The Linux desktop has always been a balancing act between convenience on the one hand and security and privacy on the other. However, Ubuntu's recent decision to add results from Amazon to desktop searches creates such an imbalance that I wonder just whose convenience is being considered -- Ubuntu's, or the users'?
The issue erupted with the announcement that, starting with the 12.10 release next month, searches on the Ubuntu dash (menu) would display suggestions from Amazon along with results from the local system. And the launcher would include an icon to connect to Amazon.
Both these changes are part of a revenue-sharing agreement with Amazon -- possibly the first of many, considering Ubuntu's attempts over the last couple of years to do what so many have failed to do: manage and profit from the Linux desktop.
The reaction was immediate. Within two days, the news was on Slashdot. Within three, Ubuntu co-founder Mark Shuttleworth and Ubuntu community manager Jono Bacon had posted defenses of the announcements.
These defenses corrected a few mistaken assumptions. For example, Amazon will not receive information directly, but only through Ubuntu. However, neither Shutleworth or Bacon were especially successful in defending the new features. This was largely because neither noticed the objection was not so much to the features but the implementation of the idea, both generally and specifically.
Making Money on the Linux Desktop
Bacon suggested that part of the objection was due to the fact that Ubuntu stands to profit from the new features. "Some of you may have a fundamental objection to Canonical making money from Ubuntu," he says. "When I hear this feedback, I usually translate it in my mind to 'I have an objection to a company abusing a Free Software Operating System with revenue-generating content.'"
In practice, however, few who took the trouble to comment actually mention the revenue sharing, whose details have not been released anyway. Similar attempts at revenue sharing are far too common on the Linux desktop these days for anyone to consider them unusual.
Few object, for instance, to Linux Mint's brief affiliate program with the DuckDuckGo search engine, or to the Calibre ebook manager receiving income from a portal site listing DRM-free publishers. For that matter, the fact that Ubuntu itself links to the Ubuntu One cloud services, from which users may purchase additional storage, receives with very few critical comments.
Admittedly, some of Ubuntu's previous efforts to generate money from the desktop have been criticized. However, the objections were generally to the specific deal, not the practice itself. For example, Ubuntu's proposal -- which was never implemented -- to make Yahoo the default search engine had more to do with doubts about the quality of Yahoo's results compared to Google's than with the deal itself.
Similarly, Ubuntu's efforts to raise money through the Banshee music player was not criticized in general. Instead, the objections were to the fact that Ubuntu was proposing to unilaterally redirect money that would otherwise have gone to Banshee developers. Even Shuttleworth admitted that "we made some mistakes" in the handling of the matter.
Just as with these past efforts, the objection to displaying Amazon results is not about the simple fact of the deal, but with how it has been carried out.
To start with, it does not help that both Shuttleworth and Bacon insist that the new feature is not about adding advertising to the desktop when that is obviously what is happening. "We're not putting ads in Ubuntu. We're integrating online scope results into the home lens of the dash," says Shuttleworth, as if giving a technical description alters anything.
Bacon, too, attempts a redefinition, insisting that the suggestions from Amazon "are not advertising, they are search results that relate directly to the content you are searching for in the dash."
In these statements, both Shuttleworth and Bacon are technically correct. What they say is true, and no ad is displayed in the traditional sense that explicitly urges users to buy. As Shuttleworth says, "These are not banners or spyware" (although no one claims they are). Nor are they "paid placement."
But their display is still an implicit invitation, conveniently linked to a page where you can buy, which by any sensible definition is an ad. To suggest otherwise is to appear disingenuous, and only creates the appearance of deception where otherwise there would be no reason to suspect any. In attempting to anticipate critics, Bacon and Shuttleworth have evoked suspicion rather than quieting it.
Actually, Ubuntu is better defended by the original statement by Olli Ries, the directory of technology at Canonical, Ubuntu's commercial arm. Unlike Shuttleworth or Bacon, Ries simply explained that the new features "will generate affiliate revenue that we can invest back into the project," adding that "We have found affiliate revenue to be a good method of helping us to continue to invest in maturing and growing Ubuntu."
Although Ries does not defend the features, neither does he attempt to claim that they are anything except what they are.
However, even if Ubuntu representatives had been more candid about what they were doing, they would not have quieted objections altogether.
The problem is not just that for many long-time users -- including me -- the Linux desktop has been a refuge from the brand-names that clutter proprietary desktops. Or that, for such users, displaying Amazon results on the desktop seems one step beyond even the notifications and updates that are constantly clamoring for users' attention on Windows.
Instead, the problem is ironic when you consider how Ubuntu has emphasized suability and design principles in the last four years. With its new affiliate program, Ubuntu appears to have abandoned those principles altogether.
Bacon claims that the new features allow ease of use for him as a consumer. "Here Ubuntu has helped me find interesting and new content without having to perform countless Google searches, navigate through various music websites and all their advertising and other such nonsense," he wrote.
He has a point, but he exaggerates the difficulty of searching specific sites, and something of the same convenience could be had by adding a link to the desktop, or a few bookmarks, or learning the advanced search features of a search engine. In many cases, he is talking about a few mouse-clicks, although those can quickly add up.
At the same time, while I have no data, I suspect that many users will find the search results less delightful than Bacon does. Otherwise, why is Ad Block the most popular Chromium extension and Ad Block Plus the most popular Firefox add-on?
When people are doing their best not to see ads in their web browsers, I have to wonder how many will welcome them on their desktops.
Moreover, Bacon overlooks the fact that the main purpose of the dash is not for users to find consumer items. In most cases, Unity's dash is used to find local applications and files. Anything that detracts from that purpose should be avoided.
Yet after Ubuntu's insistence on simplifying the desktop, in the case of this revenue sharing feature, Ubuntu is adding needless complications to a core feature.
Bacon himself may find the new features "neatly and unobtrusively integrated into the dash," but I suspect that anyone using the dash for its main purpose is unlikely to agree. To the contrary, the Amazon results are more likely to be a distraction in average circumstances.
Another basic design principle that the Amazon results ignore is the implementation of privacy. Shuttleworth does argue that "you trust us with your data already," to say nothing of Debian, the distribution from which Ubuntu takes its packages, and the greater community which builds all the applications on the desktop.
Of course, Shuttleworth is right, but he is also beside the point. Trusting community-built applications that have often been vetted by several people is one thing, and trusting the developers of those applications with personal data about your preferences and purchases is quite another.
Moreover, basic security suggests that being reluctant to hand out information is only sensible. Not that there is any reason to mistrust the developers, either collectively or as individuals. It is simply poor security practice to give out information unnecessarily, under any circumstances.
Just as importantly, it is poor practice to default to giving out information. A more secure practice would be to have users explicitly permit each sharing of information.
Granted, Ubuntu has announced that it plans to have search results transmitted encrypted, although for now they are send in the clear. All the same, security is stronger when more than one defense exists -- and users generally feel safer when they have control.
In fact, several commenters on Bacon's blog suggest alternatives that would improve the implementation immensely, solving all the objections I have made here. One suggests that users should have to choose to share information with each query. Another suggests a separate shopping lens, so that local searches would remain uncluttered by Amazon results.
One particularly interesting suggestion is for a preference dialog for the dash, which has already been formally filed for development.
Should this suggestion be implemented, not only could local and online searches be separated, but the user could choose exactly which sites or local directories would be searched. These search specifications could be varied with each lens. Such a preference dialog would not only be far more powerful than what Ubuntu is implementing, but would answer the objections by placing control firmly in the hands of the users.
However, whether any of these suggestions will be implemented remains to be seen. Bacon suggested that giving separate permissions for each search "would be a pain," perhaps envisioning the annoyance of confirmation dialogs. Yet others might view confirmation dialogs as a small price to pay for greater control and security, while the other suggestions would make the proposed features even more powerful -- to say nothing of removing most of the objections.
The Example of KRunner
Ubuntu is not the first project to consider web integration on the Linux desktop. KDE has been doing it for several years.
Most of KDE's efforts have been in widgets, which are optional. However, as Sebastian Kugler points out, KRunner, a sort of advanced menu, implemented online search results into the desktop several years ago.
Kügle notes that the KRunner implementation included many of the principles suggested here. Although KDE does include the opportunity to link to some proprietary web services, Kügle says that "we'd never silently send them data when it's not clear to the user or explicitly asked for. Basically, we won't send anything across the net without the user explicitely requesting us to do so."
Kugler also notes that KDE has entered into an affiliate-agreement that makes DuckDuckGo the default search engine in Konqueror.
True, KRunner's implementation would need more of an interface before average users would use it. Still, it does show that a balance between security and convenience is possible.
Unity's integration of web searches on the desktop will undoubtedly improve. In fact, Shuttleworth hints at future expansions, lightly suggesting that critics might want to wait to see the final results.
I appreciate the warning, yet I'll also be rash and say what seems obvious. Unity's integration of web searches on the desktop is a promising idea, but has so many shortcomings that it deserves serious questioning. With any luck, in another release or two, that questioning will result in a feature that balances security and convenience more successfully than the current implementation.