I really dislike it. We do time based reporting and it messes up all the calculations at DST change. All kinds if weird bugs that trigger once per year, especially when time skips backwards. It's also dumb how trains have to stop and wait for an hour at the winter DST change in order to maintain the schedule. It would be a breath of fresh air when it's finally dropped in the EU.
It was, and did, but that was based on a time when lighting drove total electrical load. That's no longer the case due to increasing energy efficiency in lighting and the growth of other common electrical loads.
I still think there's a wealth of useful data of the same type that can be found at timezone borders. Many will have the transitions as a slight confounding factor, but it doesn't completely invalidate the applicability of the data.
Comparing no-transition borders with transition borders might also yield some interesting findings.
If you want to compare the effect of permanent daylight vs permanent standard then just look at cities near the boundary. Drive a few miles away and voila you're 1 hour ahead or behind - throughout the year.
If one is better than the other, it will show up there.