HN's software uses the canonical URL when it finds one and in this case it was bad (https://www.openread.academy/paper/reading?corpusId=null). We've reversed that now.