> You might provide answers for how you feel that night, which may not be reflective of your larger perspective. [...] So that’s one of the big challenges: understanding what someone is really trying to say when they’re answering questions about their preferences.
> There could be two different interpretations for the question, and you just answered one of them. [...] Knowing this can happen, we use the algorithms to help us understand the statistics behind each question, and we’ll try to identify questions that are the most likely to be mistaken in this way so that we can remove them.
These are both issues with the question system that users have raised concerns about, and generally assume that the question-matching data is just going to be messed up by these. So, to learn that they were actually addressing these issues in code with statistics and analyses, makes the whole system seem a lot more sensible.
He also talks a lot about ethics and fairness and grassroots ideals - and with OkC it did seem like those were more than just corporate talk, they realy were a user-focused, transparent company for a long time. He seems to have left before the Match.com acquisition (I'm assuming) though, and unfortunately there have been complaints after that that the company is gradually moving away from those ideals.
OkCupid in their OkTrends blog where they discussed interesting findings on their data once had a post about how the data showed that paid online dating was a losing proposition for the customer. That post was quietly taken down after the match.com acquisition.
It seems to be in mid-jump. There were two major changes recently: mandatory real first names and you now need to "like" someome before you can write a message. This like generated a notification for the other user, but with the liking user anonymized for non-paying user. The latter change seems to be designed to create more pressure to pay.
“Likes” have always been anonymized for non-paying users. You’re right that this shift will cause more notifications, but they made other changes to messaging that I think make the service better for its popular users (mostly women).
The way it works now is that a message you send to someone who hasn’t liked you back appears on your profile page (if they they view it) instead of their inbox. In addition, I believe once you’ve sent a message, that user’s profile is hidden from you unless they choose to respond. So for the receiving user, they don’t have to take any action to pass on a mountain of “sup girl”s, and their inbox will only contain the people they’re actively messaging.
I don’t know if there’s an indicator that someone has sent you a message which you can see from the profile thumbnails, but if there is it wouldn’t surprise me if that’s for paying users only. They could have skipped the “like” requirement and just implemented the other messaging changes, but I think it’s a pretty good balance of free vs paid features overall.
It's recently changed a handful of features (messaging, swiping) which has felt extremely anti-user. Finding out they've been acquired, makes the recent changes make sense.
>But in general, we focus on making it an experience that doesn’t discriminate and encourages people to be their best selves.
Well, you failed miserably. Not just OkCupid but online dating in general.
I think it is one of the few aspects of 21st century life where the Internet/technology made us more close minded instead of more open minded. If you are at a party and meet five different people, you will immediately notice something that you don't like about them, but you will give them the benefit of doubt and engage them in conversation for at least five minutes and allow them the chance to show you their best selves.
To add to your point, I’ll say this: as a Asian guy I’ve interacted with hundreds of non-Asian female profiles and not a single one had mutual interest in me. Not one. And I don’t say this because I only date non-Asians so much as I don’t care about a persons ethnicity, a belief that isn’t shared by others online.
Meanwhile in real life I’ve gotten to know and date extremely successful, smart, non-white women. Why? Because by having conversations with them in real life prior to moving onto the next stage, I likely broke down generalizations about Asian men and got to show my best self.
One of my best friends who is white and extremely successful (successfully went through the top medical programs) once said to me she never dates Asian guys from online sites because she doesn’t know if they’ll speak fluent English.
One of my good looking Asian male friends has made the similar observations to me. He’s only met any of his SO’s from parties and get togethers because people online treat him differently.
I've talked to all sorts of people at parties... mostly it was small talk, or talk about something I wasn't really interested in, and very occasionally about some mutual interest.
Odds are that many of these people actually shared one or more interest with me, but neither of us had any idea because we didn't know anything about each other at the parties, so never talked about it, and so never found out we might have something in common and so might be compatible. Thinking of all these wasted opportunities makes me sad.
With online dating, by the time I meet with someone, we'll already know that we have a bunch of things in common, and already know what many of those things are. So there's a lot less guessing and blindly stumbling around trying to find something of interest to both to talk about. We can cut to the chase and immediately start talking about our (hopefully many) common interests.
That's not true. Online is not infinite. The number of matches I had went from a dozen a week to maybe one every couple of weeks. Lowered my standards every day too. I live in Seattle though. Other cities might have more fish. :*(
This paradox/abundance of choice is artificial though. Dating apps could throw up a counter of the total number of people you will be permitted to interact with over a window of time, which creates scarcity that forces the user to increase their efforts accordingly.
It’s entirely possible to encourage positive social behavior with technology.
Tinder does this to some extent by limiting how many "right swipes" a non-paying user has every 12 hours. I don't know if that is enough of a scarcity to really change any behavior though.
At the very least it discourages the strategy of blindly swiping right as fast as possible and then choosing a mate from the intersection of "those who also swiped right on you" and "people you'd ever consider dating".
Would that make sense to implement from the perspective of a dating app though?
Users could just move to another platform without artificial scarcity and... stay there, because there is no scarcity.
I'm not sure that coercing people to settle is positive behavior, but regardless, enforcing artificial scarcity isn't a feasible strategy for a dating app.
Everyone settles, realistically speaking, as you’re optimizing for a local maximum when dating, not a global maximum.
You’re equating more choices with better outcomes. I argue that isn’t the case. Unless we’re going to feed every data point we can into a ML black box, and out will come your arranged ideal partnership (there’s even a black mirror episode about this).
Getting to know people sounds more fun, but I’m old shrug
What you're saying is true, I just think users would feel they're being forced to settle if their choices are limited.
To be fair though, if people could be convinced to buy into the platform with its restrictions, it could make for an interesting experience.
While Tinder profiles are built for speed, with only a few pictures and blurbs, profiles on your platform could have much more detailed biographies, videos, etc...
It would be a more immersive experience.
If done right I think your idea does have potential.
Well most dating apps suffer from the problem of actual scarcity, in which they don't reach a critical mass of people to make the effort of making a profile and actively searching worthwhile. Tinder is one of the few that has gone "mainstream" and has reached that point.
> This paradox/abundance of choice is artificial though. Dating apps could throw up a counter of the total number of people you will be permitted to interact with over a window of time, which creates scarcity that forces the user to increase their efforts accordingly.
The problem is that's not what causes the problem.
What happens is that different people are different levels of desirable and everyone messages the highly desirable ones first. Then those women ignore most of the flood of messages they got, even the ones from other highly desirable people, because they don't have time for so many messages.
So then all the senders, dejected, realize the majority of messages won't get a response and become less selective in who they send them to. So then the second most desirable set of partners get flooded with messages and it repeats on down the line.
Which creates the illusion of choice, because even if you're a "6" you'll get messages from many people, including an "8" or three, but you can't actually have them. They only messaged you because they messaged everybody. So then the 6 wastes their time responding to the 8 instead of the other 6s they actually had a chance with, who then get no responses from other 6s.
If you limited the number of messages then everyone would only message the most desirable partners. So the 10s would still be flooded and not respond and the 6s would go from having too many messages from unavailable senders to having zero messages from anyone.
Some sites do the "user's mailbox is full" thing to try to avoid this, but that doesn't really fix it either. It just accelerates the problem because then the most desirable partners all have full mailboxes, so the second most desirable partners get flooded with messages even sooner.
What you need is some way to match people with the people they actually have a chance with (and who actually have a chance with them). But that's easier said than done.
I think that’s the crux of the problem. In the real world, 6s do have a chance with 8s, 9s, 10s. In dating apps, because of the scattershot approach, they don’t.
“At scale” is f*ucking with how humans operate, and not for the better.
> In the real world, 6s do have a chance with 8s, 9s, 10s. In dating apps, because of the scattershot approach, they don’t.
It's not that they never do, it's that they on average don't, in either case. The average person will choose an 8 over a 6, and the average 8 will have that choice. And I mean it as an overall rating including things like intelligence and wealth, not just physical appearance. It's not that a 6 can get an 8 because she's interesting, it's that being interesting makes her an 8 too.
So interactions between 6s and 8s (or, especially, 3s and 8s) are much more likely to be a waste of everyone's time than interactions between 6s and 6s, but that's what dating websites cause to be common.
Interesting. I often ponder how to approach creating a training set for a match making AI. It seems like the most difficult part is getting long term feedback. Analytics on first date compatibility would be easy, but metrics on long term relationship success would be very difficult.
> And in fact, for a very long time we resisted allowing people to filter by race—we felt it just wasn’t appropriate.
> But then we learned about some use-cases from the other side—someone who is Filipino who wants to find other Filipinos easily. We found that that’s a pretty legit reason to search by race, so we added that feature.
I don't understand. What other side? And why are Filipinos special? How is that fundamentally different from any other race-based selection?
I can't tell if you're playing devil's advocate here. I'll translate what the CTO was saying;
"And in fact, for a very long time we resisted allowing people to filter by race—we felt it just wasn’t appropriate." --> A significant portion of our users wanted to be able to filter by race, but we were afraid it'd create an internet shitstorm.
"But then we learned about some use-cases from the other side—" --> So here's the semi-PC thing we invented to cover our ass, just to give our users the feature they wanted anyways without being the target of a social activism campaign.
Thank you. I wasn't playing devil's advocate. I am not from the US, and the specific hangups and dos and donts Americans have around the issue of race are a bit alien to me.
(That's not to say I am free of hangups, but mine are different)
Yeah, they could have allowed filtering by language like Tagalog (something learned) rather than race or "nationality", which would have been just as easy. The other place I can see them doing this is Brazil, and there I'd admit that European vs South American Portuguese would be different. People like to segregate though...
By other side he meant they didn't want to give people the tool to remove certain races from the results. I know that logically this is the same exact thing but from a PR angle it's not.
In Las Vegas I knew several Filipinos. Most were immigrants but also native born as well. They have a local subculture akin to any of the others that employ specific dating sites.
I guess. But it seems like a lot of pretending. Maybe I'm too cynical, but men have a pretty basic algorithm for choosing a mate:
1. Is she acceptably attractive?
2. Is she acceptably not crazy?
Match those two things and you are 99 percent of the way to the chapel. Women have more complicated criteria mainly because what they really want (someone at the top of social pyramid) is ineluctably scarce.
Well, let's say that those factors are the most immediate ones for attraction. This doesn't translate much to "the chapel".
I've been thinking a bit about online dating (heh) and dating in general. My latest idea is that a successful match is not much a matter of state, but rather a matter of process. Two people might be well suited for each other but to get beyond the most superficial attraction you need time spent together and meaningful experiences.
They could create a dating website that encourages "matches" to go through a set of random activities, that would be fun (and, I guess, successful).
...this is called 'speed dating' or 'club 18-30' where a drink-related task is set and the interaction goes from there. I think they do meaningful walking holiday things for the more mature folks and participants get a reasonable chance of finding someone through this process.
Lucky punters might have to go to Ibiza as part of this ritual to meet the guy from five miles up the road and back in their home town they do not live their life on a daily basis like how they spent their holiday in Ibiza that time.
There may not be messaging online but there can be clumsy 'my friend fancies you' messaging not attempted since early teenage years.
In these organised events it is clear that everyone is there to find a partner, so there is no time spent chasing those that are not in the same 'available state', it is straight onto 'process'. I don't believe the numbers get better, if real world online organised dating was that good there would be more of it than there is, I think it peaked before everyone had smartphones.
The problem you have to solve is convincing people to invest time in a stranger. And forgo safety. After a few online dates people learn that writing skills don't always translate to personality. Stunning photos don't mean stunning in person.
The safety aspect can be managed (much as it is already done in online dating). But yes, you definitely have to invest a bit of time- it might be worth it if it works better than tens of casual encounters that go nowhere.
It's rather exaggerated, but it's true that men are immediately attracted by looks and women by social status (or promise thereof). And that the craziness factor is a turn off, for both.
It's also true that we tend to judge rather harshly those who don't appear to have an interest that goes beyond these immediate and basic factors.
I don't know if women value social status as much as they did twenty years ago. And men now care more about earning power of their partners.. In tech, I've met a lot of co-workers with husbands in less prestigious jobs.
I think it's more like "[a lot of] men [you know] have a pretty basic algorithm for choosing a mate". That's an OK thing to say (because it may be true!) but it's a pretty bad postulate to assume casually while engineering a human system (because it has clear drawbacks and isn't obviously inevitable). WLOG, the same criticisms hold for what you have to say about women.
> You might provide answers for how you feel that night, which may not be reflective of your larger perspective. [...] So that’s one of the big challenges: understanding what someone is really trying to say when they’re answering questions about their preferences.
> There could be two different interpretations for the question, and you just answered one of them. [...] Knowing this can happen, we use the algorithms to help us understand the statistics behind each question, and we’ll try to identify questions that are the most likely to be mistaken in this way so that we can remove them.
These are both issues with the question system that users have raised concerns about, and generally assume that the question-matching data is just going to be messed up by these. So, to learn that they were actually addressing these issues in code with statistics and analyses, makes the whole system seem a lot more sensible.
He also talks a lot about ethics and fairness and grassroots ideals - and with OkC it did seem like those were more than just corporate talk, they realy were a user-focused, transparent company for a long time. He seems to have left before the Match.com acquisition (I'm assuming) though, and unfortunately there have been complaints after that that the company is gradually moving away from those ideals.