AMO Zamboni Planning Underway

Back in November, Wil Clouser posted about some substantial changes to AMO’s codebase and development process in 2010, most notably switching from the CakePHP framework that we’ve been developing on since mid-2006 to Django, a Python framework. Those of you who have become accustomed to seeing an AMO release announcement every 3 weeks with cool new features may be wondering what we’ve been up to the last 2 months.

Our development team has been hard at work building up the Django framework to be able to interact with the existing database and read CakePHP’s sessions so that we can have both sites up in production at the same time. Once this is finished, they’ll begin porting over sections of the site to the new codebase, and ideally users won’t even realize that some parts of the site they’re using are powered by the new system.

This process will take several months, and although we’re still fixing important bugs in the existing site, we’re focusing on this new AMO 4.0 framework, codenamed “zamboni”. Since we’ll be porting over every page on the site, I’m seizing the opportunity to throw in some often requested changes and cool new features to be implemented as we move each page. We’re still working on this product plan and I hope to post it shortly for everyone to give feedback on, but I wanted to highlight a couple planned changes to some longstanding practices on AMO that might be unexpected.

“Recommended” add-ons will become “featured” add-ons

What’s the difference, and why bother? In English, there’s a slight difference between the two words. Recommendations are usually when someone suggests that you try something based on what they know about you, or when they give reasons for why someone would be interested in the recommended person, place, or thing. On the other hand, things that are featured are promoted to everyone, regardless of who they are or their interests, and often come with no explanation.

We’ve been content with keeping highly promoted add-ons labeled as “recommended” across the site, but this year we plan to add support for personal recommendations — if you tell us what add-ons you use, we’ll tell you what new add-ons to try based on collections data and what we know about add-on usage from others who have opted in. In order to differentiate between these tailored recommendations and the site-wide list of promoted add-ons, we plan to change the nomenclature to “featured” add-ons.

Daily Users will become public for every add-on

We first introduced active daily user counts for add-ons in 2008 with the Statistics Dashboard for developers. With the exception of total and weekly downloads, this data has been private by default. It was an arbitrary decision at the time, but since then, a few things have happened:

  • Daily user counts will be a lot more reliable soon. Firefox 3.6 specifies the type of update check it’s doing for add-ons, and AMO will soon only count the daily background checks, which should stop the slight daily inflation and the huge release-time inflation.
  • Other platforms like Facebook and Chrome Extensions have started displaying user counts publicly, so it’s a much more common practice now.
  • There are several lists of add-ons ordered by their number of users on the web, so it’s not difficult to estimate the number of users of an add-on in between two other add-ons that have made their stats public.
  • In a Stats Dashboard survey from July, developers overwhelming said they would be fine with this number being public:

    Chart of responses

Weekly downloads are good at telling users what add-ons are recently popular, but they don’t do a good job of indicating which add-ons are actually holding on to all of those new users. So, with zamboni, we’ll probably make the latest day’s daily user count public for all add-ons on their display pages. There would also be a new sort option for this data in search and browse listings.

These changes are still a couple months away, and much more detail on all of the zamboni product plans will be available here and on the Add-ons Blog.

  • Yuri

    Will there be an option to make the daily count private? Despite the pole, I think there are a number of developers who will not want this info public.

    • http://www.fligtar.com Justin Scott (fligtar)

      It’s possible if there is a good reason to keep this number private, but I haven’t heard one yet. The only thing I can think of is that it might reflect poorly on add-ons that aren’t actually retaining users. If you or anyone else has a reason, please let us know.

      • http://meattle.com/jay Jay Meattle

        Hi Justin! there may be business reasons not to publicly expose the daily counts, and much more relevant to non-casual developers.

        Chrome’s gallery shows the weekly count, not daily. If the decision to move ahead with public counts is indeed finalized, it *may* make sense to do weekly (or even monthly), and not daily to maintain some sort of uniformity. Daily’s tend to fluctuate quite a bit because of weekend lulls.

        • http://www.fligtar.com Justin Scott (fligtar)

          Hey Jay,

          I agree that a weekly average or weekly count would be better, as the daily fluctuations can be quite drastic. Thanks for the feedback!

  • fsync

    It would be interesting to know, out of all the people who hosted add-ons, how many chose the non-default setting for privacy of the usage stats.

    Although, to get the best info from that, I guess you’d have to run it both ways (with one value as the default in some cases, like for even-numbered add-ons, and with the other value as default in others.

  • Pingback: Categorically Popular « fligtar's blog()

  • http://shaver.off.net/diary/ Mike Shaver

    new codename sucks, no fish or peeing

    • http://www.fligtar.com Justin Scott (fligtar)

      I’m not sure what you’re referring to — we made it quite clear the fish was giving birth to add-ons and nothing else!

  • http://translate.org.za/blogs/dwayne Dwayne Bailey

    I’ve never thought about the ‘featured’ vs ‘recommended’ addon but the idea of doing proper recommendations is exciting. I commented elsewhere on this but I’d love to see addons being promoted based on geolocations. Someone browsing from Namibia or South Africa is a ripe candidate for using an Afrikaans spell checker, but that will never make a global ‘Feature’ list.

    I’d love to see this sort of locale adaptation. It might even make more sense to see if an addon is localised into the users language as an input as to whether a locale is featured for them.

  • Pingback: Revamping the AMO Statistics Dashboard « fligtar's blog()

  • http://watchesdiscounted.biz Alpis

    I've never thought about the 'featured' vs 'recommended' addon but the idea of doing proper recommendations is exciting. I commented elsewhere on this but I&#0e9;d love to see addons being promoted based on geolocations. Someone browsing from Namibia or South Africa is a ripe candidate for using an Afrikaans spell checker, but that will never make a global 'Feature' list.

    I'd love to see this sort of locale adaptation. It might even make more sense to see if an addon is localised into the users language as an input as fo whether a locale is featured for them.;