Onboarding

[simonv3]

I was thinking earlier today that what really set Slack apart was its onboarding, and it feels like what we're doing is still just throwing a sign up form at people. Should we maybe be a bit more radical than that?

@pdurbin pointed at this issue (openhatch/oh-mainline#711) and this subsequent PR (openhatch/oh-mainline#1617) for openhatch. I think we could definitely come up with something crafty that gets people using Slick in a fun way but also prepared for using IRC. I don't think we need to rely on a "slackbot" like entity to do this.

The main pain point might be detecting whether its the first time someone logged in to Slick on a specific server. Can we use cookies/localstorage? We could also throw up a "are you new to IRC?" question before log in.

[jancborchardt]

We could just require a password in any case. People are used to putting in email address, nickname and passwords nowadays. Sure, not needing a password is also cool but it might be confusing.

[simonv3]

I wonder whether the appeal of just showing a single input (just 'username') is versus showing a form and password. Do people sign up faster? Email newsletters for example that just show your email address. Does anyone know of any research on that? Does needing a password right off the bat create an illusion of "signing up to something" that might be a detractor?

There's a lot to be said for getting people to just use your interface without any hurdles (username is the bare minimum here) and then getting them to commit more further down the line. Slack's slackbot is a great example of that. As is mailing lists that ask for confirmation and more details later.

[matbee-eth]

You can avoid even entering a username, and generate a random one. Then just make it obvious how to select your own username.

[raucao]

@simonv3 Just using the username as first step and then handling NickServ if it's registered from there on is exactly how I want to implement it in Kosmos. Got the idea from momentumdash. See my Hyperchannel issue here: 67P/hyperchannel#39