When users perform a search or browse a category in a package manager, there is still a lot of options to choose from. In my user studies, all of the users were able to eventually install the software they needed, but sometimes installed other unrelated software first. When users tried to install an mp3 player, other software installed included an audio editor, a console media player, a streaming mp3 server, and a variety of music-related python libraries.
Package managers provide features to help users tell the difference between different search results or category items. The features in the package managers I studied were:
Short Description All of the package managers except for Apple’s App Store gave a short description of each package in the search results or category contents. In every case except for Software Center, the short description was generally a brief description of what the software is, such as “Plugins for the Totem media player” or “Volume control and mixer for KDE”.
Many of the short descriptions in Ubuntu Software Center described what the software does or is used for, such as “Listen to Music” or “Edit your videos”. This was not consistent across all packages. Many packages displayed in Ubuntu Software Center still had descriptions which described what the software is rather than what it is used for.
In Apple’s App Store, which had no short descriptions, users spent more time looking at the detailed descriptions and screen shots of applications that were completely irrelevant to them. In all package managers which had them, users skimmed the short descriptions as a first step in narrowing down the search results or contents of a category and spent much less time looking at offerings that were obviously irrelevant. This points to short descriptions as a key filter for users trying to find the package they want.
Long Description All of the package managers except for Puppy Package Manager and QuickPET provided a detailed description of the software, either in a separate pane of the main view or in a “more details” view. Long descriptions were generally very helpful to users, although the quality and usefulness varied across package managers. Greater details on my findings will be presented in a future post.
Screen shots Ubuntu Software Center, Synaptic, Mint Software Manager, and the Apple App Store all provided screen shots accompanying their long description. Not all users had comments on the screen shots but those who remarked upon them appreciated having a screen shot.
In Synaptic, the user had to click a button to load the screen shot. Most users did not click to load the screen shots. Participants who had expressed appreciation for the screen shots in other package managers but did not use them in Synaptic reported that they did not notice that there was a button in Synaptic to load a screen shot. This points to automatically loading screen shots as the better design choice for package managers.
Reviews Mint Software Manager and the Apple App Store provide a mechanism for users to leave reviews, and display user reviews in the long description area. Users that used these package managers were asked whether or not the reviews were useful for identifying which packages were the most relevant, and noted that the packages with the most reviews tended to be the ones that were most relevant to them.
However, the users indicated that whether those reviews were positive or negative did not have any effect on their decision of what software to download. One user indicated that the content of reviews was unimportant to him for free applications, since he could try them himself, but that if he were paying for the application he may examine the reviews more carefully. He also indicated that, there usually weren’t enough reviews to draw any conclusions about the quality of the application. Given these findings, a simple up-vote/down-vote mechanism may be just as effective as the review and star rating systems currently in use.
Featured Mint Software Manager and the Apple App Store provide the ability to access a category containing “featured” applications. QuickPET does not have an explicit “featured” category but it provides only a limited selection of applications, which can be considered to be all “featured” applications. As was found in the sorting of search results, the most popular items tend to be the most relevant ones. If the contents of the “featured” category were based on popularity, it seems logical that the “featured” category would be a useful means of choosing a relevant package to install.
However, the user studies did not support this logic: none of the users used the “featured” category at all. It is possible that the “featured” category may still be of value to users who are browsing aimlessly to see what is available, which is a use case we considered but did not study.
The usefulness of these features varied, with short and long descriptions being the most influential, followed by screen shots, then reviews, and finally “featured” applications. Even with the variation in usefulness, none of the features got in the way of users, and there is a weak correlation between the number of these features that a package manager supports, and how usable that package manager was considered.
Ubuntu Software Center has four of the above features, and Mint Software Manager had all of them, and the users generally considered them the best package managers they tested. Apple’s App Store is somewhat of an outlier here, as it has four of the five features listed here, but was not considered as usable. Its lack of short descriptions proved to be a hindrance to users. In a future post, I will discuss descriptions in greater detail.