Honestly, I harbour just a smidgeon of hope that cases like this will nudge companies to reevaluate the whole “find text that could be a link with a missing pro

Honestly, I harbour just a smidgeon of hope that cases like this will nudge companies to reevaluate the whole “find text that could be a link with a missing protocol” feature and realise that it’s a terrible idea that causes more trouble than it solves. If you want a link, write a proper URL. The embedded tweet shows another problem, too: > Grrr... Because .zip is a valid TLD, it's impossible to know whether http://t.co/webB2l1Y9w http://t.co/webB2l1Y9w should be a URL or a filename. Twitter mangle your links to use t.co including the presented text, so that embedders have no way to determine that the text was supposed to be “example.zip” without following the link or the tweet link. There may have been some purpose to t.co quite a few years ago, but the reasons justifying it vanished completely a few years ago, leaving behind just something that is completely hostile to users and security common sense. (This also reminds me of Cloudflare’s “email address protection” feature, which catches and mangles (in a you-need-to-run-our-JavaScript sort of way, so it doesn’t actually affect most people) various things that aren’t and can’t be email addresses, like package-name@1.2.3.)