This has probably been true for all unix fs's for a really long time. My take on it is that the amount of people that make better decisions than fsck are probably on a first name basis with eachother and I am certainly not one of those, so when fsck says "should we try to fix this?" the only question for me is "did I 'dd' this raw device to somewhere else or not?" but when I have that answer I might aswell let fsck do its best because I am not going to hexedit something into working shape that fsck could not figure out how to fix.
If the data is super important, I'd buy another drive and dd this one over to it somehow, so I can make 1-3-5 attempts at getting it fixed, but when fsck runs, I would let it run.
