Can we learn something about which package managers users prefer based on published reviews?
Given this lack of academic resources on the subject, the primary writings on the subject are found in the popular press and in reviews of specific products. For example, a number of reviews of package managers are available on the web. A number of comparative reviews also exist for the application stores used on mobile devices.
The reviews that exist are based on the opinion of a single reviewer who is typically a tech blogger. Does the opinion of that reviewer accurately reflect the opinions of end users? In a study performed at the University of Waterloo, I systematically collected data from a range of individuals using a variety of package managers. The results of testing with real users showed major conflicts with what you might believe if you based your opinions on the published reviews.
The points of comparison in the reviews are primarily the technical merits of the back-end implementation, the selection of applications, and how developer-friendly the repository maintainer’s policies are. Only one of the test users, a system administrator, expressed any preference based on the back-end implementation. No user commented on the number of applications available, or developer-friendliness of the repository maintainer. The users I tested with cared about how easily they could find the software they needed, so they could get on with their task. The quality of the search and categorization system trumped all of the considerations expressed in the reviews.
I also found that the preferences expressed by tech bloggers poorly matched the preferences of more typical end-users. For example, reviews of Synaptic Package Manager were generally very positive, but most users I tested with hated it. For example, one Softpedia review of Synaptic Package Manager makes a couple of statements about the usability of Synaptic Package Manager:
Synaptic’s GTK+ user interface is probably the most user-friendly UI I’ve ever encountered that even the most occasional Linux user can figure his/hers way through Synaptic’s pallet of functionalities and features. Let’s stick our nose a little inside the mind-boggling package database. But this shouldn’t be confusing, as it benefits of a very clever and well-organized management.
About half of our users found their way through Synaptic with relative ease, while the others struggled, finding the interface very frustrating. The organization was considered anything but “very clever and well-organized”, with Synaptic’s categorization scheme being considered the absolute worst out of all of the package managers studied.
So how good are the published reviews at identifying which package managers users prefer? To borrow a quote, “A blogger is a blogger because he writes well, not because he knows anything.”