Microsoft's position has been the same for a while now. They are focusing on meeting the needs of their users by attempting to fix the bugs that manifest themselves most often. Additionally, they are trying to add support for the most requested features.
They say in your first link "As we said at MIX10, we will continue to build standards support into the IE9 platform and as a byproduct our ACID3 score will increase." and I have heard them say this elsewhere as well.
They are NOT "coding to the test". They are not going to attempt to fix the particular parts of the ACID 3 that are failing. Rather, they are attempting to improve IE 9 as a whole by addressing the failures they see as most important. Only as a consequence of that will the ACID score improve. I think this is a good strategy. If it takes 1000 programmer-hours to fix an oddball rendering error in ACID 3, but that same amount of time could fix 2 or 3 really POPULAR bugs that ACID 3 does not address (it's not designed to be comprehensive), I would think their resources are better spent on those more popular bugs.
Obviously what their priorities ought to be could be endlessly debated. As can the amount of success they are having. So let's not even go there. :)
Update: I'd like to back up my statement that Acid 3 is not designed to be comprehensive with this citation: From http://www.webstandards.org/2008/10/02/dowehaveawinner/
"Acid3 was not meant to be the one and only indication of a browser’s performance. In fact many other test suites are far more important."
and
"Many subtests are high on a developer’s wish list: Full CSS 3 selectors support, media queries, SVG fonts. Admittedly a few others test edge cases and more esoteric features – but the test was supposed to be a significant challenge!"
Although I am not an IE user, it's better for us all if Microsoft sticks to improving the important stuff and skips the tests that check for "esoteric features".