I thought I'd read something about it, but when I googled, what I did find was this old HN comment: https://news.ycombinator.com/item?id=8514519 https://news.yc

I thought I'd read something about it, but when I googled, what I did find was this old HN comment: https://news.ycombinator.com/item?id=8514519 https://news.ycombinator.com/item?id=8514519 > UTF-8 + gzip is 32% smaller than UTF-32 + gzip using the HN frontpage as corpus.