discuss: Using Webbug to track hits
Subject:
Re: Using Webbug to track hits
From:
Kwan Lowe ####@####.####
Date:
24 Apr 2004 12:06:25 -0000
Message-Id: <1082808392.32572.55.camel@darkstar.digitalhermit.com>
Not responding to anyone in particular, but I do want to put my opinion
in the record:
The idea is that a tracking image can be inserted into documents on the
TLDP site so that the popularity of documents can be gauged. The IP
addresses of the readers will not be saved in order to prevent abuse or
malicious mining.
I disagree with this proposal for the following reasons:
1) I do think that gauging the popularity of documents can be useful to
focus resources and see what people are looking for. I do this very
often by going through my apache logs and:
a) looking for search strings
b) Looking at the hit counter for particular documents in webalizer
From this existing data store I have a pretty good feel for what
information is desired. With the number of hits that tldp itself
generates, a statistical look at the log files can easily be
extrapolated to the population in general. You can probably get better
than a 98% confidence by just peeking at existing logs.
2) None (or very few) of the documents currently contain this tracking
GIF. This will require rework. See also 3a.
3) The document tracker will be optional for many documents (I need some
clarification here). This will invalidate any meaningful statistics
pulled from the tracking numbers. I.e., if you're only tracking a
portion of the documents, you cannot extrapolate to the population of
documents.
a) If you intend to track all documents you will need to contact all
authors (including those that have written privacy-oriented documents)
and convince them that tracking GIFs are a good idea since this is a
revision to current policy.
b) The existing log files will still contain more valid information.
I.e., it is better to poll a small random sample than a large biased
sample. Your population must, of course, be large enough for your
desired confidence interval, but I think the TLDP will certainly have
enough hits that this won't be a factor.
4) Tracking GIFs, though their intention may be benign, are still
tracking GIFs. Many people, not just authors, object to them on
principle alone without considering their benefits. These people can be
quite vocal about their disapproval.
I also have some privacy concerns. Truthfully, they are not the main
concern, but they do exist. For one, I and many others that I work with,
have static IP addresses. As a matter of course I turn off images in my
email clients so that trackers cannot easily validate my address or know
what I'm reading. E.g., with the current political climate many people
may not feel free reading something such as an Encrypted Filesystem
HOWTO or a GPG guide knowing that it is somehow tracked. The idea, again
for a good ultimate purpose (improving the library), seems somewhat at
odds with the philosophy (or at least my understanding) of the library.
In other words, I don't see a compelling reason to use a tracking GIF
when the existing logs can provide more accurate data.
This said, should TLDP decide to track documents via transparent images,
I would like to see the proposed logging solution to see the
implementation and suggestions for administration. Again I'm not saying
that I cannot be convinced, but I would like to hear more arguments
addressing my concerns first.
