This site uses information from two separate servers. The address pwall.net resolves to a server run by a hosting service — currently GoDaddy in Arizona. That server holds all the content for the website except for the full-size photographs, the Open-Source Software files and some files related to the private login pages, all of which are held on my home server in Sydney.

I do not keep any server logs for the hosting service, and I am not aware that the company concerned keeps logs for its own purposes.

My home server does not have a DNS name; I use the static IP address of the server to create references to content on this server. The current IP address, and the port number I use, is 60.241.114.220:9090.

I do keep a log on the home server, storing the date and time of the request, the URI, the IP address of the requesting machine and the identifer string (the “user-agent” string) of the software that made the request.

In addition, the site uses a CSS technique to display an indicator showing whether the home server is online. A small CSS file is loaded from the home server for each page — if the file is loaded successfully the indicator shows green, and if it fails to load the indicator is red. A side-effect of this process is that an entry appears in the home server log for each page loaded while the server is online.

I use the log information to determine which pages are being read, which images are being downloaded and what parts of the site require further attention. The information has already led to improvements in some of my pages.

Shortly after I started using my home server to serve the images for this website I noticed a request to download this image:

I was a little mystified — the image had at one time been on the main page of the website, but by the time of this download it had been relegated to the Picture Archive pages. And the requesting IP address resolved to somewhere like Kansas. Why was someone in Kansas downloading this image?

After two or three more downloads of the same image, all to equally unexpected locations, I decided to investigate. On a hunch, a tried a Google image search for “geese flying”, and there was my image in the first page of search results (when I tried again recently, I had to page down to about the third page to find it, but it's still there).

Since then, I have become accustomed to seeing requests from all over the world for various images and pages, and it's not always obvious what kind of query led to my site showing up in the results. So I have added a small piece of JavaScript to each of my pages, with the following effect: if the referrer (the page containing the link to this page) is not one of my own pages, call the home server to load a tiny image file, with the entire referrer string in the parameter to the call. This will cause an entry to appear in my server log, detailing the search that led to my page showing up in the results.

As with the rest of the log, I use this information to improve the content of my pages, and to try to ensure that this site is not appearing inappropriately in search results (or conversely, that it is appearing when it is relevant).

If you're unhappy with even this level of logging, you're welcome to block the use of JavaScript and to suppress the sending of the referrer string — the site will continue to work, with only slightly reduced functionality.

This site does not currently use cookies. If you receive a request to store a cookie from one of my servers, the cookie is being generated by an intermediate proxy (for example, a caching system). As far as I am concerned, you may (and probably should) refuse the cookie. What that does to the proxy is anyone's guess.

It is possible that I may use cookies in the future to store preferences, for example the preferred size for the return of images. If I do decide to do this, I will clearly state the purpose of the cookie before I send the cookie to the browser.

I will never make any part of this website dependent on the use of cookies — if I use them at all, there will always be a fallback mode of operation which does not require cookies.