Ep40. 假如我要設計一個dating app

Welcome,

please enter the PIN and we shall get started with soul matching.

You'll be matched with a total of 5 candidates based on your location, mental capacity and lifestyle preferences, and each will be given a score of compatibility for your reference.

Please list your preference after the beep.

大家好嗎？唔知大家喺疫情下點樣識朋友嘅呢？有冇用過坊間嘅dating app呢？想當年我都算用過一段時間。當時我好熱衷舉重同埋高強度運動。如果呢度有CrossFit 嘅朋友仔可以相認吓！我通常first date會約對方去gym，都嚇親唔少對象😂 宜家應該改做去《好青年荼毒室》行返轉買返本書先。唔知大家有冇試過，睇相好似唔錯，網上面傾計都好傾，但到真人見面嘅時候，就似一個空洞嘅軀殼，無話題，亦都唔知點樣打發約出嚟嘅時間。如果你話我哋認識朋友嘅時候，可以認識咗佢嘅內在先，你話幾好呢？所以就萌生念頭，諗如果我可以設計一個dating app我會點呢？

不過喺講我嘅奇思異想之前，想同大家睇吓呢篇2018年The Atlantic嘅文章《The Five Years That Changed Dating》by Ashley Fetters【1】。雖然已經係三年前嘅作品，但係內容都寫得幾好！根據文章，dating app早期serve嘅係gay community，早期嘅app係喺2009同2010年開發嘅Grindr and Scruff。Tinder喺2012年推出市場，好快就成為咗市面上最受歡迎嘅dating app，文章甚至話世上有七成嘅智能電話都下載咗Tinder添！然後就有好多唔同嘅dating app推出市面。文章話：「Tinder has indeed helped people meet other people—it has expanded the reach of singles’ social networks, facilitating interactions between people who might never have crossed paths otherwise.」佢話Tinder（其實都可以應用喺其他dating apps上面），令用家嘅社交網絡擴展，令一啲正常唔會遇到嘅人可以認識對方。但係都有啲trade off嘅，譬如你記唔記得上次同陌生人傾偈係幾時嘅事？文章其中一個受訪者話：佢去party真係淨係expect同朋友玩，佢唔會random咁同女仔講嘢，然後之後約佢。佢話：「It wouldn’t be an abnormal thing to do,” he says, “but it’s just not as common. When it does happen, people are surprised, taken aback.」簡單來講，而家會俾人歧視覺得係creeper咁。作者就話，以前佢嘅年代，講緊十年前，識人was the point of going to parties。但係反而依家喺dating app上面，個bio反而係「If I know you from school, don’t swipe right on me.」究竟我哋有啲咩唔見得光，要避開我哋識嘅人？受訪者話佢知道以前嘅年代，認識另一半大多數都係喺學校、公司、朋友、親戚等等，但係而家約會反而係一個孤立、同自己平時嘅日常生活無任何交集嘅圈子。

之前見過一個app叫happn係會根據你去過嘅地方，配對返同你去同一個地方嘅人。例如你一個鐘之前去過呢間cafe，同樣去呢間cafe 嘅人就可以見到你個名，message你；但係唔使講都知真係有啲creepy啦⋯⋯但係Google Play有成100萬download。除咗happn之外，都有唔少app會依靠你嘅所在地去推薦對象俾你。文章話：「The relationship economy has certainly changed in terms of how humans find and court their potential partners, but what people are looking for is largely the same as it ever was: companionship and/or sexual satisfaction.」文章引用一位研究attraction and romantic relationships嘅心理學教授Madeleine Fugère話，通常如果喺學校或者返工認識嘅人，都會同你有好多common ground，但係如果你同對方一樣嘅單純係地理位置，咁大家會有多啲唔同嘅地方。Madeleine話「People who are not very similar to their romantic partners end up at a greater risk for breaking up or for divorce」，好似阿媽係女人咁，就係如果大家有好多唔同，就容易分手或離婚。另外，因為我哋同對方嘅距離好remote，所以我哋越嚟越唔識或者冇基本嘅manners，例如隨時爽約、冇禮貌、或者拋低一句「下乜原來你個樣係咁架？」就走咗去。

文章結尾拋低一句：「"It’s like, Ugh so many dates, and they’re just not that interesting,” Finkel adds with a laugh. “It used to be hard to find someone to date!」我覺得用algorithm可能會efficient啲，但係唔知大家會唔會將相遇當成一種緣份，定係視之為一個我哋用dating app嘅必然後果，呼之即來揮之則去呢？文章都幾light，大家都可以當飯後娛樂咁睇下

好喇講咗咁耐，究竟我心目中嘅dating app係點架呢？隨著網絡世界嘅連結越來越廣，我哋嘅交友模式已經冇辦法返轉頭。就係全球肺炎肆虐嘅時候，喺2020年6月，就有一個新嘅tiktok x tinder咁嘅dating app推出咗！呢個app叫Lolly【2】，係入面就係每人都錄一段短片去展現自己嘅性格，然後你可以係自己個feed度睇到其他人嘅短片，然後私下同對方傾計。所以真係冇辦法返轉頭啦！假使我哋以後都會繼續依靠呢種模式搵對象，咁我哋點樣確保我哋可以搵到一個同我哋有足夠嘅「相似度」嘅人呢？咁啱我一路而來幻想中嘅程式，就某程度上解決咗呢個問題－－如果我哋依靠自身嘅社交圈子嘅原因純粹係因為想大家諗嘢相似、有共鳴啲，咁不如就將dating app集中配對思想。雖然同location tracing一樣都係有啲dystopian嘅但係諗吓無壞吖！

但係set programme or algorithm之前首先要搞清楚嘅係點樣先叫一個good match：

籠統嚟講我諗三觀意指世界觀、人生觀、價值觀都要夾。至於有無生理心理感覺就好難預測到，所以呢個先放埋一邊先。於是我嘅問題就縮窄為點樣可以match到三觀呢？再退後一步諗，我地嘅三觀係點樣形成嘅呢？又點樣體現嘅呢？

我第一樣諗到嘅就係word association。事先聲明呢個係冇科學根據嘅，淨係我嘅想像。想知道對方同自己有冇common ground，我嘅諗法係對比大家嘅mind map。首先：每一個參加者都要喺限定時間內做聯想。題目應該係fix咗幾條以作比較，以random order出現，一分鐘之內寫15-20個字。當然呢個方法唔好嘅就係對於一啲語文或聯想能力比較強嘅人會容易啲，但係個idea都係想對比吓大家對於同一樣事物嘅睇法同聯想有冇相似嘅地方。用家可以set parameters選擇想配對到同自己好相似或者好唔同嘅對象，例如可能有個setting叫「相似度」，你可以由0-100%咁調整。又或者你可以選擇人手揀唔同嘅mindmap，邊個啱感覺就可以過一關！

第二就係睇吓對方嘅brainfood。呢個係我自己改嘅名嚟嘅，意指餵自己個腦食啲咩。通常我都係講啲咩書。如果我哋可以知道對方睇咩書，都可以大概估到佢嘅世界觀同價值觀？如果書太冷門，我就會研究個app可唔可以嘗試吓extract Spotify 或者其他audio player嘅資料，睇吓大家聽嘅內容。又或者如果呢啲都太冷門，都可以show唔同嘢食嘅相，然後睇下對方鍾意啲咩！同樣，我哋可以手動揀或者由得algorithm根據我哋set嘅「相似度」幫我哋配對。點解要俾用家自己set呢，因為可能我哋唔想陷入一個echo chamber，只係認識同自己相似嘅人。我覺得50－60%相似咁就啱喇！

如果第時更加dystopian，甚至可以即刻靠隨身帶著嘅Apple Watch或者其他可以感應到心跳、流汗嘅工具，即時monitor你睇到個profile嘅時候有冇一啲自己意識唔到嘅感覺；亦都可以量度腦嘅轉數速度，咁樣就可以確保兩個人可以有相似嘅思想速度。呢度唔係要話夾IQ，首先因為智力有太多唔同嘅定義，而唔同範疇亦都可以有唔同嘅能力，所以智力唔係咁重要，反而重要嘅係大家嘅速度：一個太快一個太慢會難啲係現實生活中溝通。

如果你都係外貌協會嘅一份子，其實我呢個app都可以幫你配對！我諗喺交友方面唔係要去選美，唔係要鬥靚而係點樣可以搵到一個compatible嘅。而根據今年三月五號我收到嘅MIT Technology Review嘅newsletter入面Tate Ryan-Mosley寫嘅文章《I asked an AI to tell me how beautiful I am》【2】，入面提到一間公司Qoves Studio有個「facial assessment tool」： an AI-driven system that promises to look at images of your face to tell you how beautiful you are—or aren’t—and then tell you what you can do about it. 如果你肯俾錢，佢會gen個report出嚟提議你可以點樣改善你嘅樣貌添！但係呢個唔係重點啦，重點係algorithm可以將你嘅容貌用統一定義評分，咁樣如果你嘅樣貌係8分，你可以選擇配對一個你自己設定嘅range，例如係5至10分。

以上嘅狂想嘅目的係為咗等大家見面嘅時候可以有話題，去模仿或取代我哋以前依賴朋友、同事、家人嘅圈子。呢個唔係一個好ideal嘅做法，因為咁樣就要將好多唔同嘅細節設定為一個linear scale同埋二元切割，同埋綜合好多大數據先做到。不過我諗大家唔知嘅就係而家全球大部分出名嘅dating app都係under同一間公司嘅呢？例如Tinder、OK Cupid、Hinge、Match.com等等都係under Match Group，而呢一間公司係上市公司，喺2020年嘅收入足足比上一年多咗19%！【4】假設我哋somehow都有將我哋嘅個人資料例如Facebook or any identifiable account details link咗去呢啲app，分分鐘佢哋已經有我哋每個人好全面嘅資料！靠用personality traits、spotify playlist、pinterest boards、geolocation、deliveroo or foodpanda嘅喜好等等配對，唔係難事啊！第時會唔會愛情都變成售賣商品，例如邊個俾得起錢就可以配對到一個compatible啲嘅對象呢？

我覺得真正嘅relationship，無論係友情定愛情，我相信係現實嘅交流係無法取代嘅。正如《Before Sunset》入面女主角話：「You can never replace anyone because everyone is made up of such beautiful specific details.」又或者係《攻殼機動隊》入面女主角話：「正如要有林林總總的部分，才能組成一個完整的人。想要構成迥然不同之人，所需要的東西多的千差萬別；異於他人的面容、下意識裡的聲調、夢醒時凝視的手掌、兒時的記憶、未來的期盼」雖然佢嘅對白係講緊咩係「自我」，但係對於乜嘢先構成一個獨立嘅人同《Before Sunset》有著異曲同工之妙啊。想認識一個人，無論係做朋友定係情侶，係咪真係可以靠algorithm呢？呢個問題就要等我哋喺未來發掘。

【1】https://www.theatlantic.com/family/archive/2018/12/tinder-changed-dating/578698/

【2】https://www.ypulse.com/article/2021/02/17/this-new-dating-app-is-tiktok-meets-tinder/ https://www.bustle.com/wellness/lolly-gen-z-tinder-social-media-dating-app-tiktok

【3】https://www.technologyreview.com/2021/03/05/1020133/ai-algorithm-rate-beauty-score-attractive-face/