Right now if you want to look at, for example, first touch Attribution, for an action that has occurred during the last two weeks, it is very likely that you will have a high proportion of that action attributed to none. This is because the look back window is calculated from the date range to a default (depending on if you are using visit or visitor). However, this current behavior assumes that you will always want the same look-back window across the board. In reality there are times where you will want to allow for a much longer look-back window and times when you will want it tighter. What is for certain, is that in MOST cases, you will not want a good chunk of your metric attributed to "NONE."
I believe a solution to this is to have the look back window (somewhat) independent of the time frame that you choose for your table (or visualization). It can anchor at the start date but should be variable as far as if you want to look back 30, 60, 90 days or even much more. From the documentation, it seems like someone at Adobe decided what they believed would be the best fit for this (a static look-back window applied across the board). However, this has led to confusion and often doesn't match use cases. For example, if we are looking at a period of time spanning a quarter vs a period of time of only one week, we're going to want to consider different scales of look-back. The same applies across different metrics, there are times where one metric may have a longer look-back vs another.
The solution, as far as I see it, is to have a date range in the attribution windows (whether you are specifically looking at only attribution or if you are changing from the default attribution on a right-click) that allows you to define look back window independently from the date range of the action you are measuring (so that you are still looking only at actions that occurred in your set date range, even if the look-back window is a larger window).