26 February 2015
I was about to pack up and go to my part-time job before I randomly checked my website (kenrick95.org) stats at Google Analytics and was surprised because there is a spike of visitors at 15 February.
As you can see in the chart above, it has 90 sessions on 15 Feb whereas normally it was less than 5 sessions per day (and usually they’re the source who send me spam e-mail, since I publish my e-mail address there).
I investigated further by viewing the location reported by GA. Majority of them was from Japan!
Then I navigated to “Behavior” and saw that most of the visitors were from referral, and to make things more complicated, most of the referral are from kenrick95.github.io!
Checking for more details, I navigated in and saw that 100% of the referral from kenrick95.github.io were from http://kenrick95.github.io/c4/demo/ which is a Connect-Four game that I implemented some AI algorithm in.
Aha! I found the culprit. Wait, but why?
I remembered that I’ve ever publicize about the game on Hacker News (around 15 January 2015) and it ever went to top-10 of the HN front page.
But that did not generate any spike to my personal website. And currently the discussion has been dead for at least a month. (https://news.ycombinator.com/item?id=8886897)
As you can see on the chart above, the peak was just 4 visitors, but it has no clear spike as the other days (22 Jan-29 Jan) were also just fluctuating randomly. Hence, my publication to HN has no impact on my website visitors. I recalled that publishing on HN only has impacted my GitHub repo stars (gained around 30 new stars) and traffic (around 300 views, which a lot of them view the underlying script.js)
Now I went to the stalking tool, Google, to search for what has caused the game to be popular again.
Well, I saw someone posted a screenshot of winning the game.
— Danny Garcia (@dannygarcia) January 14, 2015
But still, it was around 15 January 2015, so it was from HN.
And then on the 2nd page of Google Search. (like finally navigated to 2nd page :P) some Japanese links shows up.
I went to the links and seemed like the articles there were republished from another website called Gigazine. So I searched for the article on Google for the Gigazine article and voila: I found the same article, only this time was from the original author. (http://gigazine.net/news/20150215-c4-ai-game/)
I saw that the date of publication was 15 February 2015. Yes, this should be it. I finally found out the real culprit. Turned out that Gigazine is a quite popular Japanese news website.
No I don’t read the article, because I do not understand Japanese, but since there were a lot of pictures (even a video showing the game playing process!), I assumed that it informs the reader on how to play the game. Yes, I admit that when I show the game to my some of my friends, they do not know how to play (as on the game it lacked of playing instructions).
At this time, I was so satisfied and happy. Satisfied because now I know where the spike coming from, and happy because someone published a detailed instructions on playing the game on a popular Japanese news website.
I never expected this game to be this popular since it was just a side-project that I did during December 2013-January 2014 semester break, after attending HTML5 Bootcamp at NTU and implementing what I’ve learned from CS188.1x Artificial Intelligence MOOC offered by BerkeleyX at edx.org during February-May 2013. Well, thinking about it, maybe the nature of hard AI but still-possible-to-win is one of the reason that make people replay it until they win the game.
Hopefully, I can do some other stuffs (or games) that uses my knowledge gained from what I’ve learned in MOOCs and in NTU.