Thank you for contributing to this discussion. I think we can split the problem in relatively autonomous modules and then find agreement and solutions for each module.
Agreeing that gender in translations is a problem
Without consensus about the root problem, the discussion about implementation will be unnecessarily complicated. It seems that all participants in this discussion have been men using software primarily in English (not just Discourse, in general). Experiences of fluent English speaking men cannot be taken as a reference for how is the user experience for non-male users using software in languages with gender differentiation.
About half of the World population would like to be addressed as a female if possible. I didn’t count speakers, but languages with grammatical gender are more frequent than not. This seems to be true for the current list of languages available for Discourse, a clear indication of which speakers are using Discourse today.
Agreeing that it is not an urgent problem
Indeed, I am not suggesting that we should stop everything to solve this problem. Discourse probably doesn’t have that many strings subject to grammatical gender in other languages, and in the meantime translators can find workarounds (as they have probably done in the meantime). Recognizing the problem and a willingness to solve it would be a big progress in itself.
One suggestion is to accept reports of problems related with grammatical gender as bugs, while the implementation of a technical solution is agreed and implemented. Suggestions for workarounds would be welcome (i.e. adapting a translation whenever possible) but they would be considered just that, workarounds in the interim. Welcoming these reports would help understand better the problem.
Gender information and privacy
I bet we can agree that preferred gender identification would be a user preference only, used for the purpose of providing a better user interface. That information would remain private, not visible in user profile, user lists, or any public page.
Defining gender preference
Our goal is not to identify people’s gender orientation (a rather complex topic) but to simply offer a language that corresponds to the gender preference of the user. The question is not “what is your gender” but “How do you prefer to be described?” (MediaWiki), “Choose the pronoun you prefer” (Phabricator) or something along these lines.
This setting could be disabled by default, since English is the default locale and (as far as we are aware) it doesn’t have this problem in Discourse. Then specific languages could have it enabled out of the box (Hebrew, Spanish…)
This would be a preference in the user settings. I would personally love if @discobot (gender neutral by design) would politely say something like “By the way, how should I refer to you…” and the answer would define the user setting. Most users will meet discobot before looking at the user preferences, and wouldn’t it be nice that one of the first reactions of non-male Discourse users would be to see that Discobot and the forum they are just joining “gets it”?
How to fix buttons and other text strings in the UI
I believe the solution is not to have separate locales for male and female (a probable nightmare for translators) but to introduce variables in strings where gender variation happens. However, I am not a software developer, so it is easy for me to talk.
The good news is that Discourse is not the first project dealing with this problem, and recyclable solutions probably exist. MediaWiki (probably the free software product translated to more languages, supporting grammatical gender since 2010) has developer documentation for PHP and JavaScript. If you are interested, I can put you in touch with their developers specializing in internationalization.
How to fix discobot
I hit the problem of grammatical gender when translating narrative-bot/server.en.yml
aka discobot’s conversation. That is probably the file where grammatical gender problems appear in a highest density in Discourse, because it reflects a conversation between two individuals. As said, it is also important because it is a key component of Discourse’s out of the box experience.
Conclusion
So yes, fixing the entire problem is not simple, but I have no doubt that it is worthy. Discourse is a very user friendly software for civilized communities. In the past, non-male participants in web forums were hard to find (and non-male software developers too, who might have a special incentive to work on this), but things are luckily changing. As @sam mentions, very successful “forum” (aka “social network”) products are very good at handling gender diversity and (surprise surprise) their communities excel at gender diversity among their membership. Discourse would benefit from taking the contemporary products and the contemporary trends as reference, not the products and trends that were designed and thriving in the so-male-centric Internet of 10-15 years ago.