Online Dating Will Teach You What You Need in Digital Era
Tinder? eHarmony? Match.com? Nowadays there are many matching sites and apps. Some of my friends are addicted to Tinder, while some of them complain that they don’t get good matches even though the matching quality is said to be increasing. To be honest, I know how they feel. Personally, I started to use dating apps several months ago back in my country, Japan. But it didn’t work well. It took long time to find girls who have appearance, background, and features which match my preference. (I’m not picky though haha) Even though I found attractive girls, they sometimes just ignored me or end up with nothing after several conversations. And I had to pay a lot to get the membership for approaching more girls. So, I’m really looking forward to improvement of the matching service. Nowadays, machine learning is progressing rapidly, but why don’t matching apps suggest best match for me immediately?
People You May Know (PYMK) is similar with those matching apps. Facebook is famous for its suggestion, but many people say that the quality of their PYMK is quite low; Instagram under Facebook is said to be worse than Facebook. [1] Also Linkedin has PYMK service for their customers to get connection with new people. As Suja Viswesan talked in the lecture, Linkedin developed complex but sophisticated infrastructure for their PYMK. [2] To achieve the good quality, they collect massive data about a person’s job interest, their connections or even web pages they visited. Then computers calculate those data to suggest PYMK.
What about matching apps? It seems more difficult to collect data from customers because no one is sure what kind of data is critical for good relationships. Moreover, data regarding their personality is not as easy to handle as their job interest. For example, even if one guy says that he likes “outgoing” girls, his definition of “outgoing” can be different from matching app creators’ definition. Therefore, it will be a concern what kind of data we collect in what way. And how we interpret the data is another problem. There is an article which says that the similarity of partners’ personalities accounts for only half a percent of how happy they are. [3] To put it simply, it implies that collecting people’s surficial features and calculating similarity don’t work.
There seems to be many people who believe that as AI or machine learning develop, they can get better matches more frequently. But it’s not true. It depends on our “judgement” ability. It is still we who decide what kind of data we collect and how to build algorithm. As Prof. Charles Eesley says, “judgement” is the last part which we’re still responsible for in this digital era, while other actions can be replaced by AI or robotics. [4] In coming digital era, what we really need may be high literacy of judgement.
[Sources]
[1] https://www.usatoday.com/story/tech/columnist/2017/07/30/why-facebooks-people-you-may-know-makes-some-weird-suggestions/521264001/
[2] Suja Viswesan, the guest speaker at MS&E 238 on 20/7/2018
[3] https://www.washingtonpost.com/news/the-intersect/wp/2015/11/11/the-one-thing-about-matching-algorithms-that-dating-sites-dont-want-you-to-know/?utm_term=.761f0c7615f3
[4] Prof. Charles Eesley, the guest speaker at MS&E 145 on 11/7/2018
Users who have LIKED this post:
4 comments on “Online Dating Will Teach You What You Need in Digital Era”
Comments are closed.
Love in the digital age still needs a look at! There are so many more rules of attraction that come from in person interactions with a prospective partner. Dating applications and services need to look beyond the base level of attraction (visual) and look into deeper metrics if they wish to be successful in the long run.
Don’t you think that this online dating apps can have negative consequences as well? There is the safety aspect, how do you guarantee that the person on the “other side of the screen” is what he or she is promising to be on the app description? This can lead in to some serious safe consequences that are hard to track and prevent. Honestly I would be really apprehensive in going in one of those dates without knowing the person. People can describe them self as they want and create a perfect profile full of lies. Even though the app do the matches it is still the humans that input the information and this information can easily be manipulated.
Given the dating apps scenario that you exposed, I personally agree with you when you say that in the end it actually depends on each user’s “judgment” ability, because they are the ones who choose which data to enter. This leads us back to the discussion: do machines have consciousness and self-awareness to make decisions and predictions? The answer is no. Behind each application or program there is only a lot of coding, statistics and data, which were all placed and build by humans. So, in this sense, the only ones who can influence how the algorithm will do the “match” with others’ profiles is the user himself. The machine will only collect that data and run an algorithm in order to perform what it was coded to do.
I wonder if these dating apps could be improved if the ability to “debrief” users’ interactions with potential mates were somehow integrated into their data feed. I think this might be a way of improving the kinds of matches made by the algorithms much better, and maybe make the process seem less complicated.