Oh, you're not talking about him dismissing news of practical attacks as a direct threat, you're talking about him being dismissive very early on about the idea that he was doing a bad job of picking a hash.
From the threat it looks like Linus' threat model for git is simply different from the common use today. Of course a judgement under a different threat model would make it seem like a bad tradeoff!
https://www.metzdowd.com/pipermail/cryptography/2017-Februar...