I've got a number of tiny projects. Realistically if they're not hosted on GitHub, they won't get contributions. "Setup another login on my gitlab instance" would close to ensure nobody collaborates on it. This allows me to both disconnect them from a centralised service and remove the "you need a separate identity" step.
Oh come on. It's not like making an account is overly difficult or demanding. You are not required to send two forms of ID or a blood sample by mail.
Anyone who really wants to contribute to the project (and scratch their itch) will not have a problem with entering a username, password, and perhaps use a validation link from an e-mail. It might discourage people with just fleeting interest.
And as a hypothetical FOSS project maintainer, I know I'd rather have one of the first kind of contributor than ten of the second kind. Especially if the alternative is to be beholden to the likes of Microsoft.
Of course, the best solution would be something like ForgeFed, but I can't see that meaningfully taking off (there's no way Github will ever adopt it). I'd love to be wrong, though.
The step is not difficult. The step exists and that's enough. Search through HN comments and see how many people see creating a new account as annoying.
And I understand it - If I have a one line fix but the relevant system requires a new account, I don't bother. If it's a significant fix, but the project requires extra signed document (CLA?), they're getting a description in an issue instead.
There's a reason for a recent trend where new services let you jump right in and start using them. Then require amount creation only when you do something significant.
And for a small project that potential drive-by fix is valuable. It's not like I get to throw away most people to select the best ones. I get one potential fix a year - either it's a seamless experience or they disappear.
