It seems you stole my topic!
But I couldn't of written it better, so I guess that's better - no misunderstandings that will occur because I still don't understand the subject very well. Though the articles you gave me helped a lot, thank you - I'm still going through them throughly.
Though I will try my luck explaining first and then giving an example based on that explanation:
Si seems to take great care about the past familiarities of frozen parameters and snapshots of reality (like words said) and thus a place to point you at in case you need to recall data. Kind of like when you copy a math exercise to your notebook, you have to look several times at the book (which is the snapshot of reality) in order to capture it better and then go back to writing (which is the prediction).
As Ne seems to look at patterns on the here and now, it gives a higher priority to the connections that seem to underly just beneath the physical world and learn relationships between objects as part of a whole bigger picture. An example I could give here would be looking at a rug from up close first (realizing the object without great detail), watching carefully for the pattern it creates
while taking a step back and see how it acts to create a bigger picture (the rug itself). The act of prediction through Ne would probably occur when you see another rug that seems similar from up close (realizing the object) and then probably knowing the resulting bigger picture.
The prediction with both Si and Ne would probably vary in the amount of "stepping back" of the Ne and the detail intake of Si, but I would assume that it would be a lot like the Ne process only sometimes getting close back again just to make sure you are on the write path. The Si guides the Ne like EyeSeeCold says.
My personal experience with future interactions within the past and present (I still like calling it prediction, because it is a common definition in Mentalism) can be described here:
(I described it really shitty, so I wouldn't be surprised if you wouldn't understand what I mean here)
Walking through the same road to work stored details about the road, how fast I walk it, how fast does the traffic lights change while looking at the objects as a whole - the road as a path, the speed of me moving as a correlation with the crowd and the traffic lights changes as parts of the huge traffic system.
trying to predict how could I walk the road with efficiency I used my Si to collect the parameters I chose and the experiences of the relevant parts in the system, switching to Ne I tried to find the patterns that are relevant to my path right now accordingly with my Si parameters as the variables available to me right now - it results in finding traffic lights patterns that are more efficient then others and thus choosing them over others along with the distance from the target and the speed of walking in the crowd.
Just something I noticed from that experience. I seem to freeze it as a snapshot so I could use it in further instances because logic says the road isn't going to change, the traffic lights are supposed to be the same everyday if I'm walking at the right hour and the only thing that changes is the crowd and my walking speed which needs to be adjusted to predict the future. But that's only after there is a pattern to my pattern finding, which is possibly even more complicated because it relates to a system of a system of prediction?