[thu2111]

As someone who has worked on the front line of the fight against phishing and account takeover in the past, I can assure you and others that you're dead wrong. Making this change was a recommendation I made to the Chrome team years ago because the number of people who would reliably type in their username and password to a site hosted on hacked web servers (supershop.co.hk/account_login.php etc) was just so high. And when those accounts got hacked, scamming and sometimes even extortion would follow.

Your side view mirror metaphor is unfortunately not clear at all. The side view mirror is simple and performs its function correctly as designed. It can't really be improved without totally replacing it with something else like a camera. Now of course not everyone will use the URL bar even if it's redesigned to work correctly. But right now the bar is practically designed to look as intimidating and useless as possible.

Perhaps you're so used to parsing URLs in your head you don't realise it, but URLs are a baroque and absurd design that nobody without training could properly figure out. It's basically random bits of webapp memory and protocols splatted onto the screen in a large variety of different encodings. In a desktop app dumping RAM straight onto the screen would be considered a severe bug. On the web it's tolerated for no good reason beyond history.

To give just one example that has regularly confused people in the past: URLs are read left to right except for the domain name (the important part) which is read right to left. You don't stop reading a domain name at .com, you stop reading it at the third slash or possibly a colon, but that form is rare.

[franga2000]

As someone who has had to teach grumpy old high school teachers how to not fall for phishing and mitm attacks, I really can't see the problem here.

The way I used to teach was very simple and very effective: there are 3 parts to a URL - the first part tells you if the connection is secure, the second part tells you who you're connected to and the third part tells you where on that site you are. The first part needs to be httpS, the second part needs to be the site you're expecting and the third you can ignore. They're even shaded differently to make it easier to read. "If you're going to Google and the black part ends with anything but google.com, call IT" made sense to even the oldest and most reluctant people I've had to deal with. The problem was actually getting them to check every time and not forget.

It seems to me that this change will not help people without training, change nothing for people with training, and make sharing links even more confusing for everyone.

Are you saying someone is less likely to get phished on "supershop.co.hk" than on "http://supershop.co.hk/account_login.php", even where the http:// part is replaced with a red padlock and /... is grayed out?

I see only one real solution to phishing: don't let users type passwords manually. WebAuthN and password managers both automatically read the domain and won't try to authenticate on a domain that isn't a perfect match. I've had more success with that than any other anti-phishing measure I've tried deploying (history-based domain trust, explicit trust on first use popup, detecting unicode gaps and domains in credential fields...).

[thu2111]

Sure, absolutely. People understand domain names, they're found on billboards, adverts, business cards, all over the place. And it's a simple text match. Does the bar say "google.com" or "google.co.uk"? Yes? Then you're on Google. So when it's simple people get used to checking and can be reasonably told they're expected to do it.

The greying out and replacement of padlocks etc, the anti-phishing training, it's all just working around a historical design problem in browsers. There's no need for it to exist. Notably, mobile apps don't have this problem.