Hi Tim,
You are on the right track in analyzing Dick's paper. You need to pay meticulous attention to the details just as you are. However, when you get stuck, you definitely need Dick's help. Don't pay any attention to the advice Yanniru is giving you; his latest post contains a number of errors of misunderstanding.
For example, the unknowable data is not unsampled data, as Yanniru suggests. If you want some handle to help you visualize what it is, you should think of the unknowable data as being some explanation or theory that successfully explains all the knowable data. Explanations and theories can all be expressed as sets of numbers too, you know, so to "come up with" the unknowable data, you just have to imagine that someone has worked out a theory that successfully predicts all the knowable data you have and that that theroy has been reduced to a set of numbers.
Now, as to exactly how that data is added to the original observations, I'm a little over my head so you need to get the correct answer from Dick. But until he returns, and to help divert you from Yanniru, here's kind of how I think it works.
The unknowable data is added to the data in each observation, but it needn't be exactly the same set of data for each observation. It would, however, represent the same "theory" or "explanation" that is used for each observation. I don't know if you picked up on it, but Dick stipulates that the unknown data must follow all the same rules that the observation does. So the combined set of data making up an observation along with its component of unknowable data would represent something like one example or instance of a theory or explanation in a specific case of one observation at one particular time.
If that doesn't help, or is too confusing, just forget it and wait for Dick. But, please don't give up on trying to understand Dick's result. I think it is truly profound and I think your effort will be well rewarded if you stick with it.
Warm regards,
Paul |