In its (admittedly weak) defense, if the submitter edits the title back during the few-minute window after submitting where this is possible, the normalizer does not kick back in. But it is a bit opaque, in the sense that the poster needs to: realize that the title has been changed; know that they can edit the title (but not too slow!); and know that it won't be filtered through again.
I have no idea what the website would look like without it, but I have a feeling it does more good than harm.
It's more of a gut feeling than anything else. I don't know what the normalization rules are, but I think we mostly notice the normalizer when it mangles something [1], and not when it's just quietly humming along. Someone with more spare time could probably figure out a dataset of original to normalized titles and measure that.
[1]: Example from the frontpage right now: "How Big Are Factorials?" probably got normalized to "Big Are Factorials?" originally, based off other previous manglings I have seen previously.
huurtehoog · · focus · HN ↗
Also, (2011)
GranPC · · focus · HN ↗
badc0ffee · · focus · HN ↗
GranPC · · focus · HN ↗
I have no idea what the website would look like without it, but I have a feeling it does more good than harm.
hunter2_ · · focus · HN ↗
GranPC · · focus · HN ↗
[1]: Example from the frontpage right now: "How Big Are Factorials?" probably got normalized to "Big Are Factorials?" originally, based off other previous manglings I have seen previously.