The "from" address in email is in no way authenticated. There's nothing stopping anyone from using any email address to register using this proposed scheme.

The reason services send emails to users is to verify they can receive email, which does require a password to the account at that email address.

If you already have a service, test it out and measure its impact. That's going to be way more informative than all of our educated guesses combined.

Because you loose high % of users on each hoop they need to go through. Removing friction is one of the major ux tasks. This solution would have 80% falloff if it's not something users really want and can't find anywhere else.

This is why 90+% of our users login with one click google, it's just too much to manually insert data.

mailto: links are broken for a lot of people since they were invented in the days of desktop e-mail clients, not webmail.

Not to mention that even when they work, they might open in the wrong client -- e.g. your personal email when you need to send from your work email.

And when they don't work, you have to manually copy and paste the address and then manually copy and paste the code and it's a huge drag.

Also now you have to handle people spoofing email addresses, which means you can't include users sending emails without certain security verifications.

It's better UX as well as more secure the way it's done now.

UX is way worse if they don’t have a mailto handler set up (I don’t on my work and personal laptops, for example, and on my phone it probably won’t open the email account I want to use with you).

This also relies on them having various antispam things set up in their DNS, otherwise you haven’t verified anything.

As other have said, it's not ideal, but if you are strapped for cash it's well worth considering.