It touches on that, too. It's also a hard metric to work with because we're getting better at _detecting_ cancers, including more benign and earlier stage ones. I believe the SSC post found that most of the improvement in cancer mortality was more an artifact of detection than actual improvements in care. i.e.: if you look _only_ at equivalent Stage 4 cancers, then the death rate from there hasn't changed much over time.

If you assume really old people don't have higher death rates, I suppose.

Both the Slatestarcodex article and the links you provided in your comment use deaths per 100,000. In fact, your article reaches a similar conclusion to Scott Alexander, that mortality is declining mainly because of smoking declining.