In dropping fallback mode, GNOME made the right decision -- and utterly failed to communicate that fact.
You have to pity the GNOME project these days. Even when it does the right thing, it does so in a way that maximizes controversy.
I'm talking about the project's recent announcement about dropping support for fallback mode. Since it was first introduced with the GNOME 3.0 release the fallback mode has provided an approximation of the GNOME 2 desktop for users who lacked the hardware acceleration needed for the latest desktop environment. Now, GNOME developers have announced that the upcoming 3.8 release will not include the fallback mode.
From any perspective, this decision was correct. Although described in the GNOME 3 release notes as "an excellent experience [that] incorporates many of the improvements contained in the release," fallback mode has actually been a crippled version of the last GNOME 2 releases.
Far from attracting users, fallback mode appears to have been widely regarded as a contemptuous gesture to those who either lacked the necessary hardware or who preferred not to use the proprietary drivers needed for the hardware acceleration required by the GNOME 3 release series.
Part of the problem may have been, as GNOME developers suggest, that "Some distributions labeled this mode as something other than fallback mode. This caused some grievance as it gave an impression that it was intended to provide a GNOME 2 experience."
But whatever the case, few are likely to miss fallback mode. When the news reached Slashdot, one commenter described fallback mode as "non-functional." The most positive comment was, "I hate fallback mode, but it's better than nothing."
Moreover, while fallback mode was inadequate from the start, in the last year, GNOME Shell extensions, and Linux Mint's Cinnamon and Mate have all begun offering an experience far closer to that of GNOME 2.
How much such considerations played in the decision is uncertain. For GNOME developers, a large part of the decision was simply practical. According to the initial proposal, fallback mode was not being tested and was simply in maintenance mode.
Just as importantly, an increasing number of applications, including Totem, Empathy and Cheese no longer worked in fallback mode, and some core features, such as keyboard configuration, would require an overhaul if they were to continue to work in fallback mode.
In addition, as users of the latest Fedora release have found, a new technology called llvmpipe offers a way around video drivers without hardware acceleration.
Despite a small reduction in performance with llvmpipe and its lack of support for a few chip architectures, the efficiency of maintaining only a single desktop environment speaks for itself.
Admittedly, GNOME never made clear to users (although they might reasonably have guessed) that "fallback mode was always meant to be a temporary stopgap."
Still, the alternative would have been to increase efforts to support something that simply "does not provide an acceptable experience" and was no longer needed anyway.
Despite all the reactions to the decision as GNOME's abandonment of users, it was obviously nothing of the sort. The switch to llvmpipe improves user experience, and the decision to drop fallback mode was a well-considered decision made over several months.
True, you might argue that GNOME should never have implemented such a half-hearted decision as fallback mode in the first place -- or perhaps not depended on hardware acceleration. But at the very least, the project is finally correcting its original mistake.
The Marketing Fail
Properly handled, this decision could have been at least a minor marketing triumph for the project. Ever since the start of the GNOME 3 release series, some users have complained that GNOME ignored users and that its developers cared more about their own vision of the desktop than users' needs.
Yet here was a case where GNOME was making a decision that benefited users as much as developers. Without any exaggeration whatsoever, the project might have announced the decision as an improvement of the user experience.
The project might then have suggested that users nostalgic for GNOME 2 check out GNOME Shell Extensions, perhaps providing a list of a half dozen or so extensions needed to re-create -- so far as that is possible --GNOME 2 on top of GNOME 3.
In fact, Emily Gonyer did blog along these lines after the news appeared. However, Gonyer is a relative newcomer to the project, and her comments went unreported.
But from the GNOME project itself? Nothing. No news release was issued about this moderately important piece of news. The Marketing team, if any of its members were aware of the stakes, never discussed the possibilities on its mailing list, much less did anything.
Instead, the final decision surfaced in a posting to the release mailing list -- a perfectly reasonable choice for a purely technical decision, but one that removed all control of a potentially important marketing message.
Meanwhile, what were GNOME's leaders talking about? Most were not talking about the project's image problem at all. And those who were discussing it were mostly complaining about the problem and offering excuses for it.
Karen Sandler, GNOME's executive director, expressed the "disappointments" of recent reporting about GNOME. In particular, she was responding to Linus Torvalds' latest comments about KDE and GNOME.
Then she pleaded that "it really takes time to get things right" and, citing the fact that GNOME 2 also took time to be accepted, added that "this is how that happens in a true free software community run project -- through slow incremental improvements that may only be acknowledged as afterthoughts."
Federico Mena-Quintero expressed similar sentiments in his blog, but more aggressively. According to Mena-Quintero, developments in GNOME have been fiercely attacked since the project's earliest days.
Although he acknowledged that many of the attacks on GNOME have been from ordinary people, Mena-Quintero particularly blames "yellow journalists" -- poorly informed writers who give the most sensationalist interpretation to every development.
Calling them "hater bloggers with a job," Mena-Quintero went on to say, "They pick up the latest flamewar, however minor, and make a big deal out of it. They summarize blog posts and quote things with not enough context . . . . They predict the decline and fall of a software project because there is a flamewar going on. They build an ongoing, not entirely consistent, self-serving narrative of the soap opera that they want free software to be."
The point is not whether such comments are justified. The point is that they are poorly timed. At a time when GNOME had the opportunity to deliver positive news, the chance to counter the usual criticism with something concrete, it missed the opportunity -- so thoroughly that I can only assume that GNOME's leaders were not even aware that there was an opportunity.
The result was inevitable: When the news was picked up by sites such as LXer and Slashdot, the reporting lacked any perspective from the GNOME project.
By default, the stories became another center for the same old discussions by users annoyed by GNOME 3: suggestions about what other desktop environments to try, criticisms of GNOME's design philosophy and comprehensive dissections of the problems with recent GNOME releases.
None of the comments were in any way new. However, they probably became that much truer in many people's minds because of the repetition -- all the more so because no one from GNOME offered any opposing views.
Of course, adding a GNOME perspective would not have entirely eliminated such comments. But at the very least, GNOME's input might have focused the discussion on fallback mode rather than on general grievances.
Just as importantly, input from GNOME would have provided an alternative viewpoint. But the total absence of comment only makes the often-repeated criticisms seem truer. From the perspective of GNOME members, it is harder to imagine a worse outcome.
A Mixed Problem
I don't want to exaggerate the important of dropping fallback mode. Still less do I want to attack GNOME mindlessly. But I have gone into some detail on the subject because I believe that GNOME is missing opportunities. Nor is it the only project to do so.
Yes, marketing is a foreign mindset to many developers. Many see it as a form of lying. Yet this was a case where a project could have said nothing except the truth and perhaps helped itself -- but apparently failed to see the opportunities, much less take advantage of any of them.
Instead, as the opportunities slipped away, hurt feelings were expressed. Such feelings might have been understandable when the first complaints surfaced, and the members of GNOME were disappointed by the reception to all their hard work. But two years later, the expression of hurt feelings has become as routine as the criticisms for which they are meant as a response. Even more importantly, they are unconstructive, if not outright harmful.
GNOME cannot control what its members say in public. Its members would understandably object if it tried. But, like so many other free software projects, it is badly in need of some official responses, or at least some coordinated ones.
The brutal truth is that GNOME has a marketing problem as much as a developmental one. And, until it realizes the fact, no amount of counter-attacks or analogies to the past will help it move beyond the user revolt that has already lasted far too long.