Comment

Comment on Lyft now lets drivers prioritize matches with women and nonbinary riders

It is zero sum, giving preference to one group in the list requires changing preferences to another groups.

Imagine you’re at Walmart queued for checkout. You form a first-in-first-out line and the first checkout available takes you. With this women+ system there’s a person sending women+ to the express checkout (>20% of total capacity) unless there are no women in line, in which case the express serves men now. At busy times both checkouts serve customers, but the women+ lines are always at least as busy as the other lines.

Obviously the algorithm for location and time based matching with ratings is different than queuing at walmart, but the principal is the same. If you have a system where A >= B it is possible to fall on the equal region, but at low traffic times or in less ride dense areas this is objectively unequal.

It is objectively always better to be in the women+ group than outside of it.

source

Sort:hotnew top

flumph@programming.dev ⁨9⁩ ⁨months⁩ ago
You’ve changed the perspective to potential wait time for male passengers. That may be true but it doesn’t have an adverse impact on male drivers, which is what was stated in the comment I replied to.

It is objectively always better to be in the women+ group than outside of it.

Based on Ubers data, women+ are raped five times as often in ride shares. “Objectively” I bet a lot of women+ would choose “maybe a longer wait” over “5x chance of being raped”.

source
- cheery_coffee@lemmy.ca ⁨9⁩ ⁨months⁩ ago
  I’m talking about the impact on the drivers, not the waiting time of riders (though that would also become longer on average for men).
  
  I ran a simulation study of the queuing+matching system described (a variation of M/M/c queue) and there is a clear negative impact on male drivers. There are three situations: too many riders, balanced demand, and too many drivers (there’s also zero demand but we’ll ignore it).
  
  Too many riders: in this situation men and women+ perform the same because the matching rarely happens. Women+ riders match to women+ drivers 23% of the time. Fares for women+ drivers are 53% women+.
  
  Balanced demand: women+ match 23% of the time, yielding a benefit of 5-9% more fares to women+ drivers. Women+ riders match to women+ drivers ~41% of the time. Fares for women+ drivers are 80% women+.
  
  Too many drivers: Women+ always match in this situation, making 5.5x more fares than men. Women+ riders only match to women+ drivers. Fares for women+ drivers are 85% women+.
  
  The disparity can in theory go up to 8x more fares for women+, but the scenario where that happens has women always available in the system.
  
  The actual outcomes of this would vary in real life of course, and queuing theory isn’t really my thing. I assumed all women+ drivers opt in (because why wouldn’t they?) and I’m using Lyft’s own published numbers. The state of the system will oscillate between these outcomes, but in theory it should skew towards the two biased results.
  
  Now to your point: obviously women raped and financial impact to men are two entirely different things and we can’t even begin to compare them in this way. Rape and sexual violence is abhorrent and we should take actions to reduce and stop it, the question is always: which actions are reasonable and fair.
  
  This system is financially biased against men, and significantly so. It would reduce the event of sexual violence (by reducing male-women+ interactions). Is the system a fair tradeoff, I don’t know. My gut feeling is that I don’t like it.
  
  source