I have seen some really good results using presets as long as you are maintaining a good zero set procedure.
Will you ever get 100% perfect presetting ? probably not with today's technology but really close is a good start.
The issue I see often see is that you are presetting with ink that has not yet been emulsified with water and this causes differences when you start running. Most preset software should come with a learning feature which looks at the ink setting for each key that the software gave you for start up and then looks again to see where that key ended up and make some form of software fine tuning. Again not a perfect situation but better than nothing.
On a commercial web press I would have thought this would be a valuable addition to help reduce paper waster/time
Yes, zero setting is critical but very hard to do accurately with presses with ductors. Can be done accurately with "digital inkers" such as with Goss's and the one QuadTech has now. The zero setting is just a matter of stopping the ink feed with their ink zone pumps.
IMO, emulsification is not an issue even though it seems to be. The problem with different levels of emulsification in the roller train is that these differences affect the ink feed rate. When forcing the ink feed rate to be independent of changes in water, the print density is very consistent and therefore can also be predictable.
There are some ink key software that does attempt to have self learning capabilities. It is a nice idea and seems to make sense but the problem is that it is not mathematically valid. If one has the same problem to solve, then a self learning process can help to fine tune the result on successive tries. But presetting of offset presses is not the same problem for each setup. Each time you start up a new job, the problem of presetting is totally different mathematically. Therefore if you try to use a software that is trying to learn what to do, you will always be wrong. It will chase its tail.
If you have software that was mathematically capable of doing the needed calculations, then it would have to know all the issues that are needed to make an accurate calculation of what the presetting values should be in the first place. So why not do that.
There is no need to be perfect. If you have a tolerance for your densities that are acceptable if you run within those ranges, then the target is a large barn door. If your density tolerance is +/- 0.05 then your ink tolerance is about +/- 7% of ink volume. That is a total range of 14%. With newspaper press inks, that range is much more. So if you start the press and the densities are out of tolerance, it means you can not even hit the 14% barn door.
If you had a consistent and predictable ink feed with good zero setting and an accurate calculation of presetting ink keys, you would have very low set up waste. Much lower than what could be obtained with existing press technology and in-press density closed loop systems.
Of course I love trying to understand this topic and looking at valid solutions. Unfortunately the industry does not. :-(