They deployed a bugged service worker long ago (beginning of 2020 or something), but service workers can be hard to replace because they tamper with requests an

They deployed a bugged service worker long ago (beginning of 2020 or something), but service workers can be hard to replace because they tamper with requests and cache. "Technically" it's been fixed in production for a while, but it doesn't propagate so many still have the bug.. Still seems very strange to me that they allowed this situation to happen for such a huge (in traffic) website. Not sure how exactly the service worker was broken (I didn't have the curiosity to check in devtools, and now it's gone), but I had this exact same issue and saw other people having it right here on HN, and this was the way to solve it too.