Improving scroll bar interaction

Swap out the tracking filter when the user converges to a scroll bar.

Note: By converge, we mean the user has driven velocity to zero and the current reported x y position, and the x y position shared between tracker via placement of cursor. So both tracker and user agree that velocity is zero and the cursor identified visibly on the screen.

OK, scroll bars have one dimensional degree of freedom, the the user still his his 2-3 bits of precision on velocity at comfort level. We want an entropy tracker that quantizes along the one dimension,  using most of the IO 'IQ' on the one dimension. The other dimensions shows out of bounds,  and have one velocity setting to do it, yield the other line of symmetry to the scroll bar. When exiting the bar, we swap back to the 2-D tracker, keeping the same comfort precision (or IQ).

The one dimensional case means extra velocity settings for that scroll bar, the user can jump it, smooth it, stop it on a dime, jump it up and down. More speed setting, with the one perpendicular speed setting needed to exit the bar. There is a slight bit of inertia on exiting, we accept that trade.
With control of IO we can do this neat thing, we can swap the tracker need on for serious graphics design work, boost the tracker 'IQ'.

Then, I do not have to tell you what the gamers are doing with trackers. They will swap out trackers constantly, based on the real time movement they need on the screen. Trackers slightly ballistic. Trackers that have position quants, user can jump at warp speed, like a tab. Add in joy sticks, complete mechanical simulators, and exposed the IO system and the tracking codes.  Big win for linux. Big boost for the industry all around.

What theory am I using?

TOE, the problem is one of queuing user moves through a finite channel. The entropy tracker is a queuing model, it is the WalMart checkout manager, keeping the clerks to customer flows matched and lines stable. Algebra, everywhere, is a whitening process along finite axes of symmetry. In this case, we know the axis f symmetry and the information bandwidth, we set the quants by making the Hufkman tree of his known preferences, we have tracked and computed the user's typical moves, and we have a generator; just like a plain text search engine. The tracker has a very good guess about what the user really wants on a move event basis.

A ballistic tracker

In this situation, it is all about the cursor, the cursor is an aimable, smart mover but has ballistic, inertial properties in a field. The gamer is very careful about the initial tack, he knows that the object has inertial and tend to stay on the ballistic tack.  The game interface IQ has gone up, likely to 3-4, the player is a devotee. The solution is a quantizing kalman filter, which has not yet been invented. This filter runs a fixed point kalman, but uses te redigiized math of the Huffman tree, which we can consider as a digit system. We develop quantixed math, digits with assigned paths.  Any number can be decomposed into a finite number of traversals through the Huffman graph, and we get finite bandwidth math. Hasn't quite been invented, but I will look.