Japan: On Twitter, nobody knows you're a bot

On the Internet, nobody knows you're a dog.” — The words of a well-known adage dating back to a New Yorker cartoon from 1993 have been interpreted and re-interpreted, acting over the years as a starting point for debates on privacy and online anonymity. In Japan, a country with one of the world's most popular online bulletin boards, the limits of this adage have become clear in recent years, high-profile slander cases having exposed the dangers of relying too heavily on online anonymity. Government institutions, companies, media organizations and advertising agencies have similarly learnt not to take the dog adage too literally — at least not while editing Wikipedia.

This week though saw a whole new take on the familiar “nobody knows you're a dog” adage. Blogger coconutsfine [ja], in a post entitled “The people I've been friends with for ages on Twitter were bots” (twitterでずっと仲良くしていた人がbotだった), described his disbelief at discovering that two of his closest friends on Twitter were not the kind of people he thought they were.

The blog entry, posted on March 9th, begins:

僕もtwitterを始めてからもうすぐ2年になる。情報系の大学に入ってから関わりのある人のtwitter-erが増えたのでどんどんのめり込んで行った。まあろくすっぽ大学生活をエンジョイもせずにtwitter充していたわけで、友人からの遊びの誘いを断ってまで twitterをやってるような廃人なわけだ。そんな廃人になってくると、そろそろtwitter上にも旧知の仲というか、移り変わるタイムラインでも昔からの顔なじみみたいな人が結構でてくる。昔よく見たのに最近見ないなと思っていたらアカウントが消えていたり、wassrで偶然見つけたり、まあそんなことがあったりする。

It's been nearly 2 years since I started using twitter. Since I entered a university specialized in informatics, the number of people around me using twitter has increased, and I've been getting more and more absorbed in it. I'm using twitter so much that I hardly take the time to enjoy university life, to the point where I'm like a shut-in, turning down invitations to hang out with friends so I can write to twitter. Having reached this point, I've ended up on twitter with what I guess you could call old friends, familiar faces that, despite coming and going, I've known for a long time. When the accounts of people who I used to see quite often disappear, I end up finding them back by chance on wassr, that's kind of how it goes.


Among my old friends on twitter, there are two users named @donsuke and @ha_ma. @donsuke has an icon of a cat right now, but he used to have an icon of a stuffed bear wearing a cloak, and he had this thing about ending his sentences with child-like expressions like “nano da” — quite the kidder. @ha_ma always uses the ending “nano yo”, and has an icon of a little girl, so she seems like a cute person. These two are good fun, always replying to my boring posts. When I would announce that I had arrived home, they would greet me with “okaeri”, and when I would announce that I had just woken up, they would reply with “good morning” — and I would often reply back. They were really close friends of mine.


These two were old friends of mine. Recently though, for some reason @donsuke started acting pretty strange. Last week, @donsuke replied with this total non-sequitur:
Suspicious post by Twitter user @donsuke: You're getting there! (9:11 pm March 2nd)

Suspicious post by Twitter user @donsuke: You're getting there! (9:11 pm March 2nd)


You're getting there!

こんなreplyを飛ばしてきた。まあtwitter上では誤爆はよくあることで、@donsukeは昔から大量postする twitter-erだったのでこのときもあまり気にしなかった。誤爆postしてることを教えようとも思ったが、twitterというのは発言がどんどん流れていく場所だから、そのままにしておいた。そして、一週間たった今日の昼のことである。

This reply came my way. There are more than a few off-target comments on twitter though, and @donsuke has been a heavy poster on twitter for a long time, so I didn't think much of it. I thought I might mention to @donsuke that his post was off-target, but there are so many messages flowing through twitter that thought, I'll just let it go. But then, just a week later, this came today at lunchtime.
Suspicious post by Twitter user @donsuke: You're getting there! (9 hours ago)

Suspicious post by Twitter user @donsuke: You're getting there! (9 hours ago)


You're getting there!


coconutsfine's response to finding out that @donsuke is a bot.

coconutsfine's response to finding out that @donsuke is a bot.


Is this what I think, is @donsuke a bot?

@donsuke のwebページに飛んでみる。他の一般ユーザーのtwitterアカウントだった。「どういうことだ?」と思いながらそのアカウントのwebページからたどってみる。すると、いくつかのtwitterアカウントが並んでいるページに飛ぶ。右上には「つくりました」の文字。

I rushed to check @donsuke's webpage, only to find that it was a link to another user's twitter account. “What's going on here?” I thought to myself as I followed the link to the webpage of the other twitter account, where I found a page with a list of twitter accounts. And at the top right of the page, there were the characters of the word: “Tsukurimashita.” [つくりました/”I made them.”]


“…@donsuke is a bot!!” I screamed at my MacBook, with no one around to hear.


But there was more. When I looked closely at the page, I saw another account name that I recognized. And I screamed at my MacBook again:
Tweet by @coconutsfine upon discovering that @ha_ma is also a bot.

Tweet by @coconutsfine upon discovering that @ha_ma is also a bot.


So @ha_ma is a bot too?


@ha_ma, you too?


Complete shock. I was speechless. These two were at the top of my list of friends.

おそらくだけど、僕が一番始めに出会ったbotは@bomtterだった。twitter上で「爆発しろ!」が流行るきっかけになったbotだ。そして同じようなbotである@wakatterが人気になった。しばらくすると@kyoujinのような人口無能のbotが出てきた。これらは一目でbotとわかるような特徴的なものだった。文脈無視の決まり文句だらけか、明らかに中に人間はいないような人口無能の文章生成postだったからだ。ところが、その bot達が注目を集めている頃からfollowをしていた僕の親友の二人は違った。@donsukeと@ha_maを僕が先にfollowしたのかあちらからしてきたのかは覚えていないが、僕は一度@donsukeをfollowしてすぐに「botかな?」と疑ったことが確かにあった。しかし、何故かは知らないがそのときbotでないと判断してしまったのだ。一度疑って、botでないと判断したらもうbotと思えなくなる。@ha_maにいたってはbot と疑いすらしなかった。「あれ?マジで@ha_maの中の人って女なのかな?」とか思っていたくらいだ。

I think the very first bot I ever met was @bomtter, the bot that sparked the twitter “Bakuhatsu shiro! [“Explode!”] trend. And then there was @wakatter, another similar bot that became pretty popular. A bit later on, there were artificial chatting bots like @kyoujin that came out. You could spot at a glance that these twitter users were bots, that was their common characteristic. They either posted canned phrases that were totally out-of-context, or they posted artificial-sounding phrases that were clearly not written by a human. The two friends of mine [@donsuke and @ha_ma] who I had been following in the time since these earlier bots hit the limelight, however, were different. I don't remember whether I followed him first or vice versa, but certainly I had my initial doubts just after I started following @donsuke that maybe he was a bot. But for some reason, I don't really remember why now, I reached the conclusion that he was not a bot. And once I had doubted him once, then judged that he wasn't a bot, I could no longer think of him as a bot anymore. On the other hand, I never even suspected that @ha_ma was a bot. I was wondering more about whether she was really a “she” or not.


But then, when I really think about it, I hardly have any real contact with most twitter users. They're all people I talk to exclusively through this intermediary, twitter. In the end, apart from real acquaintances, the majority of these people are not humans, but just “twitter accounts”. It really has nothing to do with whether they're bots or not. I just trust completely that a person with an anime icon actually has a really cute face, and that a person with an icon of Maki Horikita must definitely be Maki Horikita.


In any case, the fact that I was able to have such a valuable experience with someone who was actually a [computer] program makes me very happy.


In a note appended to the blog entry above, coconutsfine explains that @donsuke and @ha_ma turned out to have actually been submissions to a contest [ja] in which users try to design the most convincing Twitter bots, with each bot scored on the basis of how many times it is bookmarked. Blogger showyou explains his motivations for designing Twitter bots in an entry posted on March 10th [ja], and coconutsfine has also posted a follow-up entry [ja] in which he explains the reasons he was duped into believing that @donsuke and @ha_ma were actually human. See also an overview of the whole story at the Polar Bear Blog [ja].

This blog entry was translated in its entirety with permission of the blogger.


Join the conversation

Authors, please log in »


  • All comments are reviewed by a moderator. Do not submit your comment more than once or it may be identified as spam.
  • Please treat others with respect. Comments containing hate speech, obscenity, and personal attacks will not be approved.