What started this thought chain off was a post offering that the answer to differentiation for a mixed ability classroom is to use open ended problems. This seems to be a widespread consensus based on how often I've read the same idea. That feels problematic to me from several angles.
1. I'm not sure if there are enough open ended problems that provide different levels of depth and easy entry that span the entire curriculum. I can see such problems as being useful some time but potentially needing to be interleaved a lot of the time with less flexible material more focused on concepts needing to be gone over.
2. These type problems are almost always prescribed in mixed random groups. How far can one student pull such a group if they are diving deeper and what kind of experience does this provide for all the participants? For that matter, if a student is behind the others will they more or less rely on the group to peer teach whatever underlying concept is involved? How well does that work? While the problems themselves are good the overall structure aims at collaboration and creating a common experience for everyone rather than differentiating per se.
In search of more ideas on what an open ended problem / curriculum might look like I was reading through this paper: http://math.sfsu.edu/hsu/papers/HsuKyshResek-RichProblems.pdf
"Working in well-facilitated small groups on rich problems that are accessible puts students in the position of differentiating the content, processes, and product of their own work. When students are empowered to make natural choices as they work on rich problems together, there are almost always surprises for teachers and often for the students themselves. One of the most important surprises is who comes up with interesting ideas; it is not always who a teacher would have predicted. In this article we discuss what makes a problem rich enough to allow facilitation of this self-differentiated student work."I did like some of the offered ideas and was thinking about this one:
"Which positive integers can be written as the difference of two squared integers? For example, 17 = 92 – 82?"
The authors caution that the teacher must be ready with questions to keep students going as they get frustrated. Based on my experiences I wonder a bit of how this would work in practice? I can easily picture kids trying out sample numbers and noticing patterns but I don't see them getting from there to concrete reasoning on why this is generally true without a sufficient bit of number theory/modular arithmetic. i.e. all odd numbers 2n + 1 squared are 1 mod 4 and all even numbers 2n are 0 mod 4 and the only combinations you can get subtracting them are 0,1,or -1 which implies all odds and multiples of 4 but not the non multiple of 4 even numbers i.e. 2 mod 4.
I also don't quite see the pattern noticing section as being particularly fine grained in its accessibility. The most likely outcome would be some kids noticing it after writing a table of numbers and everyone else going "oh that looks true.."
If I were giving this problem I might start with some modular arithmetic background to provide some tools for the kids. Or perhaps treat this like a "3 act" exercise. Working on it long enough to deduce patterns and try to find reasons for them, providing the info on modular arithmetic as a group and then resuming in groups to see what could be discovered more formally.
So I found something that I might use but I'm still not certain that this is differentiating in any significant manner. I also need more observation to see how it goes with real kids. My main feeling is that group work on interesting problems remains a good practice but one that aims at collaboratively learning as a group. It still suffers from weakness when the members of the group have different amounts of "ability" or experience especially as the gulf widens.