Peak Chrome? Google's browser falls as Firefox, Internet Explorer stay flat | Hacker News Reader

Peak Chrome? Google's browser falls as Firefox, Internet Explorer stay flat

arstechnica.com

33 pointsabraham21 comments

These number are very different than that shown by StatCounter [1] and Wikipedia [2]. If you look at Wikipedia's comparisons, they compare data from many sources. NetApplications (the dataset used by this article) is by far the one source that differs greatly from the rest of the data.

This likely has to do with the types of audiences that visit the sites they monitor, so the data should be taken with a grain of salt.

[1] http://gs.statcounter.com/ [2] http://en.wikipedia.org/wiki/Usage_share_of_web_browsers

We at NetApplications understand that our numbers are different from others who report market share. We are the only provider (as far as we know) that does weighting by country. No one has a data source that will truly represent the world accurately (percentage of traffic coming from a geographic region that is the same as the internet usage from that region), so it's an extremely important factor to appropriately report market share by continent or globally. A simple example is that China, by far, has the world's largest population and the world's largest base of internet users. While we have a vast amount of data coming in from China, we still have to factor it to more accurately represent actual internet usage. For more information about this, please read: http://www.netmarketshare.com/faq.aspx#Methodology

Indeed. As a point to back this up, 97% of our 80,000,000 daily hits are internet explorer 7-10.

This confirms is that browser statistics are rubbish!

This mostly confirms that browser statistics vary by region, audience, sites sampled, etc.

That being said, I'd guess if we sampled Facebook and Google (+ regionally strong competitors, e.g. Baidu) we could get a quite accurate sample of the general web population ;-)

Why not Wikipedia? It seems to me to be the best source of data, in terms of 'a general site everybody visits'.

I'm wondering what the demographics of wikipedia is though - I'd not be surprised if it's biased towards students and tech people. I know for sure the rest of my family does in no way visit wikipedia as much as me and my co-workers.

That's the reason why I left it out of that list. I initially was going to mention it :-)

"Chrome has dropped 0.27 points"

Is this a joke? That graph looks like four straight (xkcd-style) horizontal lines.

Why would you expect to be able to see a deviation of 0.27 on a graph that goes from 0 to 60? That is a change less than 0.5%.

The numbers quoted aren't read from the graphs, but rather the graphs and the numbers are derived from a table of data.

Two things:

1. I wish the author would define "market share" - is it % of users, percent of pageviews via that browser, or something else.

2. The adoption visualization really shows the difference between release strategies.

The article was using the data from NetApplications which is % of users. Most % of page views, such as StatCounter, still show Chrome as the top browser.

Given the mechanics of Chrome prefetching, % of page views figures are somewhat suspect. (Yes, they try to correct for that, but are the corrections correct?)

Wouldn't prefetching actually bumps up the page view?

Yes. And when you artificially bump Chrome's page views, Chrome looks better.

Chrome is still moving up on Clicky...

http://getclicky.com/marketshare/global/web-browsers/

I switched from Firefox to Chrome out of frustration because Firefox kept breaking the tree-style-tabs extension. I used to be a big Firefox defender before that but I gave up and moved to Chrome.

One thing I've noticed over the past year is that Chrome's automatic updates break functionality (OSX) and then Chrome needs to be restarted, but doesn't realize this and prompt the user to restart. Now whenever Chrome malfunctions I check for updates and 90% of the time an update was installed and is waiting for a restart.

Yay, lets go back to pretending Alexa and it's derivatives are actually relevant to reality!

Some confidence intervals on these graphs would really make this a lot more useful...

Nice plots..what software did you use to make the plots?

The line graphs are ggplot2 in R (http://ggplot2.org/)

How can we know the numbers are correct?

Reply on news.ycombinator.com