Me and my friend have been discussing recently on how to build a "women only" platform & I was wondering if this can be solved with tech, it should be cheap, does not compromise identity(other than gender), fast and feasible.
Why? - A friend was looking for a flat along with a roommate on a very popular facebook group in India called Flats & flatmates, but around 85% on the posts were from brokers, and some of these posts looked very shady. She had to hand out her phone number or call the broker in order to contact the person regarding the flat.
She was very hesitant to give out her number and I understand as we Indians practically invented "bobs & vagene". Plus, she said we as developers should have solved this by now.
So we were thinking of ways someone could prevent this.
Do a realtime verification by going and meeting the person.
Drawback - Too expensive in terms of time, resources & money. Might work for large corporations. Your address is exposed here too.
Do a ID/Aadhaar verification, the Indian Social Security Network counterpart.
Drawback - Most people will not just use their ID to verify for a website/app especially if it's new & Adhaar is already so controversial due to it's security loopholes.
An "invitation only" system where current users can invite other women.
Drawback - Difficult to scale quickly, but seems the most sensible way.
How would you solve something like this?
Latest comments (6)
As far as I understand it facial recognition technology is roughly at the party-trick level, able to guess not quite as well as humans can, and we ourselves are hardly infallible. Algorithms so far have also tended to display training bias and are worse at categorizing women and people of color, to say nothing of trans and gender-nonconforming people. It's completely unacceptable for something this important, but to my point elsewhere no automated general-purpose solution is acceptable for what fundamentally isn't a technical problem.
Whilst you have a very interesting idea, I feel like once gender starts becoming involved, lines staff to become blurred. There are humans which struggle to find their own identity within themselves, and trying to create a program which does that for them could very quickly end up getting pretty heated.
Gender is difficult to govern, it's not like age where there's an absolute value which is true, it means something different for each person. For that reason, I think you should rule out anything involving computers analysing people.
Instead, try gear it more towards community moderation, such as having a referral program. Maybe see if you can force implement some kind of photo profile picture, so users can see who they're talking to, and if there's no photo then they can move on.
Your friend could use Google Voice to create disposable phone numbers, possibly?
As for automating verification -- you'll sleep easier if you don't try. You don't solve this with technology; it's a social not a technical problem. Any general-purpose "solution" will come complete with gaps, incorrect assumptions, edge cases, and more, and the consequences of those here run much deeper than in most applications.
I think option 3 because theres so many variables to consider due to current nature of the word “gender” (maybe more here in the US) but also other “trust factors”
I have a similar prototype for a ‘members only’ app, wherein the members are all invite only and referral based. There is a publicized rule where if the conduct of a member deems dismissal, their referrer is also dismissed. The only way to rejoin is of course from an existing member. The ‘referral tree’ is an ever changing data schema. The MVP uses firebase for sms 2-factor login.
This is very interesting actually. The referrer dismissal will make sure existing users won't break the system.
Thanks for the suggestion.