Two Firefox stories were posted during the middle of August 2005. The first story was a report by Spread Firefox that the number of downloads had reached 80 million and that Firefox "was taking back the web". The second story reported that Firefox market share had slipped in July, for the first time.
Both stories were met with a healthy dose of skepticism. The first because the measurement being reported has very little to do with adoption, and the second because the method for collecting the data may not generate statistically representative data.
Leaping to Unsupported Conclusions
However, if we are measuring "taking back the web" the count of downloads must surely take a back seat to market share as measured at the server. What was most interesting was how ready most Slashdot readers were ready to reject the statistics and come up with unsupported reasons for the downturn. Some proposed that it was the sale of new machines with IE installed that created the market share decline for Firefox, but none posted links to PC sales data for June and July. Others thought that Windows updates might be forcing users to switch back to IE, however there had been no previous discussion of what would have been a very noticeable effect of Windows updates. Finally some suggested that the IE7 beta testing was affecting the outcome, but surely that would have been reported in the referenced article as a causative effect.
Math Error Trumps Logic
Perhaps what was most interesting was the debate, spurred by the article's summary on Slashdot, of the market share number's statistical relevance. Hemos wrote of the 0.64% decline in market share, "a number change like this is statistical noise". This prompted responses that the decline in Firefox's market share was certainly statistically relevant, especially in light that this was the first ever decrease. Unfortunately the first few responses of this nature confused the 8% market share number quoted in the summary with the 7.3% decline computed by multiplying Firefox's previous market share by 0.64. This resulted in people rejecting the premise that a near 8% drop in market share deserves to be understood. Further confusing matters was Hemos contention that the measurements themselves were flawed being in direct opposition to his statements that he believed the absolute numbers.
What should have been a strong call to understand the reversal of market share trends so that they could be counteracted in the future became a victim of the same viral marketing that created the initial market share to being with. This is a textbook example of an organism so dominating its environment that it needs to eat its young.
The success of these campaigns rely in the members feeling good about their choice. That feeling is supported in a large part by the ever increasing number of people jumping on the bandwagon. If the bandwagon stops being the most popular place then many who jumped on without understanding their choice will question their choice and hop right back off. Therefore any discussion that there are significant issues, and possibly sacrifices, with being on the bandwagon must be immediately quelled.
Either Spread Firefox needs to embrace the cause of this downturn, explain it to the masses and counter its effect or it will be destined to become another of the many good ideas that couldn't overcome the power Internet Explorer's two main assets. One, being preinstalled on nearly every consumer computer sold. And two, as a result, being favored by web developers to the detriment of other browsers.