The Dating Brokers: An autopsy of online love
datadating.tacticaltech.org
datadating.tacticaltech.org
The whole thing was built with affiliate marketing and also engineering was increasingly driven by the marketing team and their sometimes outrageous ideas. I remember the best idea that we had was to move to recurrent billing (profits went through the roof because people forgot to cancel the service) and not long after, it was acquired by T-mobile. That's when I left since the work suddenly was all about internal API's and not really interesting for a nerd like me.
The craziest place at friendscout was the customer care. We had one student working there who did a master thesis in psychology and chose sado-masochism as subject. She was on a roll there with all the people asking crazy questions to customer care. She was kind of a student of the human condition I think and I had some very cool conversations in my time there. Friendscout wasn't targeting this (S/M) niche.
However there were plenty of users that came from all sorts of dark corners and S/M was by far not the darkest. One guy (he was internally called the shit-eater because well no need for explanation) kept coming back with countless new pseudonyms every week ("brownsauce", "smellyjelly", "sh1tinyourmouth" and other words, very hard to filter out upon registration). He was an isolated individual no doubt but kept half of engineering and all of customer care busy for weeks.
Most users were fairly normal though, albeit (looking back) vulnerable in a way that still makes me sad. At least those who engaged with the customer care.
They were men (mostly) that hardly ever had a girlfriend whose name hasn't ended in dot-jpeg or the type of girls that would "power"-date 7 guys in as many days to find their "perfect love". For many it was like tinder but without labeling it as a casual dating site. But casual hookups were probably the main use (at least for guys).
Since it was early days the product wasn't very specialized or settled into a specific vertical, that's why we had all types of crazy people mingle along "normies". (what is normal? I have no idea but I knew that place certainly didn't feel normal)
The shit my colleagues from customer care have seen though was another level. Once they compiled the craziest stories into a book (verbatim conversations from users copy pasted for the amusement of staff) and handed it out to all employees as a Xmas gift. It was funny admitted but hardly ethical to put these customer-support emails into a book and then laugh about your stupid users at a company gathering.
Thinking back, my biggest regret about the place was not leaving earlier.
After the acquisition the founders also moved on to other projects. Founders were relocating their personal assets to Switzerland and learned how to capture all of the market. They formed various shell companies (e.g. Insparx in Luxembourg) through which to bootstrap much darker and more sinister "dating" ventures. They formed a company called c-date (c = casual dating which was the German front for Insparx). This allowed them to capture the lower segments of the market while keeping their names clean and also save some cash by funneling services through the low-tax structure instead of Germany. They also pushed into the upper segments like match.com (total market domination seems the point for any company so be careful what you get into because eventually they'll pivot into new domains and bootstrap it with what they have) ... but this isn't what I'm talking about here.
According to one of my old colleagues from friendscout who moved on with the managers to build C-date (again actually Insparx), C-date was using the affiliate marketing scheme they had at friendscout but applied in this way: advertise on websites for women (fashion magazines, interior design, basically all the Grune+Jahr publishing brands - most of Grune+Jahr were also affiliates with Friendscout). The messages women saw when visiting burda.de would be something like "looking for Mr Perfect?" or some other drivel that only fashion magazine readers would be gullible enough to respond. The women would click on it and sign up in an iframe of the original (girly) site (and not have a clue that this was c-date). And men would be targeted with a more honest and different message (on xhamster, redtube, youporn ... you name it) which read: "looking for horny sluts in your area that just want to get f!cked - click here <blinky blinky text>". Absolutely fucking degrading.
Both Jan Becker (and his mentor & former friendscout CEO Andreas Etten) managed to keep their names out of controversy and am sure are well respected individuals of the Munich & Zurich tech scene. But they're until today the names behind the Luxembourg shell companies and who signed off on these predatory strategies. They (and others) deserve to be burned at the stake for tricking these female (and quite frankly male) users with these growth hacking techniques.
If you have read this from beginning to end you might have noticed that it all started out with good intentions: Bring people together based on match making rooted in science. It ended as one of the most disgusting things that I've ever come across. Meanwhile I've seen countless similar ideas go from best intention to a dumpster-fire (sharing economy LOL). Whenever somebody claims that their idea will make the world a better place, run!! Because these are the worst.
If it's stupid and it works...
I'm actually surprised Tinder and co haven't started selling an option to bid money to have your profile be seen first in your city. Same as any other real time bidding in advertising. Have users try to outbid each other to be within first x swipes in the city, instead of "potential match in position 10574" that nobody will ever swipe to.
I realize it might mess with their products' Aha! Moment of seeing only super attractive members within your first few minutes of the user journey, but there's likely a ton of money in it, and a way to strike a healthy balance. I know I would probably end up using something like that.
On an unrelated note, seeding your app with "fake users" siphoned from your peer companies sucks if you're charging for the service. You end up having your users spend hours swiping or paying money to connect with fake profiles. Can't be ethical.
This probably sounds as quaint as arranged marriages nowadays.
She was a new PhD student, and I was the departmental sysadmin. To be fair to her, she knew vi just fine, but we had these messed up DEC keyboards with no ESC, where ESC was mapped to F11.
https://stackoverflow.blog/2017/05/23/stack-overflow-helping...
Glad it wasn't the answer in your case, of course. :)
> but we had these messed up DEC keyboards with no ESC
One of these things? https://deskthority.net/wiki/File:LK201AA_top.jpg
I'd have trouble getting out of vi with that too.
Online you get to learn a bit more about them before you invest your time: do they have similar values, similar interests, compatible lifestyles and socio economics? Not all of it is written as text, a lot of it can be easily inferred from the photos, which is why Instagram is so popular these days in online dating as a lifestyle resume. Yes, it's heavily edited to show the best parts, like a resume, but you can still draw some conclusions from it. Like with hiring for your company, the mode you do it, the better you are at guesstimating what that person is like from their profiles. Several hundred first dates and your model gets trained with a lot of data.
There's also a lot about attraction and compatibility that's not only nonverbal but inarticulable. "Chemistry", goes the analogy. The club environment is stacked in favor of the attractive, but chemistry usually wins the day.
I know this probably is somewhat less in other cultures. Certainly I seen no profiles with no picture in the USA but I still see 80% or so of profiles with no useful info in them, just generics.
Not suggesting a bar is better. I'd love to find and/or develop a better system that some how encourages better matches
It's the Red Army circa 1945 approach to dating. Throw enough bodies at the problem and eventually one of them will be the one that solves it.
If you want to make fewer higher quality matches don't go to a bar.
Which, as the non-photo portion of the prevailing Swipe-To-Date format shrinks, increasingly describes the problem with dating apps.
Similar problem with Instagram.
The interesting thing is that the success of these apps/media suggests they are how people engage and apparently behave as if they actually want, regardless of whether they might theorize something else would be better in some way.
The more you use it, the more it knows your type, the more it recommends them, then it dangles out more possibilities that are just beyond reach but could be accessible if you just throw money at it.
The prospect of matching with someone you find very desirable keeps you swiping, making you more likely to do it in the future.
I'm not sure if this is necessarily unethical, but it is clearly an example of a knowledge-based power imbalance. Because I am quite sure the majority of Tinder users are unaware of all the machinations behind the curtains.
As long as the gender ratios remain similar, I don't understand how a higher amount of users results in worse dating odds. Are you referring to the paradox where the same ultra-hot 100 people get all the messages, were the site's population 100, 100k, or 100 million?
Wouldn’t it make sense for the dating site to optimize it so that the women with the least matches get to the top of the queue, so you have a better chance of a mutual match?
Showing people with less likes would make it more fair, that is right. But the effect will be that a woman will see more „less attractive“ guys and will stop using the app. And women are the users all dating apos have to keep because of the large gender inbalance on dating apps.
Turns out that users stop noticing pretty if they only see pretty. Better highlight the contrast to keep users engaged.
I have met quite a few very pleasant people but we just mutually realized "not for me" after a date or two. We wished each other luck and parted ways. In the case of bumble in my experience there is an emotional maturity and interesting people. But apparently you are required to have a photo at Machu Picchu.
I agree, it does seem ridiculous: here in the DC area, it does seem that every single professional woman over 30 has such a photo.
But strangely, I actually have a photo just like that on my profile! It was taken at the NYC Natural History museum.
For women's profiles, the stereotypical photos are: 1) Macchu Picchu 2) at a bar with an alcoholic drink in-hand, frequently looking drunk 3) at a beach 4) in bed, cuddling with a big smelly dog 5) at a gun range These are pretty typical of 30+ white women in the DC area, though they're never seen all in the same profile (the women who like guns don't go to Macchu Picchu).
Also, you say that in your experience on Bumble, women are more emotionally mature. In my experience there, I'd say there seem to be more professional women there, but if you're not a big bar-goer and drinker, you're not going to do well there, at least in this city. Also, the women there seem to be almost all white and black; all the Asians seem to be on CMB, which is where I get the vast majority of my dates.
That's why you'd see so many great people and no match, they all left long ago. It's ghost town.
You probably get news matches when people sign up or come back, dating is very seasonal.
So what are the best seasons for it? I'm guessing December, though my recent experience has been somewhat the opposite.
Now we have AI and intelligent machines. I'm sure onine dating does a better job matching people up. (/s for anyone who is humour impaired)
Actually, what works, is putting sweaty tee shirts in a bag, and letting women sniff. Turns out that how you smell is a pretty good indication of how likely it is, you can form an effective match.
So on that score, the first online love broker who manages to capture digital nose moments, probably wins.
Right now, I'd welcome good stats on how much better online relationship brokering does than other models. If you rank by income, you probably get more successful matches for rich people seeking partners for high value service, but there might be surprises: It wouldn't surprise me if down in the dollar-shop window, people do well, but better than random?
For a dating service company not to be overly incentivized to keep users swiping forever, a membership model like with a country club could also work: a large upfront fee to join and then maybe a smaller reoccurring fee to maintain membership.
There is no good old idea, but that somebody else didn't think of it first. I bet somewhere out there is a statistician from the 1940s with a box of those cards with holes and notches you used as a DB "join" method over fields saying "...have I got a story for you..." (I used them in school btw. I had a real moment in DB class learning JOIN and thinking back to how we did selection over multiple criteria from these cards)
It doesn't mean there weren't any in-group dynamics or discrimination though. Race simply wasn't the focus. People would discriminate based on things like religion instead.
Fortunately we don't have to stay at that level and can over come it to an extent, but claiming it wasn't a thing in medieval and other empires is factually incorrect.
What do we mean when we talk about human races anyway? I've been thinking about this for a few minutes now and I am stumped.
Agreed. But I feel ethnicity is "used" in a way like "race".
> Race is defined through genetics.
Sure? I thought genetically race had a very weak basis, and that it was basically some aspects of bodily looks and location of genetical/blood origination? The main reason we needed ethnicity as a term, because race was "simply not there in the genes" and thus "mostly about looks" which gave it very little use in proper science.
A lot of it remains when group emirates but keeps being close knit.
That is ethicity.
I find it hard to understand how some people still cling to 1960s "there is no such thing as race" arguments while in the real world you can send a swab of spit to 23andme or similar services and get a precise breakdown of your ethnic background down to the percentage.
No, you can't. You can get bullshit geneology which you shouldn't take seriously.
https://www.bbc.co.uk/news/science-environment-21687013
> But in a public guide, published by Sense About Science, Prof David Balding and Prof Mark Thomas of University College London warn that such histories are either so general as to be "personally meaningless or they are just speculation from thin evidence".
> The scientists say that genetic profiles cannot provide accurate information about an individual's ancestry.
> They say "the genetic ancestry business uses a phenomenon well-known in other areas such as horoscopes, where general information is interpreted as being more personal than it really is".
https://www.theguardian.com/money/2017/nov/04/home-dna-kit-f...
>Ancestry’s DNA expert Mike Mulligan (93% Irish – everyone’s email at Ancestry gives their ethnicity breakdown) admits that the ethnicity percentage is a “top line” estimate derived from just a very small part of our DNA, a couple of letters long in the 3bn letters that make up our DNA, and that there are a lot of “inferences” made from the data. Precision is still a problem for DNA kits. Mulligan says that the Irish, Scots and Welsh are “almost indistinguishable from a DNA point of view”. Meanwhile, when it comes to western Europe “it is the most traipsed-about part of the planet The amount of DNA that has been blurred together is incredible”. Remarkably, DNA testers can’t really tell the difference between German and French DNA.
https://gizmodo.com/how-dna-testing-botched-my-familys-herit...
> I suspected the error might lay not in my family narrative, but in the DNA test itself. So I decided to conduct an experiment. I mailed my own spit samples to AncestryDNA, as well as to 23andMe and National Geographic. For each test I got back, the story of my genetic heritage was different—in some cases, wildly so.
etc etc.
All I'm saying is, this has nothing to do with the results of a 23andme sequencing run, which would provide details about your ancestry. Race (in the US sense) correlates pretty weakly with ancestry, e.g. all the early humans for most of human history were black.
I do think that ancestry can provide valuable insights into one's self, however I'm still not sold on the necessity or reality of race. Ancestry is not written on a person's face.
Skin color, facial structure and other commonly held visual traits of "race" don't map strongly to genetic identifiers that could form some scientifically valid definition of "race." Two black people living in the same area, sharing the same cultural and ethnic identity, can differ more genetically than either with a white immigrant.
In other words, that "black" is a race, and "white" is a race, etc, is entirely a social construct, which should be non-controversial given that these categories were created and solidified, culturally and politically, long before genetic science was even a thing.
I'm pretty sure that's false. What is true is that africans exhibit a great amount of genetic diversity, i.e. two africans will typically be more genetically different than two europeans. But the genetic distance between either of those two africans and either of those two europeans will still be even greater.
> In other words, that "black" is a race, and "white" is a race, etc, is entirely a social construct, which should be non-controversial given that these categories were created and solidified, culturally and politically, long before genetic science was even a thing.
It's a social construct to the same extent as colors are a social construct, which were formulated long before the theories of electromagnetism was even a thing.
A thing being a social construct doesn't mean it doesn't describe an underlying reality. That applies to any category invented by mankind. For instance, considering plants and animals, or even living beings and inanimate objects, as belonging to different categories is also a social construct. Because, after all, they're all just an association of atoms.
I didn't say Africans and Europeans, I said black people and white people.
A black person from Africa and a black person from elsewhere are considered the same race because of their visual similarities, regardless of their background. That's the social construct.
>A thing being a social construct doesn't mean it doesn't describe an underlying reality
It can be inaccurate enough that it isn't useful, though.
Same thing
> A black person from Africa and a black person from elsewhere are considered the same race because of their visual similarities, regardless of their background. That's the social construct.
And also perhaps because they actually are from the same race? Do you consider all the current inhabitants of the USA to be native americans?
> It can be inaccurate enough that it isn't useful, though.
That's true, but racial categories are really accurate, especially in this age of cheap DNA testing. As for usefulness, I'd say it's pretty useful in the medical field, for instance certain drugs work in some races but not in others due to racial differences in body chemistry.
As for medicine, while the efficacy of drugs can be ethnicity dependent, race is such a weak proxy for it that it often leads to mistakes when doctors operate by habit. Nothing trumps actual genetic screening (as opposed to "which Anglo-centric category do you fit best based on how you look").
Do you know how well dating services currently work for less desired groups? OKCupid has stats on their engineering blog and its not pretty
I mean, yikes.
Thank you for granting me this exercise in creativity... now please don't do this.
I once found a vulnerable API endpoint for a staffing company and ended up exporting their whole client database.
I'd be interested to know how they managed to achieve that. The waiver doesn't seem to appear in the html source.. so I'm not sure where it is coming from and it is now too late in the day for me to investigate it further..
in that context, the italics is sort of like how you might get a summary in italics at the start of a book chapter.
Just like printing in a smaller font to be environmentally friendly using a smaller display front is similarly advantageous - and don't even get me started on all-caps, you might as well be burning a mountain of coal if you type in all-caps.
/s
That's it? That's the news? Where's the meat? What am I missing?
Usually you have to go through the back door to get this stuff. Even advertising companies go further than this to anonymize you. This is a pretty blatant privacy violation.
If the headline was just "Match.com sites are sharing your profile info between them" this would smell bad but wouldn't be a big deal.
We believe that the $0.57 average revenue per user that
Match Group reported in their Q2 2018 Investor Presentation
is just a fraction of the user profile's real value.
Companies are (at least theoretically) valued based on their future potential. Just because "real value" of user profiles may not currently be getting realized, does not mean that it isn't already incorporated into the value of the stock.