Geoff Chappell - Software Analyst
No pages at an obscure technical site are ever really hot, but a decent selection of them attract more than 100 visits per month.
There are nearly three thousand pages at this website. A small proportion have been written for a general readership, which here means moderately advanced Windows users. The overwhelming majority of pages here are very detailed descriptions of programming functionality in Windows. They are the sort of thing that may be the key to completing a Windows program, DLL or driver, or even to knowing what might be feasible, but they are beyond arcane to everyone else. I am astonished that any of these pages get looked at even once a day.
Still, since I do trouble from time to time to extract some statistics from the server logs, it is no extra trouble to present them a bit neatly and post them here—and since the overall rationale of this website is that other people’s software should be open for inspection and criticism, it is as well that I am myself an open book.
Or so went my thinking until 2017 (see below) when a change by the website’s hosting service made meaningful statistics look too hard to collect.
A side-effect of revising this website’s scripting and styling in early 2021 is that I bothered to look at what the site’s hosting service provides by way of configuration without requiring that I learn anything very much about Apache servers running on Linux. Thus does it come about that all access to the website since mid-February 2021 is redirected to use HTTPS and the server once again has statistical summaries that are immediately meaningful. The “open book” is back in play.
Much has changed in the last few years. Use of Internet Explorer, which for much of this website’s history was used by something like 80% of readers, is now (early 2021) down to roughly 10% and falling fast, having been supplanted by Google Chrome (at about 60%) and Mozilla Firefox (at around 25%). Of course, 10% for Internet Explorer is an order of magnitude greater than the share that some commentators give it, perhaps in their eagerness for it to go away. But this is a website about Windows programming. I can easily imagine readers accessing the site from computers (or virtual machines) that have little or no software that didn’t come with Windows, including from old versions. They’ll be using Internet Explorer for a good while yet.
Perhaps as interesting is the fall in readers running Windows. Though this website deals only with Windows and with programming tools for Windows, more than 10% of readers run Linux and another 4% run MacOS. Perhaps some or all of this is from mobile phones, but I can’t help thinking that there may be some abandonment of Windows even on real computers. After all, I myself find the modern Windows so unpleasant—and unpleasantly intrusive—that I wonder sometimes about changing to another operating system for everyday use. Considering how deeply I’ve been invested in studying Windows for decades, getting me even to think about using something else is quite an achievement by Microsoft.
Research and writing for this website took a break for a little more than five years while all my intellectual output was the property of an employer (who anyway was never able to provide working conditions that left a committed worker any spare time for research of any quality). The break ended in 2016, first by being less committed so that I had spare time at nights and on weekends, and then as a full-time sabbatical on the way to rebuilding a consultancy. Not until the year’s end, however, did I review the logs to look closely for what of my new work was getting read. The results are more than a little discouraging. Nothing that was truly new in the year had yet reached the admittedly arbitrary cut-off of 100 views per month. Clearly I’m not in the business of writing for the masses, but my first thought on reviewing the year was to wonder whether I should remain in the business of writing at all.
Unfortunately for the presentation of statistics, the hosting service for this website enabled HTTPS for it in May 2017. I didn’t ask for it and I can’t see how the site could ever make productive use of it. I doubt it does any readers any good, either. Along with doing nothing I wanted, it had the unwelcome effect of collecting statistics separately for access with and without HTTPS. In the absence of any obvious redirection from one method of access to the other, it may be reasonable to reconcile the two sets of visits per page just by addition. But the corresponding counts have to be found in each set to find which of them total to at least 100. Doing that by hand would turn the collecting into more than a brief diversion. Automating it leaves me resenting that I have to. So the presentation of statistics stopped.
Even before the break for my one experience of full-time employment, research and writing for this website had stopped at the end of 2010 while I took stock of what it’s good for, how it could be continued and whether I should want to continue it—which, indeed, it turned out I did not very much want to. October 2010 was for many years the last month in which truly new material had been added to this website. Still, I was open to rethinking and I continued looking at the statistics for another two years.
Raw logs do go further back but I didn’t start examining them systematically until the beginning of 2009.