There’s an important lesson to be learned from the whole PackageKit episode.
That lesson: there is an important difference between being transparent and being communicative. Transparency is good, but sometimes it is not enough. Some issues must be discussed proactively. Sometimes, one must go and solicit feedback aggressively.
In reading through the bugzilla comments about this issue, the most insightful comment I came across pointed to a mailing list thread for PackageKit. The only two participants in that thread were Richard Hughes and David Zeuthen, the two Red Hat engineers who were most responsible for the changes to PackageKit’s default behavior.
Were they making these decisions behind closed doors? Demonstrably not. Some people seem to believe that davidz and hughsie colluded to “sneak” a change in. Examination of this thread reveals that:
1. They certainly were not in collusion, since the discussion occurred on a public mailing list.
2. They certainly were not pushing a unified agenda, since the conversation looks exactly like any important technical conversation should, with appropriate give and take.
3. The conversation was limited to two people, which was not nearly enough input to make a decision of this magnitude.
It’s easy in retrospect to see how all this happened. I’ve been in this position, too: you discuss a change very publicly, you assume that everyone who cares about the topic is paying attention, you make a decision, and then when the change hits, people go nuts in a very public way. It sucks. But it’s also a good opportunity for reflection.
Some changes are really important — more important than they may seem when you’re down in them — and it’s vitally important to solicit feedback actively for those changes. It’s an excellent demonstration of the importance of the Fedora feature process — which exists precisely to mitigate risks like this. Big changes should never, ever be a surprise.
Well-meaning people make mistakes. Especially people who want nothing more than just to Get Things Done. That’s one of the strengths of our model: we make mistakes in a way that allows us to recover from them, and if we’re smart, to learn from them. I think we’re seeing a lot of learning now. That’s a good thing.