The screen you get when the link in your invitation email is invalid mentions an “I forgot my password” control (button?) that doesn’t exist.
(You probably have to click on log in to see it.)
Suggested improvement:
Either include the forgot password button on this screen and keep copy the same (getter) or change the copy to provide a link to where the person can actually do something about this.
As is, the next step is not clear.
Not really a fan of the slippery slope we’re going down on this one, but OK, I changed it to:
Sorry, that password change link is too old. Select the Log In button and use ‘I forgot my password’ to get a new link.
Shall we also draw an arrow toward the upper right and describe how to move the cursor or finger toward the target? 
Hey Jeff,
I was thinking more that it could reduce the person’s load. Currently (off the top of my head), there seems to be four steps involved when the person is at this point:
Steps: 4.
Quick brainstorm on whether we can improve the experience:
Do we know the user’s email address from the link? If so:
“Sorry, the password change link you followed is too old. [Request a new one.]”
1: Person clicks the button and a new link is emailed to them.
Steps: 1 (25% of the effort)*
If we do not know the person’s email address from the link:
“Sorry, the password change link you followed is too old.
[Enter email address] [Request a new one.]”
Steps: 2 (50% of the effort)
Of course, this may seem like nitpicking — what’s 3 extra clicks and 30 second extra time in the grand scheme of things? — until you think about someone who uses their computer with just one switch control. In such an instance, we could be saving them minutes of work. (And even the seconds can add up to minutes, hours, days and weeks over the course of every feature in an app, and every app in a person’s life.) 
*Conservative estimate based on every step having the same effort. Typing out an email address, of course, is far more effort than clicking a button. Each letter can be compared to a button click. In which case, the first scenario (where the person does not have to type in their email address) would have far less effort than the others. (Again the switch control user is a good one to think about in these situations.)
Sure, it’s just a question of engineering time vs. frequency. I don’t think users run into this expired password link scenario that often, so text telling them to log in again and click “forgot password” isn’t super onerous.
I don’t oppose the changes you are describing, and we’re also open source, so if anyone wants to hop in and do the engineering work, feel free!
Of course, totally get that.
OK, putting “set up Discourse dev environment” on the todo list for after our pre-alpha release
