I’m learning new characters while referring to a physical copy of “Remembering Simplified Hanzi 1”. The HTML5 beta now shows me how to draw each stroke of new characters (which is great; I missed that from the Android version - thank you!)… but the characters themselves seem to be from anywhere in the same chapter, leading to much page-flipping and searching. 
Hmm, we do add a bit of a randomness value to the readiness calculation for items. It is possible that for new items that don’t show up immediately due to an existing backlog of items in the due this is happening. When you’re getting these new items are you noticing it immediately after adding or at a later time when they appear?
It seems to be non-random again now. It was coming up with new characters randomly within a section for a day or so.
I’m happy with randomized reviews, but randomized new-adds (they were within section 22 of Heisig RtH1) were really annoying.
A problem I’ve encountered with Skritter is that characters from lists occur in a non-random order (e.g. adjacent characters sometimes have similar radicals or, in the Heisig method, “primitives”) and Skritter often won’t “shuffle” them enough. It’s fine to see them in some structured order for the first time, but later on that is actually a hinderance. You wind up leaning on test-taking skills to answer questions instead of actual learning. Users should be able to choose the amount of shuffling with a slider bar in the settings. Incidentally, this should also fix the problem mentioned at the top of this thread, because one could also set the randomness to zero.