I don't think that this will cause any large impact on the GRE's role in the admit process simply because students could always send only the most recent GRE by slightly altering there information when registering for the exam (for example, by changing one's address, or including a middle initial). I discovered this myself when I took the GRE four years apart (so my address had changed) I was upset when I realized on the most recent score was sent out as part of my "free" score reports, and I had to call ETS to have them merge the records (I hit the 800Q on the second one, but suffered mild drops in AWA and V, so I wanted both sent out). So all that scoreselect does is codify a loophole that's presumably been well known for a long time (or at least known by the type of student who is also likely to abuse scoreselect).
But perhaps I am underestimating the impact of moving something from unacknowledged loophole to legitimate part of the process. It also raises the issue of the misaligned incentives of ETS, being a monopolistic, private company (albeit non-profit), and academia. It's in ETS's best (short-term) interest to have people take the test even more times, no longer afraid that the newest score could sink them. But will this negatively impact the application pool and process? Unclear, since the GRE is a pretty crude instrument as it is. This change on the surface benefits those that are bad test-takers and freaked out on one test, and it also helps those that can afford to take it N times. But it also helps those that are less in the know and didn't know about the pre-existing loophole.