It's funny... doing all this 'hacking' to create profiles with awesome match %'s... is pretty pointless. In my personal experience, a match below 80% is a red flag (serious incompatibilities), but above 85% there aren't any big differences.
Like he said, he went on 55 dates, but only three second dates. The 55 dates really isn't too hard to do without the hacking, it's just a question of time. And the "three second dates" means his filter wasn't even that great -- he probably would have done better just simply browsing on the site, and only directly messaging the girls he found interesting in the first place.
But the real interesting thing here is the clustering into 7 types of women -- that's fantastic! I'd love to read more about that -- if he could write it up in a blog, OkTrends-style, I feel like it could get a huge number of hits. I think tons of people, including myself, would be interested in the details, especially if he did it for both men and women.
I spent about one year on OkC @ approx 40minutes/day on the site, and got about 20 first dates, 10 second dates, 2 third dates, and 2 fourth dates. They were almost all 85+% matches, and weren't that hard to find. However, I would say that my manual filtering probably took way less time than he spent implementing his models.
He probably learned more engineering and advanced math while I learned how to read subtle messages of profiles, project the right ones of my own and know what's worth talking about ahead of time by having conversation online first.
He probably got the satisfaction of 'hacking' the system, while I had more efficient expenditure of money (dates can get expensive on average for guys when you decide to at least offer to pay).
None of my OkC prospects looked like they would work out long-term, though I'm still good online friends with over half of them. I flippantly shut down my account 3 months before I would have to leave the country, because even if I met someone I didn't want to be in a long-distance relationship, so I decided to save myself the trouble.
And then met a girl the very next day (not kidding) that I really hit it off with when I wasn't looking, and we're planning to go traveling in Asia next month (it's been 7 months).
I'm a diehard hacker and nerd and all, but when things like this happen, it's hard to not wonder if the traditionalists do have a point when saying you can't figure these things out with numbers. At least not when people are gaming their numbers :)
The math is super-fascinating though, and I hope OkC team does some research into it and integrates some of the ideas: obviously they wouldn't want to support uber-profile optimization for multiple groups, but maybe help find the right groups, etc.
I do think his biggest win as a male on OkC was being able to generate all the inbound traffic he would want with no invested marketing on his part (besides writing his scripts). That's something even the most skilled male OkC connoisseurs find difficult to do.
> The 55 dates really isn't too hard to do without the hacking, it's just a question of time.
That I disagree with, 4 years and hundreds upon hundreds of hours sunk into OKC, and I've had maybe 2 dozen dates from it.
Knowing how to optimize the site would be very useful. From pictures (OKC has MyBestFace that helps with that) to profile text. This guy obviously had good pictures (great pictures alone can make a profile, for either male or females), a decent profile, and the ability to write good opening messages.
Without at least 2 out of 3 of those, online dating becomes nearly impossible.
I thought HN of all places would celebrate this - a growth hack by a down and out mathematician. He just increased his inbound funnel by many multiples. He's now "killing it" and has been acquihired by his target. ;)
Well, he did it the standard way but was only able to go on 6 first dates in 9 months of trying. By hacking the match % he was able to have more women view him and then message him and go on 55 dates. Article also states that when he messaged a woman it took 3-5 messages to get to where he could set up a date but when the women messaged him first it was as easy as a single reply to begin setting up a date.
"The 55 dates really isn't too hard to do without the hacking"
I think it depends where you live. In some places you can have as many dates as you want, in others, it's even difficult to have someone read the messages you sent. I experienced both.
"he probably would have done better just simply browsing on the site, and only directly messaging the girls he found interesting in the first place."
He'd already tried that, the article claimed. The problem is that he didn't know what was important to the women he was interested in, and so didn't know what parts of his life to mention in the OKC profile, to attract first dates from women he, as you note, might well have messaged anyway.
I think the main point here is that one has to enable his luck. I mean, meeting somebody you'd fall in love with is luck. You can not use data for that. However, you could use data to enable yourself to meet enough people so that the luck had chance to happen. Of course, it is not the only way - it's just one of the ways. And, I guess, a way that appeals to the HN crowd :)
Agreed that the clusters is interesting! Having bots visit girls with high match percentage and thereby direct in-bound messages from girls seems huge as well. Seems like girls are more likely to message you if you have such a high match. He definitely drove a ton of first-dates with minimal per-unit effort. Your point about the conversion rate of 2nd-date is still valid.
I've been off the site for many, many years now, but I could find a difference between 85% (which I would have considered low), 90% and 95% in most cases, but what I found worst was that I couldn't weight major red flags enough so even a 97-98 could have one or more.
> It was first date number 88. A second date followed, then a third. After two weeks they both suspended their OkCupid accounts.
You know... there's a damn big chance you find someone worth having a relationship after 88 dates. Something tells me his technique was no better than just dating at random.
I think you are missing the point of his strategy. Before this he had 6 dates in nine months, his primary problem was the top of the funnel. I don't think this strategy helped him improve quality, but it did lead to quantity which at the end of the day got him this result. And I think with online dating quantity matters as the signals you are using to select are so poor.
"First dates" in the context of this article sound more like "approaches" in real life, rather than actual dates. In real life dating, you'd go up to someone first, chat for a bit, then ask for an actual date. OKCupid removes the physical/verbal interaction and visual cues (along with body language), which are really important to most people.
I wonder if the next generation of dating sites will match people based on facial expressions and body language (captured by a Kinect, for example), instead of all these stupid quiz questions.
Or, let's admit it, he sucked in real life human relationships. I mean we all sucked in first X dates, just, hopefully, took less than 88 to realize that whatever you are doing is just wrong :)
I did something similar 4 years ago. My OkCupid profile was receiving about 3-10 visits/week from women, and I had only a small handful that I matched 90+% with. I deleted all my answers and answered just the minimum required with complete honesty but zero possibility of controversy. My match % was 95+% with everyone and my inbound views turned into 100+/week.
The match percentage was useless as a filter, but who cares? The new filter was my profile, and women who liked it messaged me.
Within a few weeks I'd been on several dates — I'm now married to the last woman I dated from then. She messaged me.
I'll give the guy credit for coming up with a creative idea [1], but for a guy who was overly concerned with efficiency, going on 55 first dates and only 3 second dates strikes me as absolutely insane and a waste of time. It would be similar to having 55 in-person interviews and only making 3 offers. Something is going wrong in the funnel.
I don't see how his system was better than just using the site as it is intended, nor do I think it should be romanticized as much as it is in this article.
[1]: Though it fails the categorical imperative. If everybody did this okcupid would be much worse off.
Those numbers didn't surprise me - in my online dating marathon, not aided by big data, I had roughly 100 first dates and less than a dozen second dates, before meeting my now-wife (via okcupid).
Hard to say the process didn't improve things for him, though - maybe he's distinct enough that compatibility for him is more rare than average.
This is a little bit off topic, but although I agree that [1] does fail the categorical imperative, it isn't because "If everybody did this okcupid would be much worse off."
The categorical imperative is not a consequentialist motivation. It doesn't say, "Don't do X if doing so would make Y worse off."
Also, here's my .02 re-posted from the comments section:
Rather than answer the questions that were important to him he decided to find a set of people he thought he would like then only answer the questions they care about, and not even the way he would naturally answer them, rather he used an algorithm to determine the weight that would be best to get the highest match %. The fact of the matter is he could've spent a fraction of the time just answering all the questions honestly and with his honest weights and he would've found high matches too. Furthermore, he could've narrowed it down to just the kind of people he wanted through a normal search and then filter their questions based on what's important to them (which is a normal question filter on OkCupid).
So in fact what he did was pretty bad, violates OkCupid's TOS in numerous ways and at the end of the day wasn't honest to himself as he created specific profiles for his targets.
Honestly, we should not be celebrating this.
To recap what he did:
1. Didn't want to answer questions, so let's find all the questions that are important to everybody from the categories of people he thinks he likes based on clustering and then browsing a profile or two of people in that cluster. (He did this by creating numerous fake profiles and having those bots answer all the questions so he could scrape his targets question)
3. Create specific profile for his targeted group. With words and information that he knew they would like.
2. Answer ONLY the set of questions deemed important to those people. He answered these with weights determined by an algorithm that determined the best weight to achieve the highest match% rather than honestly.
3. WIth new found 99% matches go on dates with these people and follow normal dating process.
Now that we see the above broken down we can see that it's really not good. In fact, he was only answering what they wanted, and created profiles for them. But he wasn't being honest with himself or with his answers. If we're trying to match with everyone, which is essentially what he did, it's not that difficult to do. The fact that he eventually found someone is great, but the information used was faulty. Obviously there's no way he would be 99% with that many people normally.
I celebrate this. Two people found love and are getting married. I think it's great. He answered all questions truthfully but let the code determine the weight of each answer. It's not like he wasn't giving honest answers. And yes, he created two profiles, but then deleted one. He wasn't sure which group of women he'd match up with best. I guess you see this as fundamentally dishonest but I think the couple says it best:
“People are much more complicated than their profiles,” she says. “So the way we met was kind of superficial, but everything that happened after is not superficial at all. It’s been cultivated through a lot of work.”
“It’s not like, we matched and therefore we have a great relationship,” McKinlay agrees. “It was just a mechanism to put us in the same room."
You're right that under the assumption that the OkCupid algorithms actually work, the best strategy is to just answer the questions honestly. If they don't really work all that well, just gaming it to give good results would be a rational strategy.
Though unrelated to this post in particular, I figured I would post a method of finding dates that worked particularly well for me.
Post a "Blind Date" message in the M4W section of CL. In the message describe your self as honestly as you can, while still being interesting and flattering to yourself. Ask the women to describe themselves to you in the reply. Say one or two interesting things about your self, and what you are looking for. Request that the responding girl does NOT send you a picture, and wait for the messages to roll in :)
I picked up quite a few dates that way, all the girls were beautiful, smart, and very interesting to talk to. Because we weren't a "100%" match, we actually had some different points of view, which lead to fun conversations.
You might think this would lead to you perhaps going out with girls who are not very good looking. First of all, you can have lot's of fun with a girl, even if you are not sexually attracted to her. But in reality, only girls who are very beautiful and confident in their appearance would actually reply to this message.
In any case, it worked great for me. I met lot's of cool girls, and eventually found the love of my life.
"Amy Webb was having no luck with online dating. The dates she liked didn't write her back, and her own profile attracted crickets (and worse). So, as any fan of data would do: she started making a spreadsheet. Hear the story of how she went on to hack her online dating life -- with frustrating, funny and life-changing results."
Man... I kept listening to her go on and I couldn't help but to think "I'm picky, but why am I so lonely." "I don't want to do anything, its everyone else's fault."
OkCupid is how my wife and I met. She was one of two women I exchanged messages with, and the only one I met in person. We've been married for 5 years and have had two children.
I have no idea what our match percentage was, and there were a few things in my profile that were turn-offs for her, all of which appeared in my only picture on the site: Me, with a fresh buzz-cut, jogging up a hill with my dog. She's allergic to dogs, doesn't run because of her asthma, and prefers long hair.
All of that was superficial, and she was able to look past it. She engaged me mainly because of the descriptive content in my profile. I just went nuts explaining who I was, in a chatty, stream of consciousness manner.
In the end, I re-homed my dog with her own parents, and let my hair grow to 21", which I started growing out again after we'd been married for a year. That wasn't all her; I had hair that length in high school and chopped it off to help me stay employed.
What's my point? The content matters most. You can optimize your approach to searching for matches, and you can go on lots of dates, but you can't force a good real-world match. If it's there, you'll know. No mathematical model of searching, nor red-pill-esque approach to building self confidence will be more effective than an open exchange of ideas between a couple. Get to the messaging. Give her a chance to be disinterested, because that's a hurdle you'll need to cross at some point in the relationship... assuming, that is, that you're looking for love.
What's interesting to me is how this guy inverted the gender roles on OkCupid. Normally (says OkTrends data) men take the initiative, and women filter incoming messages. In the later stages of this setup, the guy set up bots to get himself in the "inboxes" (list of profile views) of lots of women, getting enough incoming messages that he could take the normally female role in this online dating dance.
At least one useful take-away: k-modes clustering. I initially thought it was a typo/miscommunication, but the soybean reference seemed bizarrely specific. It apparently actually exists, and is an extension of k-means to categorical data, using Hamming distance instead of Euclidean distance.
i did something like it 6-7 years ago with friendster when it got 'who looked at me' new feature then. i scrapped million ids of teenagers to below-age-30 women. my friend list went from below hundred to full 3000 (the max at the time) in a week or two. facebook was not popular in my country (not usa)
the scrapper was written in newlisp (save search result pages with curl, use regex to match and collect the ids). it's probably easier to write in other languages, but that's what i knew.
i used wget and curl to loop over the ids but it's too slow because they download the whole page. later i found out about 'curl -i' (header only) and a million ids was done in about hour or two (i moved my operation from my home's 64kbps to my colo datacenter mbps internet).
my account is no longer exist (probably banned); however, i do still have a screenshot of me having 3000 female-only-friends and 70000 non-hidden-females 'look back at my profile'
initially, i talked to any interesting woman; however, later i made a strict rule to only respond to women who wrote at me. there were just too many fake female-accounts.
i got a couple of dates from this feat; however, i met my wife in a traditional catholic youth retreat. when i let her know about this friendster thingy, she just laughed. now i'm happily married with a 15-months-old boy.
I am feeling sorry for all the women subjected to all that spam, not only by this guy. And talking about this guy, let's be honest, he was different from other "spam" contacts by implementing a bot-assisted spamming, nothing he should be praised for.
Math genius? More a spambot writer, but not for money, for an advantage.
Everyone has interesting talents that can attract mates. And most people can look good if they dress well and develop proper grooming habits. Cheer up.
I think the most interesting part of this is that an algorithm told people they are 99% compatible and that changed their behavior enough to make it easy for them to go on a date. Funny that an artificially high number on a computer screen completely changed people's behavior and perception of the world.
I've dated around on OKC since the early days. His lack of success on first dates points to a problem with him; maybe he was too picky or had some other issues which turned women off. If he was looking for that Hollywood-style instant spark, he was mistaken; that doesn't always happen.
When I was on OKC, about 1 in 20 messages would result in a first date; but you bet that more than 50% of first dates turned into second dates.
When women first meet you, it's almost like they're going through a checklist in their heads: is this guy a creep? A rapist? A jerk? etc. etc. (this is just the impression I got). If you don't trigger any of the alerts, you're golden.
One of the biggest mistakes nerds (like myself) make on their profiles and on dates is that we try to impress the woman with our encyclopedic knowledge of some esoteric subject. That's a sure turnoff.
Maybe we should have an "ask HN" on dating ... :-)
Once upon a time, I was seeking a Filipina gf. At the time, DateInAsia didn't have any good search interface but it was a good place to meet people if you weeded out the scammers.
I was looking for very specific things (Catholic, educated, no kids, 25 or older, etc). I scrapped (slowly) the site content and threw all the fields in a database so I could query it locally :)
I filled out my profile as complete as possible about myself and who I was looking for. DateInAsia lets people know when you viewed their profile... so my Python script automatically viewed all the profiles that matched my search queries. Many of them viewed my profile in return and those who were interested messaged me. I met some nice ladies that way but it turned out to be a Filipina lady I met in an unrelated chatroom who I fell in love with.
My own more humble attempt at mixing geekdom with love :) but love comes in unexpected places not ruled by math
It's a happy story, and I'm sad to see all the naysayers in this thread. I really don't see what's unethical about his behavior. He never lied about his views and his fiancee knows the whole story.
It seems like a totally rational response to the insanities of online dating. Especially as a guy, you have to message hundreds of women to even get a handful of replies. Through automation, he's equalized the playing field so that, like women, he has the opportunity to filter only amongst those who have already expressed interested in him. No more time/effort wasted on women who never reply.
I actually was working on an automated framework for batch messaging and a/b testing on OKCupid (https://github.com/morgante/abcupid) before realizing I don't have time for a relationship.
There are other ways to hack dating sites that use other means completely.
Have several friends on the site, all with their real profiles. When you like a girl, have a friend with the most compatibility message her, and introduce you two. If such a "cool" guy vouches for you, how great must you be? She is intrigued. She hasnt gotten messages like this. And you go out w her. Not only that but you start w a warm introduction and something to talk about.
And if she doesn't respond to him, you can message her yourself, and bam - double your chances.
[+] [-] crazygringo|12 years ago|reply
Like he said, he went on 55 dates, but only three second dates. The 55 dates really isn't too hard to do without the hacking, it's just a question of time. And the "three second dates" means his filter wasn't even that great -- he probably would have done better just simply browsing on the site, and only directly messaging the girls he found interesting in the first place.
But the real interesting thing here is the clustering into 7 types of women -- that's fantastic! I'd love to read more about that -- if he could write it up in a blog, OkTrends-style, I feel like it could get a huge number of hits. I think tons of people, including myself, would be interested in the details, especially if he did it for both men and women.
[+] [-] keerthiko|12 years ago|reply
He probably learned more engineering and advanced math while I learned how to read subtle messages of profiles, project the right ones of my own and know what's worth talking about ahead of time by having conversation online first.
He probably got the satisfaction of 'hacking' the system, while I had more efficient expenditure of money (dates can get expensive on average for guys when you decide to at least offer to pay).
None of my OkC prospects looked like they would work out long-term, though I'm still good online friends with over half of them. I flippantly shut down my account 3 months before I would have to leave the country, because even if I met someone I didn't want to be in a long-distance relationship, so I decided to save myself the trouble.
And then met a girl the very next day (not kidding) that I really hit it off with when I wasn't looking, and we're planning to go traveling in Asia next month (it's been 7 months).
I'm a diehard hacker and nerd and all, but when things like this happen, it's hard to not wonder if the traditionalists do have a point when saying you can't figure these things out with numbers. At least not when people are gaming their numbers :)
The math is super-fascinating though, and I hope OkC team does some research into it and integrates some of the ideas: obviously they wouldn't want to support uber-profile optimization for multiple groups, but maybe help find the right groups, etc.
I do think his biggest win as a male on OkC was being able to generate all the inbound traffic he would want with no invested marketing on his part (besides writing his scripts). That's something even the most skilled male OkC connoisseurs find difficult to do.
Edited: to add last thought.
[+] [-] com2kid|12 years ago|reply
That I disagree with, 4 years and hundreds upon hundreds of hours sunk into OKC, and I've had maybe 2 dozen dates from it.
Knowing how to optimize the site would be very useful. From pictures (OKC has MyBestFace that helps with that) to profile text. This guy obviously had good pictures (great pictures alone can make a profile, for either male or females), a decent profile, and the ability to write good opening messages.
Without at least 2 out of 3 of those, online dating becomes nearly impossible.
[+] [-] Qworg|12 years ago|reply
[+] [-] jcc80|12 years ago|reply
[+] [-] BioGeek|12 years ago|reply
He released a 37 page eBook with an easy-to-implement blueprint: http://www.amazon.com/dp/B00HY351S2
[+] [-] yodsanklai|12 years ago|reply
I think it depends where you live. In some places you can have as many dates as you want, in others, it's even difficult to have someone read the messages you sent. I experienced both.
[+] [-] randallsquared|12 years ago|reply
He'd already tried that, the article claimed. The problem is that he didn't know what was important to the women he was interested in, and so didn't know what parts of his life to mention in the OKC profile, to attract first dates from women he, as you note, might well have messaged anyway.
[+] [-] smsm42|12 years ago|reply
[+] [-] jeremyis|12 years ago|reply
[+] [-] blueskin_|12 years ago|reply
[+] [-] sjs382|12 years ago|reply
[+] [-] statguy|12 years ago|reply
[+] [-] CodeCube|12 years ago|reply
[+] [-] hcarvalhoalves|12 years ago|reply
You know... there's a damn big chance you find someone worth having a relationship after 88 dates. Something tells me his technique was no better than just dating at random.
[+] [-] technotony|12 years ago|reply
[+] [-] ahomescu1|12 years ago|reply
I wonder if the next generation of dating sites will match people based on facial expressions and body language (captured by a Kinect, for example), instead of all these stupid quiz questions.
[+] [-] cperciva|12 years ago|reply
[+] [-] jules|12 years ago|reply
[+] [-] gedrap|12 years ago|reply
[+] [-] diminoten|12 years ago|reply
[+] [-] jchung|12 years ago|reply
[+] [-] mehkcupid|12 years ago|reply
The match percentage was useless as a filter, but who cares? The new filter was my profile, and women who liked it messaged me.
Within a few weeks I'd been on several dates — I'm now married to the last woman I dated from then. She messaged me.
[+] [-] Pxtl|12 years ago|reply
[+] [-] birken|12 years ago|reply
I don't see how his system was better than just using the site as it is intended, nor do I think it should be romanticized as much as it is in this article.
[1]: Though it fails the categorical imperative. If everybody did this okcupid would be much worse off.
[+] [-] tunesmith|12 years ago|reply
Hard to say the process didn't improve things for him, though - maybe he's distinct enough that compatibility for him is more rare than average.
[+] [-] mgraczyk|12 years ago|reply
The categorical imperative is not a consequentialist motivation. It doesn't say, "Don't do X if doing so would make Y worse off."
[+] [-] asnyder|12 years ago|reply
Also, here's my .02 re-posted from the comments section:
Rather than answer the questions that were important to him he decided to find a set of people he thought he would like then only answer the questions they care about, and not even the way he would naturally answer them, rather he used an algorithm to determine the weight that would be best to get the highest match %. The fact of the matter is he could've spent a fraction of the time just answering all the questions honestly and with his honest weights and he would've found high matches too. Furthermore, he could've narrowed it down to just the kind of people he wanted through a normal search and then filter their questions based on what's important to them (which is a normal question filter on OkCupid).
So in fact what he did was pretty bad, violates OkCupid's TOS in numerous ways and at the end of the day wasn't honest to himself as he created specific profiles for his targets.
Honestly, we should not be celebrating this.
To recap what he did:
1. Didn't want to answer questions, so let's find all the questions that are important to everybody from the categories of people he thinks he likes based on clustering and then browsing a profile or two of people in that cluster. (He did this by creating numerous fake profiles and having those bots answer all the questions so he could scrape his targets question)
3. Create specific profile for his targeted group. With words and information that he knew they would like.
2. Answer ONLY the set of questions deemed important to those people. He answered these with weights determined by an algorithm that determined the best weight to achieve the highest match% rather than honestly.
3. WIth new found 99% matches go on dates with these people and follow normal dating process.
Now that we see the above broken down we can see that it's really not good. In fact, he was only answering what they wanted, and created profiles for them. But he wasn't being honest with himself or with his answers. If we're trying to match with everyone, which is essentially what he did, it's not that difficult to do. The fact that he eventually found someone is great, but the information used was faulty. Obviously there's no way he would be 99% with that many people normally.
[+] [-] jcc80|12 years ago|reply
“People are much more complicated than their profiles,” she says. “So the way we met was kind of superficial, but everything that happened after is not superficial at all. It’s been cultivated through a lot of work.”
“It’s not like, we matched and therefore we have a great relationship,” McKinlay agrees. “It was just a mechanism to put us in the same room."
[+] [-] jey|12 years ago|reply
[+] [-] oafitupa|12 years ago|reply
[+] [-] x0054|12 years ago|reply
Post a "Blind Date" message in the M4W section of CL. In the message describe your self as honestly as you can, while still being interesting and flattering to yourself. Ask the women to describe themselves to you in the reply. Say one or two interesting things about your self, and what you are looking for. Request that the responding girl does NOT send you a picture, and wait for the messages to roll in :)
I picked up quite a few dates that way, all the girls were beautiful, smart, and very interesting to talk to. Because we weren't a "100%" match, we actually had some different points of view, which lead to fun conversations.
You might think this would lead to you perhaps going out with girls who are not very good looking. First of all, you can have lot's of fun with a girl, even if you are not sexually attracted to her. But in reality, only girls who are very beautiful and confident in their appearance would actually reply to this message.
In any case, it worked great for me. I met lot's of cool girls, and eventually found the love of my life.
[+] [-] pella|12 years ago|reply
http://www.youtube.com/watch?v=d6wG_sAdP0U
"Amy Webb was having no luck with online dating. The dates she liked didn't write her back, and her own profile attracted crickets (and worse). So, as any fan of data would do: she started making a spreadsheet. Hear the story of how she went on to hack her online dating life -- with frustrating, funny and life-changing results."
[+] [-] monksy|12 years ago|reply
[+] [-] ceautery|12 years ago|reply
I have no idea what our match percentage was, and there were a few things in my profile that were turn-offs for her, all of which appeared in my only picture on the site: Me, with a fresh buzz-cut, jogging up a hill with my dog. She's allergic to dogs, doesn't run because of her asthma, and prefers long hair.
All of that was superficial, and she was able to look past it. She engaged me mainly because of the descriptive content in my profile. I just went nuts explaining who I was, in a chatty, stream of consciousness manner.
In the end, I re-homed my dog with her own parents, and let my hair grow to 21", which I started growing out again after we'd been married for a year. That wasn't all her; I had hair that length in high school and chopped it off to help me stay employed.
What's my point? The content matters most. You can optimize your approach to searching for matches, and you can go on lots of dates, but you can't force a good real-world match. If it's there, you'll know. No mathematical model of searching, nor red-pill-esque approach to building self confidence will be more effective than an open exchange of ideas between a couple. Get to the messaging. Give her a chance to be disinterested, because that's a hurdle you'll need to cross at some point in the relationship... assuming, that is, that you're looking for love.
[+] [-] vellum|12 years ago|reply
http://forward.com/articles/170925/hacking-jdate-to-find-the...
[+] [-] azernik|12 years ago|reply
[+] [-] llasram|12 years ago|reply
[+] [-] wjnc|12 years ago|reply
[1] http://stats.stackexchange.com/a/58920
[+] [-] hs|12 years ago|reply
the scrapper was written in newlisp (save search result pages with curl, use regex to match and collect the ids). it's probably easier to write in other languages, but that's what i knew.
i used wget and curl to loop over the ids but it's too slow because they download the whole page. later i found out about 'curl -i' (header only) and a million ids was done in about hour or two (i moved my operation from my home's 64kbps to my colo datacenter mbps internet).
my account is no longer exist (probably banned); however, i do still have a screenshot of me having 3000 female-only-friends and 70000 non-hidden-females 'look back at my profile'
initially, i talked to any interesting woman; however, later i made a strict rule to only respond to women who wrote at me. there were just too many fake female-accounts.
i got a couple of dates from this feat; however, i met my wife in a traditional catholic youth retreat. when i let her know about this friendster thingy, she just laughed. now i'm happily married with a 15-months-old boy.
[+] [-] acqq|12 years ago|reply
Math genius? More a spambot writer, but not for money, for an advantage.
[+] [-] beachstartup|12 years ago|reply
[+] [-] vph|12 years ago|reply
[+] [-] normloman|12 years ago|reply
[+] [-] programminggeek|12 years ago|reply
[+] [-] ajays|12 years ago|reply
When I was on OKC, about 1 in 20 messages would result in a first date; but you bet that more than 50% of first dates turned into second dates.
When women first meet you, it's almost like they're going through a checklist in their heads: is this guy a creep? A rapist? A jerk? etc. etc. (this is just the impression I got). If you don't trigger any of the alerts, you're golden.
One of the biggest mistakes nerds (like myself) make on their profiles and on dates is that we try to impress the woman with our encyclopedic knowledge of some esoteric subject. That's a sure turnoff.
Maybe we should have an "ask HN" on dating ... :-)
[+] [-] Paul12345534|12 years ago|reply
I was looking for very specific things (Catholic, educated, no kids, 25 or older, etc). I scrapped (slowly) the site content and threw all the fields in a database so I could query it locally :)
I filled out my profile as complete as possible about myself and who I was looking for. DateInAsia lets people know when you viewed their profile... so my Python script automatically viewed all the profiles that matched my search queries. Many of them viewed my profile in return and those who were interested messaged me. I met some nice ladies that way but it turned out to be a Filipina lady I met in an unrelated chatroom who I fell in love with.
My own more humble attempt at mixing geekdom with love :) but love comes in unexpected places not ruled by math
[+] [-] morgante|12 years ago|reply
It seems like a totally rational response to the insanities of online dating. Especially as a guy, you have to message hundreds of women to even get a handful of replies. Through automation, he's equalized the playing field so that, like women, he has the opportunity to filter only amongst those who have already expressed interested in him. No more time/effort wasted on women who never reply.
I actually was working on an automated framework for batch messaging and a/b testing on OKCupid (https://github.com/morgante/abcupid) before realizing I don't have time for a relationship.
[+] [-] kqr2|12 years ago|reply
http://christophermckinlay.net/
Also a link to the kindle edition of his analysis:
http://www.amazon.com/dp/B00HY351S2
[+] [-] EGreg|12 years ago|reply
Have several friends on the site, all with their real profiles. When you like a girl, have a friend with the most compatibility message her, and introduce you two. If such a "cool" guy vouches for you, how great must you be? She is intrigued. She hasnt gotten messages like this. And you go out w her. Not only that but you start w a warm introduction and something to talk about.
And if she doesn't respond to him, you can message her yourself, and bam - double your chances.