I think things are getting better. Teams are starting to figure out what to look for in goalies. It only took them 50 some years to start to figure it out.
The voodoo portion comes from those who try to design predictive models for drafting and those predictive models fail miserably at identifying goalies. Because being a goaltender is about more than your stats. Most of those predictive models don't do one thing. Actually watch the players. It's a lazy excuse to not actually watch the players and identify with your own eyes which ones have the necessary ability to become good.
I think the problem is that there are no good predictive models for goalies of draft age (because you need such a large sample of shots against to get a prediction you can have confidence in).
There are pretty good predictive models for skaters. And scouts have always been better at predicting skater talent than goalie talent. As skater predictions improved with analytics supporting scouting (and vice versa), teams now have more confidence with their skater draft picks.
But goalies are still a crap shoot...
You can actually see evidence of the rising confidence in skaters over goalies at the draft. From 2000-2011 there were 31 goalies taken in the top 32 picks of those 12 drafts; 2007 was the only year there wasn't at least one goalie taken.
In the 13 drafts since then, only 5 goalies have gone in the first 32 picks (Samsonov in 2015, Oettinger in 2017, Knight in 2019, Askarov in 2020, and Cossa and Wallstedt both in 2021); 9 of the 13 drafts had no goalies go in the top 32 picks.
The last goalie picked in the Top 10 was Carey Price, 5th overall in 2005 - that was 20 year ago! From 2000-2004 there were 7 goalies taken in the top 10 picks (DiPietro 1st, Krahn 9th, Leclaire 8th, Blackburn 10th, Lehtonen 2nd, Fleury 1st, Montoya 6th).
Goalies were always voodoo. Analytics can't help for draft age goalies, but it can for draft age skaters. Teams have moved away from throwing away draft capital on goalies because skaters are a wiser investment.