While I agree with you somewhat, I also find value in an iterative learning approach - one that, in theory, constantly improves over time.
My problem with his approach, which I think you hint at, is that to properly improve you need to make sure you're optimizing the right metric(s), and those goals need to be clear. In Mark's case, I don't think that he has a clear goal he's trying to optimize all the time, so the different iterations on the system seem like they're swirling around a drain: slowly getting better and better, but sometimes just moving orthogonal to the most effective direction.
his paper reinforces a desire in me to be way more thorough in my reading; my head was overheating from newfound motivation
Mark Forster stuff is interesting, but he never stuck with this, he ended up doing version, have version after version of this... and is still coming up with new versions to this day.
I almost feel bad for him, because it seems he is obsessed with finding the perfect system, but despite dedicating a huge amount of his life hasn't been able to do so.
From his blog it seems he has been very ill recently, due to cancer and he is still trying to find a system and still trying to find something that works. Every system he publishes has worked for a few days or a few weeks, but then he moves onto something different.
I think although his insights are helpful, I think he is the perfect example of what happens when you focus on productivity as an end result too much rather than the real world.
I've always liked his "Final Version" method better, where you select tasks you want to do, and then do them in reverse order.
The "little and often" line is a takeoff on the job of a fireman for a steam locomotive.
So I've started using this, and it's really nice for getting into a flow of knocking out tasks. It's kinda like NADD candy or something.
1 nt m -> in lb
I mean can't even come close to doing it.