January 24, 2005
the graphs (part 1)
One of the most frequent requests I get from people here and over at SpreadFirefox.com are for more detailed download information. The cumulative download graph is the most often requested and fairly easy to put together.
This graph shows the total number of downloads since the day of the release. As you can see, the line is pretty darned straight. That's highly encouraging to me as we've never before seen this kind of linear growth over this length of time with any prior Mozilla product releases.
In this graph, it's not real easy to see, but things had started to dip a bit right before the New York Times ad ran on December 14. The slope turned upward a bit for the following two weeks. Then it fell back a bit over the Christmas and newyear holidays but came back with a nice kick as people returned to school and work in 2005. That uptick is visible starting around January 4th.
The line runs up through early this morning when we hit 20,000,000 downloads. That's shy of 11 weeks since the 1.0 release.
I've got a daily totals graph that I'll be posting next so stay tuned.
Posted by asa at January 24, 2005 01:14 AM
Perhaps it's time to get a perl script going, and input the data daily... then people can query the data at any given period of time?
Nice toe on that curve! Sex-ay.
How many lines of code per nanometer?
(Sorry, darkroom chemistry.. in.. brain.. @_@)
Seriously though, that is a great graph and I will gladly use it when pitching webdesign standards to clients. Hope to see more of the same, thanks Asa!
Up-up in a way!!! FIREFOX!
Although you are seeing a linear increase in this chart, to see growth, switch the Y scale to log (base 10). You will see the rate of growth, and it is actually slowing. True linear growth would curve upward at the end of the graph.
FIREFOX is great whatever way you look at it.
Interesting comment Walt, never thought of it that way before.
Based on this data, is it safe to assume that in order to keep this growth, Firefox will have to do an advertising campaign every 3-6 months or so?
I very much doubt its possible for Firefox to keep this rate of growth. If it continues like this much longer, EVERYONE will be using Firefox. Who are we going to steal from M$ then?
What you have to remember from this graph is that it is software downloads, not actual users. 20M downloads != 20M users. Multiple machines, reinstalls, etc. Play a fairly significant role in these numbers. That means there's still plenty of growth headroom.
Nothing against Firefox but a cumulative chart is worthless. You could have one download a day and the chart would show a steady increase. You need a bar chart with the total for each week as a seperate bar.
OK, so the second page is what I was looking for the but comment about page 1 is still true.
Dear Walt Matwijec,
I'm not sure I see what you mean by growth must be logarithmic (Y scale marked in log10 scale) or your "True linear growth would curve upward at the end of the graph."
In the current universe one can have exponential growth, where the growth is a (near) straight line on a logarithmic scale - your "Y scale to log (base 10)." One can also have a geometric growth, where growth is a (near) straight line on a quadratic scale with y= ax + bx^2 + c. AND one can have a LINEAR GROWTH pattern, where growth is a (near) straight line on a graph with y = mX + b scale.
There are other growth patterns but they are, for the most part, just modifications of these three.
While this graph may not show true growth in Firefox's download (see Mike Z's post), it does show growth which does appear to be linear. If it keeps up a linear growth pattern, one day all users will at least have download Firefox.
If the growth curve were geometric then the time where all users will have downloaded Firefox would be much shorter. A a exponential growth curve would be even sooner! At least in "practical terms" that is P-) It's not likely that "everyone" will eventually download Firefox. There will always be "religious" fanatics who will never change, no matter what the benefits in security or usability are.
Actually, a log graph won't tell you about rates of growth in a meaningful way. To do that you need to take the derivative of the number of downloads, ie how quickly the number downloaded per day is changing. (The first derivative, in maths terms.) In fact, that's what I wanted as soon as I saw this.
So, two more graphs: number of downloads per day; percentage difference in the number of downloads over (period) compared with (previous period). In this, (period) can be day, week, month..
See how Business Week does graphs of economic growth, sales, etc. Sometimes they're doing second or third derivatives - the rate of change of the rate of change, or even the rate of change of accceleration.
Thanks for all your Fire Fox Sites
I am going to recomend you to all my friends
Keep up the good work
Regards From Stephen Bunce