Re: Removing Firefox Version Number

On Aug 16, 9:13 am, Matt Brubeck <[email protected]> wrote:
> On 08/15/2011 11:52 AM, Tyler Downer wrote:
>
> > Asa, would you mind clearly, and with supporting data, tests, metrics,
> > show What problem this is addressing, Who is affected by it, What our
> > goal is (long-term and short-term, and not generic terms like "getting
> > rid of versions") and How this solution addresses the problem better
> > than the solutions we already have in place.
>
> I'll try to answer some of this, at least the parts I understand.  I'm a
> Mozilla employee but not a member of the UX or Product Management teams.
>   I can't talk for those teams, but I can give my own opinions, and my
> own understanding of what they've explained previously.
>
> First, let me describe a similar situation:
>
> Before I started working on Firefox full-time, I worked at Amazon.com.
> Amazon's process is a lot like the new Chrome/Firefox release cycle.  My
> team deployed a new version of our site every four weeks, with a
> staggered process where new versions first spent time on our dev and
> staging channels before going out to the production site.
>
> Each bit of software deployed to the site gets a new version number
> every time it was rebuilt.  Many packages had versions in the hundreds
> or thousands.  Of course, our customers never knew or cared about the
> version numbers.
>
> Now, a hypothetical question:
>
> If someone proposed adding a version number to our web site and
> publishing it where customers could see it, would this be a good idea?
> My answer is no.  Some of our corporate customers might have liked it;
> they spent time filing support tickets, and liked to have the maximum
> amount of data about everything we did.  But ultimately it would have
> led to more questions than it answered, would cause customers to notice
> updates that didn't really affect them (and wouldn't be noticed
> otherwise), and would have been mostly useless because customers
> couldn't choose which version was deployed to which channel anyway.
>
> Many people in this newsgroup have correctly pointed out differences
> between a web site and a client-side application.  But Chrome and
> Firefox have both started moving to a model where many of the same
> reasons apply.  Except for the rare user who has a good reason to run
> old insecure software, all users should be on one of the supported
> channels and cannot choose which version is deployed to that channel.
> Normal users are not affected by what number appears in the UA string,
> and shouldn't have to care.  In a hypothetical world where the version
> number had never been displayed to users to begin with, suddenly adding
> it would generate more questions than it answered.

However, there is a major flaw behind your example, which has been
pointed out before. With a website, 100% of users are using the latest
version, Always. They have no choice, there is no action that needs to
be taken on their part. The website is deployed on the server, that is
it. With Firefox, we SUCK at getting people to upgrade to the latest
version. 66% or even worse, 52% of users running Firefox 5, and it's
been 6 weeks since it was released. That is a total failure. IF we had
a way of getting even 95% of total users (no Asa, not just the newest
users, all users) on the latest version, then we can think about
getting rid of the version number. But we don't, and Mozilla sees
unlikely to change anything in how Firefox updates until, oh, version
9 or 10. By that time, what numbers will we have? 30% running the
latest version?

So I understand you want to make Firefox like a website, but until you
figure out a way to reliably deploy the latest version automatically
within 24 hours to every installed user, then you need version
numbers.