For the past few years, this blog has been using Google Analytics. Although the site only gets a few hundred page views per day, I sometimes feel a bit vain and want to know whether real users are actually visiting. I recently happened to look at the GA analytics and noticed that the share of visitors from Taiwan is unusually high.
Out of consideration for careful handling of user data, server logs are automatically purged, and Google Analytics doesn't identify specific users, IPs, or timestamps — it only provides statistical information. This is exactly the kind of limited usage I want, and it does give me plenty of information.
GA's stats show that over the past year I had 65,000 unique visitors. Of these, 24% were geographically located in Taiwan — only 10 percentage points below mainland China's 34%. Analyzing by ISP (Internet Service Provider), 9% of the traffic came from "taipei taiwan", followed by "(not set)", and then "taiwan academic network". Only after those come the various providers from China and the US.
I myself find this blog rather boring. Most of the content is just notes on simple problems I wanted to keep for myself, shared incidentally with whoever happens to pass by. Either way, I haven't found anything particularly special about it. All the Chinese I write is in Simplified characters. The terminology is also what we use in class, and when I talk offline with friends from Taiwan, I've found that many translations are quite different. So whether comparing populations of 1.3 billion versus 23 million, or search habits and various keywords, I naturally assumed that Taiwan's share should at most be no more than a third of mainland China's.
Whatever exists is reasonable. Since "Taiwanese visitors exceeding of mainland Chinese visitors in Google Analytics" is a fact, I can only speculate about the reasons under this fact. Here are some baseless guesses.
Small sample size
As mentioned above, the site had a total of 65k visitors over the past year. This can't even be described as "niche" — it's more like "who knows whether this site even exists". Add to that the fact that the blog's topics are quite narrow; leaving aside the low quality, the field itself is one that only a very small number of people care about. In this situation, the precision of user ratios is heavily affected. The difference between 30% and 60% might just be someone refreshing a few extra times, or some big name casually mentioning it in some circle.
So first of all, this may not even be a problem — just a random outcome.
Ad-blocking extensions
The most important extensions in a browser are probably things like AdBlock and uBlock Origin, which control browser behavior. Their function isn't just ad blocking. We can also use them to clean up pages and make websites look simpler. More importantly, they can block certain special requests — such as GA, or those obnoxious cryptomining scripts that have been popular lately.
Take me, for example: when I visit this site, I certainly don't send a single request to GA. Nor do I intend to force users to tell me they visited. Furthermore, from an implementation standpoint, if some "magical method" were used, it would mix in a lot of crawler information. That would raise recall, sure, but precision would plummetprecision and recall?. So blocking GA's tracking is actually very easy.
In mainland China, there are countless websites with no integrity. Even so-called "technical people" talk about ideals while their bellies are full of business schemes, thinking others can't see it. In this environment, combined with the sample-size issue discussed above, those who are interested in coming here are mostly fairly knowledgeable about computers. In that case, one can imagine how high the extension installation rate is in mainland China.
Therefore, uncounted users may also contribute to this ratio discrepancy.
Network issues
This machine is hosted on Linode. As everyone knows, mainland China has a particularly formidable gateway. A lot of traffic from outside the gateway gets randomly dropped — figure out the reason yourself.
Personally, I think blocking various phishing sites is reasonable. I personally don't think much of various cults either, and porn and such things are judged differently due to cultural differences, which is also understandable. But what really bothers me is punishment without warning. Punishing without teaching first is a very bad tendency. "Your whole family can go to hell! Figure out the reason yourself."
As far as I know, this little site is also partially "sent to hell" on some networks, while some networks decide not to send it yet. Even more magically, some networks decide to send it partially but not the whole family. So it manifests as accessible but unbearably slow.
A study[citation needed] found that for every x% increase in police efficiency, the number of public security incidents in a city during a period can drop by y%. When public security is poor, y can be an order of magnitude higher than x.
I can't find the specific content of this study right now. But the conclusion is quite intuitive: when deterrence exists, a small amount of deterrence can push a large number of people on the edge back. Applied to this little site, I suspect many websites get similar treatment. When it and many others have been messed with like this, users may choose to give up — closing pages that load slowly, or simply routing everything through a tunnel.
Therefore, some mainland Chinese visitors may have simply given up on visiting, or viewed the pages under another country's identity. Even if GA tracking is allowed, it would just add "+1" to some other place.
Source/referrer issues
Has anyone experienced looking up something for a colleague or friend, jumping from link to link, and landing on your own page, with the problem solved on the spot? I've had the pleasant surprise of that happening. So whenever I write some notes, I sometimes believe they might genuinely help someone — that's one of the motivations for keeping this site up despite the low traffic.
The prerequisite is that they use Google.
How should I put it... Baidu and I don't really like each other. Baidu dislikes me for being unregistered and unfiled overseas; I think Baidu ranks various plagiarized reposts of my content at the top while mixing in the originals, which is rather low — making it look like I'm the one copying others. So I haven't maintained any sitemap or such for Baidu.
Recent referrer analysis also shows that traffic from Baidu here is quite low.
Additionally, some sites that "back up" this site are quite amusing — the full text is all written by me, except they've kindly changed the author name for me. Managing to mark a source in some tiny corner is already more than generous of them.
On the other hand, I do see some Traditional Chinese pages that solemnly link to this site next to the relevant topics they mention. Some of those authors are exactly the kind of people I'd love to get acquainted with.
China has plenty of regional stereotypes — Henan and manhole covers, the Northeast and machetes, Jiangsu's thirteen cities all looking down on each other. The Taiwan/Fujian gag "we're like the Northeast triads, your son is in our hands" is also widely circulated online.
But in this regard, Taiwan sometimes feels a bit like elementary schoolers. Can't compare, just can't compare.
So I figure attracting traffic is one factor too.
The above is just some casual speculation, written as a way to slack off after coding until my head hurt.
If any friend from Taiwan happens to pass by this page, might I have the pleasure of collecting some insights from a visitor's perspective? Much appreciated.
Comments · 5