Blog stats

Comparison of software

In fact the story has an intermediate stage between me manually gathering statistics and the server being upgraded. I had actually got so far as to look through the list of available WordPress stats plugins and found a few that looked interesting. Here is a quick review in no particular order:

LightStats

This seemed to offer some simple bar charts about post and comment numbers, and a nice word count per category pie chart. Given my previous interest in word counting my blog, I thought that would be a useful feature.

Quick Stats

No graphs, but it provides the very informative stats of most popular pages, referers (sic), and browser choice.

WP-Stats

No graphs, just verbose text, with information probably unnecessary for a small blog and some useful information missing.

It was relative to these, then, that I compared the choice of software made by most hosting provider, and was understandably impressed when I logged in to the stats page and was introduced to AWStats. There in front of me was every piece of information I could care to know about my website (except blog-specific stuff), all laid out neatly and logically (if we overlook the use of frames), with simple but effective bar graphs (that used the height property of the img tag to drastically reduce bandwidth). I was further impressed to read that this software did not threaten my Freedom, being licensed under the GPL.

Questions to be answered

The question remains, “How popular is my blog?”, but as I only have a week’s worth of data (which probably disproportionately reflects my own accessing of blog pages while interpreting the stats and writing this post) I don’t think it is helpful to comment yet. I will only say that the results suggest a much higher visitor count than 30 per month. A more detailed analysis of the figures should wait until at least next month, but there are some general comments about stats and some specific issues which came up for me to record here first.

As I indicated above, some of the interesting things to know are

1. which pages are the most popular,
2. which pages are the main referers (sic) to my blog,
3. and which browsers people are using to visit.

The first is useful because it adds a democratic potential to my work. If I ever feel I should be contributing more to the Web and trying to solve people’s problems, I can check what my readers find most interesting and try to write more posts about it. Unfortunately this can lead to a Duelling Banjos Effect where people find my pages just because I write about a certain subject a lot, even though the pages don’t include exactly what they are looking for. AWStats does have a breakdown of how long people are staying at my site, although obviously this is very hard to measure accurately, but if the number of people in the lowest time range decreases then that would suggest my posts are becoming more relevant to people.

Similarly the second in this list is useful for understanding people’s preferences, by telling me which page they were visiting immediately before mine. This is especially useful when the page is a search engine and I can see which search terms they are using, and in fact AWStats separates this information out and presents stats on the most common search query and search word used to find your site. It also separates out another piece of search engine related information, namely the accessing of your site by bots, crawlers and spiders. I’m not sure if this can give an indication of popularity, more likely an indication of how often your site is updated, but it does give a useful glimpse into the world of these software agents which are to some extent building strands of the Web behind the scenes.

Thirdly there is browser choice information, and AWStats also gives operating system information. This acts as a reminder that I need to test my blog on other browsers — I haven’t even tried viewing my blog in Internet Explorer 7 yet, which I should do as I pride myself on not only being able to create and maintain valid web pages, but also pages that work in Microsoft browsers. Practicality (and the fact I’m not getting paid to produce this blog) says that I can only do a limited amount of QA beyond checking the page for validity, but certainly if a large number of visitors were using a certain browser and that browser rendered my blog in a way that discouraged them from staying on or returning to my site, there would be grounds to prioritise support for that browser more. Like crawler information, though, the information is also useful for giving me a view of how the Web ecosystem is built up in terms of client software. There are other sites which have this information, but looking at my blog stats perhaps gives me the best indication of trends among people like-minded to myself.

The other stats are things like traffic on hours of the day, days of the week, and days of the month, and geographical information about visitors. As I don’t produce any particularly time-sensitive information, I don’t expect to see any traffic spikes. If I did release some world exclusive one day, I’m not sure how much these traffic graphs would tell me apart from how quickly knowledge about my blog spread over time. The geographical information might be useful as a rough indication of the native languages of people who visit my blog, but there’s not much I can do based on that apart from increase my effort to learn and spread Ido.

Problems

Those are the sorts of stats available and the sorts of questions they can answer, but I should mention a slight limitation of the software. Early on in the week the stats seemed to be suggesting that nobody was looking at anything other than the first page of my blog. This was disheartening but not unacceptable, until I realised that I myself had been looking at individual posts on my blog and these page views were seemingly not being recorded. When I examined the URLs which were recorded and compared them to the URLs of my blog, I realised what the problem was. The default for WordPress is to create permalinks (links to individual posts which do not change over time) of the form /blog/?p=123 where 123 is the post number, and AWStats is designed to strip everything after and including the question mark of a URL (in order to stop different requests from the same form being treated as different pages). As often happens, though, the solution was simple; I merely went to the Options page for Permalinks in the administration interface to WordPress and selected the third option in the list.

There was another small problem during the upgrade, not related to AWStats but worth mentioning for completion. After access to the hosting had been restored and my blog was online, I checked my body count graph and found the page blank. I methodically went through the code to see which line was failing as there were no error messages in the output, and worked out that the problem was not in my own code but in that of the graphing library I used. The obvious conclusion was that the upgrade included a newer version of PHP that was incompatible with one of the features of the library, so I sought out a later version. Sure enough there was such a later version, and I installed it, but it did not produce any different output. The newer library version did, however, contain some diagnostic files, and I used one to tell me what was wrong with the environment on the host. The answered turned out to simply be a lack of the GD graphing library. I was all set to complain about this to my hosting provider when suddenly, maybe the next day, I checked again and found everything was working without me having to change any code. It is very nice to find such a simple upgrade take place, in contrast with upgrading a certain piece of blogging software.

Philosophical question

A few of my statements above lead on to a philosophical question, namely whether popularity is a worthy goal in itself. The democratic argument I hinted at was that by being popular I would be giving more people more of what they wanted. So, assuming that people want what was best for them or what was best for the Web / world (which is a big assumption) then the more popular my blog is, the more good I am doing. There is the objection to democratic arguments that an entity trying to be popular will be forced to abandon any distinguishing features that might deter some section of its audience. This leads to all entities offering almost exactly the same choice, usually appealing to the audience’s basest and least sophisticated desires.

On the Internet, though, this is slightly different. The potential audience size, the lack of lock-in to a particular site or service (helped by standardised protocols and formats), and the low barriers to entry, mean that anyone who is prepared to settle for a small percentage of the audience (although a relatively large absolute value) can appeal to a niche and still consider themselves popular or make a profit. I am not the first to point this out, of course, as it is basically a statement of The Long Tail effect, but maybe I am pointing out a corollary to it: aiming for popularity on the Internet is less likely to have the bad results of doing so offline, as long as you are doing so within your niche on the long tail. Perhaps the advice is useful offline too, but practicalities like high overheads and entrenched monopolistic opposition make it too often impractical to stay within a niche. I also feel there is some point to be made about the Internet causing highly popular services to fail because they don’t perfectly fit the niche requirements of their users, and I think we have seen a trend towards more open communities and more customisation of sites and services, but I think this argument can wait for a bit more data first.

Finally there is the point that a blog doesn’t necessarily have to serve other people, and it can be purely a way of encouraging you to put your ideas into words to the standard of a (perhaps imagined) audience. If I felt that helping people through this blog required too great a cost then my efforts to help people should better be focused elsewhere, and popularity statistics should not influence me. So far, though, I think that the at least nominal purpose of my blog of showing people useful examples of programming and fixes for bugs, is more of a useful framework than a limiting boundary. I’d like to think that if you do what you love and you don’t compromise your principles, then your audience, however small, will appreciate that and you will benefit from it.

Epilogue

So, when am I going to plaster my blog with AdWords to “monetise my Intellectual Property”? (I feel dirty just writing that). Unfortunately the choice of whether to include advertising is a fairly mathematical one. Is the cost to the aesthetics of my site, the good will of my visitors, and my unconscious independence from the commercial pressure to leave my niche, higher than the value I would get from having adverts? (Find out when I write up the statistics after getting some more data).

Anke, kad mea Europana frato di ica texologo recevis mea komento?