They also found a "surprising number of ‘off-target’ mutations," according to Nature News.
This is the thing which scares the daylights out of me around genetic engineering generally. While some genes expressing differently will present clear dysgenic properties, others may not evince until much later in life - or perhaps even in subsequent generations, by altering gametes or epigenetic in-utero signalling or whatnot in unanticipated ways.
Yes, we can do epidemiological statistical studies to understand what individual genes affect, yes, we can do this for combinations of genes - but we can't study the affect of genes on subsequent generations, apart from in drosophila studies and what-not - which frequently end up demonstrating that yes, these edits do have unanticipated affects a dozen generations down the line, particularly when multiply crossed with other variants, and random mutations.
If human genome tinkering ends up common place, we will end up either with strict reproductive controls, or we will end up reproducing through cloning, and our evolution becomes entirely self-driven. The alternatives are likely just too dangerous.
Edit: sorry, forgot one of my points. Right now, we rely on statistical studies to understand what parts of the genome do. This could be remedied by having adequate computational power and models to actually simulate an organism from the molecular level up. Only then will we actually have some degree of control, and true understanding of what we are doing.