Horas and konnichiwa untuk semua

At least currently, Google mail’s capacity grows at 4 Bytes/second. You can watch it grows on login screen like below:

I previously wrote that I need additional 4GB per year. At the rate of 4B/s, in a year there will only be an increase of 4B x 60 x 60 x 24 x 365 = 126.1 MB 😦 Hopefully there will be rate ups from time to time.

The other day I was given this link which tells about Yahoo Indonesia and a research company announcing their research on Indonesians’ online behavior:

Internet, Kita, dan Masa Depan
http://cetak.kompas.com/read/xml/2009/03/27/03024518/internet.kita..dan.masa.depan.

I am a bit skeptical about statistics like this, since research companies like AC Nielsen, IDC etc often lack the knowledge they need to analyze the market. They are already famous so they dig lots of money, though their research staffs lack the knowledge to analyze the market.

So I decided to find out by myself.

I took Facebook as my target, as Indonesians are said to be crazy of Facebook and my observation shows it is true. I wanted to find out how long is an Indonesian online on Facebook everyday. Just logging in into Facebook makes one’s status online. I thought I needed to check who among my friends on Facebook that get online and when. I decided to use Adium, since it has Facebook client and senses when a friend becomes online. I thought Macruby would be nice since it has Cocoa natively, but after trying for a while I thought it was too much to learn and I only got a few hours. I turned to AppleScript, which is also new to me, though I used to be programming on HyperCard, its predecessor. I finally coded the stat tool in a few hours using AppleScript, Adium and Growl. Growl is only used for visual feedback, because the script must be running throughout the whole data gathering session.

The next day I ran the script and gathered the data from 7:30AM to 17:30AM West Indonesia Time. I got quite consistent results, as follows:

1. Around 25% of my Facebook friends (more than about 99% of them are Indonesians) went online that day
2. Average total online time among friends who went online that day is about 60 minutes (i.e., 1 hour)
3. Standard deviation is high (around 60), indicating that total online time varies a lot among friends
4. Average total online time for the whole population (which includes those who did not go online that day) is about 15 minutes

My tool has room for improvement indeed (e.g., I am polling once every 60 seconds, so users who went online and offline in between pollings won’t be noticed), and the sample for this statistic is too small compared to the population (the sample is 259 friends, and the Indonesian population on Facebook is around 800,000 users, according to Facebook Ad). But at least this shows that there is a good probability that 200,000 Indonesians (i.e., 25% of 800,000) are online on Facebook every working day during working hours;)

Now I wanted to know what would be the result if I measure during the night (off-hours) and during weekends. This might give some insights on whether there are significant mobile users logging in into Facebook. In Japan, more users use mobile to access Mixi, a Japanese SNS, than PC. By the way, Mixi was made by an (former) Indonesian named Batara Kesuma 😉

No, sorry, this is not April’s Fool. I really did the above research 😛

Here is the AppleScript code. Comments are welcome but be gentle, as this is my first AppleScript script and I coded it in a few hours. I originally wrote the data to file, but then changed the implementation to use Sqlite, so some garbage is still in the code.
Continue reading →