all other things (like tilt) being equal. And the maximum release speed a given user of a given finger top can actually attain will depend in a complicated way on stem diameter.
Ideally, the thinner the stem, the faster the release speed. But at some point, your fingers start rubbing against each other -- especially during strenuous twirls. For a finger top with a typical axial moment of inertia (AMI), my rubbing limit is around 3.5 mm. For a very low-AMI top, I can go as thin as 3.0 mm, but only with great care.
But 3.5 mm is way too thin to start a high-AMI top like a Simonelli -- hence his tapered stems. The idea is to grip a thicker knurled or splined part of the stem to deliver an optimal starting torque and then let your fingers climb the stem to ever smaller diameters as the twirl proceeds -- not unlike going from lower to higher gears to accelerate a car.
As finger tops go, most of mine have intermediate AMIs. These get good spin times with a splined but untapered 4.8 mm stem at least 16 mm long. For the high-AMI tops, I use the twirling technique described above with a stem stepping down from 6.0 to 4.8 mm
For spin time, the last thing you want is a stem ending in a short knob.