"What happens when we do security updates? (usage goes down as anti-virus software updates get propagated, but recovers over the following week or 2)"
Do you mean that anti-virus software is suspicious of the new version of Firefox and prevents it from running or from connecting to the Internet? If so, eep! What percent of Firefox users does that happen to? How long does the effect last? What can we do about it?
John
· 1 year ago
hey jesse -- actually, we don't know. what we do know is that in the days following new dot releases, we see a clear decline in usage -- 10% or more -- and it takes a few days to recover. working theory at the moment is that it takes a while for the anti-virus programs to recognize the new firefox is valid -- we don't really know though, so that's a good example of a question that tracking the stats raises that we need to figure out.
seth
· 1 year ago
Great post. Are there other values for multipliers other than 3? If 3 is a conservative multiplier, what is the aggressive multiplier? 3.5? Or, do some think it could be higher? Might be interesting to see the range of total users based on the conservative to aggressive multipliers.
You also mention that most all websites use a multiplier of 3 to get from ADU to total users. Just interested in learning more about the calculation of this multiplier number. Seems like an interesting calculation to understand more deeply. If you have any others sources that you trust/like and discuss this calculation, please share.
Alex Polvi
· 1 year ago
> What happens when we do security updates? (usage goes down as anti-virus software updates get propagated, but recovers over the following week or 2)
Another consideration is that Firefox does not check for updates after it already has one. This will cause users not to ping until they upgrade. Another conclusion would be that users are clicking "later" and not restarting their browser for a little while.
What's really cool about this data, regardless of the total or "multiplier", is that Firefox use is growing strong.
VanillaMozilla
· 1 year ago
Judging from support requests on MozillaZine, Internet security software continues to be a big problem. Some of it has some nasty bugs and pitfalls, and I'm not sure there's any guarantee that this stuff gets updated or fixed. That is, it might be totally up to the user to fix, in which case, much of the damage would be permanent.
There's a bug report on this ( https://bugzilla.mozilla.org/show_bug.cgi?id=33... ) that may help some, if it ever gets fixed. It might also be a good idea to buy some popular Internet security suites and give them a thorough shakedown.
Alan
· 1 year ago
"Why are so many people in China using the English version of Firefox?" Ooh, ooh, pick me! Theory: because of the Firewall in China, a high percentage of users use proxies (which tend to be in English-speaking countries). Depending on how mozilla.com determines a visitor's country (and therefore, the proper language version to offer for download), it might offer them the English version. The "Other Systems and Languages" link is not super-prominent, so they take the English version.
Harley
· 1 year ago
Also, remember that for people who get Firefox as part of a Linux distribution, automatic update for Firefox itself is disabled. This is because the Linux distribution itself will manage updates, and there is no need for Firefox to also manage updates.
So, it seems possible that using this methodology, nearly all Linux Firefox users may be uncounted.
Alan Bell
· 1 year ago
my 3 Ubuntu computers all have Firefox with updates arriving from the Ubuntu repositories, not direct, so I am using a recent build but you won't see me in your stats.
Seth Galitzer
· 1 year ago
I'm with Alan Bell. Gentoo manages updates the same way. In fact, the "check for firefox updates" is disable in the application so the distro can manage the updates. I bet your numbers are actually much higher than what you're seeing.
Carl Parrish
· 1 year ago
Another Linux user (Fedora), I'm sure your numbers are much higher than what you're seeing. May want to check with the major distros and see if they have stats on FF users.
John
· 1 year ago
hey carl, seth, alan, harley -- you're all exactly right -- i should have listed linux distros as a significant factor in the "firefox dark matter" in the world -- it exists, but we don't have ways to see it. linux use certainly accounts for at least single-digit millions every day (and maybe more), but it's hard to say exactly how many. easier to approximate this number than it is folks using it behind enterprise firewalls, though.
Alan Lord
· 1 year ago
Another couple here. FF on Ubuntu and FF built from source for my "Linux From Scratch" server based boxes. (I turn off automatic updates in the .mozconfig file). So there's two more here...
Tuxi
· 1 year ago
Are you saying that behind corporate firewalls you may have several users that get counted as one? If so, my several Linux machines and three Windows machines at home are likely counted as one because of the router. For my home anyway, the multiplier is much greater than 3.
Franadora
· 1 year ago
Another factor - I (one person) use Firefox at work (part time - internal corporate applications require IE6) as well as on two machines at home. And have at least two users on each of my home machines (one Administrator, one Limited). So - is there a separate AUS ping for each user on the same machine? Then I would probably account for up to 5 or maybe even 6 pings on some days (usually Mondays or Tuesdays when I traditionally check Windows updates.)
But even if there is not, you would get 3 user pings for one real user.
Gostak
· 1 year ago
The ping hit is a minimum number. I don't want my software to talk to the world without my knowledge. Updates get queried and installed at ***MY*** request, and then only after I let the query through my firewall!
Further, how do you count where Firefox is installed on both a Windows and a Linux partition? Is that one or two installs?
Personally I would count it as two.
Ed
· 1 year ago
I know some users who freak out whenever any security update comes, and stay off the Internet entirely until "the shakes start" - frequently around a week or so. On at least one of my computers, I only do updates if either I'm at a convenient point to close all of my windows * tabs, or I need to go to an untrusted site. On another computer, I only run updates that fix problems that would affect me (javascript bug? That computer doesn't have javascript enabled, so no biggie.)
I'd suspect that the firewall thing you mention is actually specific to proxying firewalls, which could cache the presence of an update, and therefore each actual ping you get could be one user or 10,000 users. However, most home firewalls, being simple packet filters, would have no impact.
Benjamin
· 1 year ago
John, The difficulty in measuring market share of open source applications is one the OpenOffice.org, each Linux distro, and many other communities have grappled with. If you were to generalize Mozilla's experience to create a strategy paper for other projects to follow, what would it look like? Is there a chance Mozilla would be interesting in participating with some other projects to develop their own capabilities similar to yours?
-Ben
oscar
· 1 year ago
Same with Debian, automatic updates of Iceweasel are disabled and handled via package managers. And as Firefox is the standard browser in all but a few Linux distros, you're probably missing a few million of us.
It might be a good method to count windows and mac users based on how many times a given update, say e.g. 2.0.0.8 has been downloaded. How do these compare to the update ping numbers?
VanillaMozilla
· 1 year ago
If we lose 10% of users temporarily with each update, it wouldn't be at all surprising if we lost 1% permanently with each update. That's a lot, but you would probably never notice from ping statistics.
We see a lot of users on the support forum who claim that they have configured/turned off/uninstalled their firewall, so it couldn't be a firewall problem. With sufficient persuasion most of these users eventually fix their firewall and eat their words, but that level of support is impossible for the vast majority of users. It might be a good idea to devote some resources to this problem.
Rafael
· 1 year ago
Now if you apply either $1, $4, or $40+ per user per year, that $80MM a year is kinda chump change.
Nice problem to have though.
Ask Adobe or Emmy for total number of Web users. They're within 5-10% of having every user on the web world-wide. They had been public about the numbers in the past.
Nick H
· 1 year ago
I think a lot of people are getting distracted with the nitty gritty details of converting the pings into absolute number of users. Yes, MoCo has no magical way of doing an absolute conversion from software update pings, to ADU, to users.
I do appreciate John providing so many details into how the .125 billion figure was calculated; however, the real news is that the software update pings have doubled relatively quickly. I think it's safe to assume that this is roughly proportional to ADU over the span of this measurement.
Double is double, regardless of the exact number you're doubling.
Michael Vincent van Rantwijk
· 1 year ago
I wonder why MoCo isn't comparing the IP numbers used for downloads (for installations) with the IP numbers used for updates???
Simon
· 1 year ago
Firefox comes with a default live bookmark that points to the BBC News RSS feed.
Have you asked bbc.co.uk what are the usage stats they receive?
Jay@Bangkok
· 1 year ago
Dear webmaster,
I am in Bangkok and study in IT so I was assigned to find the open source. Then I thought that Firefox is interested for presentation. And I think I would like to present the statistic of download of Firefox. However, I don't have data for compare with other browser , so would you please get the statistic of download of Firefox to me?.
Best Regard Jay
Brian Masinick
· 1 year ago
Perhaps I overlooked it in the article and the comments, but are there also any rough statistics, among the Firefox downloads and/or the usage patterns of how many Firefox users are on the various platforms that are supported: UNIX, Linux, BSD, Windows... ?
Kamahl
· 1 year ago
Simple. My firewall (McAffe) will block any unknown program from accessing the net without my permission. I had to turn on the warnings, otherwise they are silent. It mesures a 'safe app' by checking the md5sum against it's allowed database. (big!) It takes a day for someone to accept the new checksum, and add it to the database, and another day for all users to have downloaded the new database. that accounts for two or three days of blocked firefox after every update!
Romi
· 1 year ago
OMG, there's already 5 trackback from Indonesian blog here, most of them mention that Firefox users is half the Indonesian population.
"Why are so many people in China using the English version of Firefox?" In Indonesian, all users are using en-US version, mainly because it's just right on your face when you go to the download page. None are complaining, since we're already using with English Windows, and (most?) of them is pirated version.
Internet bandwidth is still expensive for the most of us, not counting limited International bandwidth, but I guess most of the 'techies' and bloggers are using Firefox. Recently, I asked the stats from a big forum and Firefox is already surpassed 50%. There is stats from a 6 month old Indonesian blogging service that speaks the same.
Bruce A. Wittmeier
· 1 year ago
My hat goes off to the Mozilla team. It wasnt enough that MS couldnt compete on a level playing field and began giving away the browser, but then to attempt to integrate it into the O/S was beyond belief.
The stamina and professionalism of the Mozilla team deserves an award for perseverance and development of a superior product.
Bruce
Nelson B
· 1 year ago
Why does market share effectively help the spread of Web standards everywhere but Korea? Why not Korea? (Presumably you mean South Korea). Is it: - because FF has NO market share there? - because FF owns the market there? - something else?
johnolilly
· 1 year ago
Ah -- I said that we're effective at it everywhere but [South] Korea (and presumably North, too, but that's not really what i meant.)
In South Korea, there are some government regulations & crypto techs that are used by the government that require ActiveX implementations, and so it's been very very hard for us to get any significant traction -- as a result , it's harder for us to spread the OpenWeb message.
unknown
· 10 months ago
I removed firefox this morning for downloading files without my express permission. I really like how they removed the option to disable automatic updates
All Comments
Do you mean that anti-virus software is suspicious of the new version of Firefox and prevents it from running or from connecting to the Internet? If so, eep! What percent of Firefox users does that happen to? How long does the effect last? What can we do about it?
You also mention that most all websites use a multiplier of 3 to get from ADU to total users. Just interested in learning more about the calculation of this multiplier number. Seems like an interesting calculation to understand more deeply. If you have any others sources that you trust/like and discuss this calculation, please share.
Another consideration is that Firefox does not check for updates after it already has one. This will cause users not to ping until they upgrade. Another conclusion would be that users are clicking "later" and not restarting their browser for a little while.
What's really cool about this data, regardless of the total or "multiplier", is that Firefox use is growing strong.
There's a bug report on this ( https://bugzilla.mozilla.org/show_bug.cgi?id=33... ) that may help some, if it ever gets fixed. It might also be a good idea to buy some popular Internet security suites and give them a thorough shakedown.
Ooh, ooh, pick me!
Theory: because of the Firewall in China, a high percentage of users use proxies (which tend to be in English-speaking countries). Depending on how mozilla.com determines a visitor's country (and therefore, the proper language version to offer for download), it might offer them the English version. The "Other Systems and Languages" link is not super-prominent, so they take the English version.
So, it seems possible that using this methodology, nearly all Linux Firefox users may be uncounted.
But even if there is not, you would get 3 user pings for one real user.
Further, how do you count where Firefox is installed on both a Windows and a Linux partition? Is that one or two installs?
Personally I would count it as two.
I'd suspect that the firewall thing you mention is actually specific to proxying firewalls, which could cache the presence of an update, and therefore each actual ping you get could be one user or 10,000 users. However, most home firewalls, being simple packet filters, would have no impact.
The difficulty in measuring market share of open source applications is one the OpenOffice.org, each Linux distro, and many other communities have grappled with. If you were to generalize Mozilla's experience to create a strategy paper for other projects to follow, what would it look like? Is there a chance Mozilla would be interesting in participating with some other projects to develop their own capabilities similar to yours?
-Ben
It might be a good method to count windows and mac users based on how many times a given update, say e.g. 2.0.0.8 has been downloaded. How do these compare to the update ping numbers?
We see a lot of users on the support forum who claim that they have configured/turned off/uninstalled their firewall, so it couldn't be a firewall problem. With sufficient persuasion most of these users eventually fix their firewall and eat their words, but that level of support is impossible for the vast majority of users. It might be a good idea to devote some resources to this problem.
Nice problem to have though.
Ask Adobe or Emmy for total number of Web users. They're within 5-10% of having every user on the web world-wide. They had been public about the numbers in the past.
I do appreciate John providing so many details into how the .125 billion figure was calculated; however, the real news is that the software update pings have doubled relatively quickly. I think it's safe to assume that this is roughly proportional to ADU over the span of this measurement.
Double is double, regardless of the exact number you're doubling.
Have you asked bbc.co.uk what are the usage stats they receive?
I am in Bangkok and study in IT so I was assigned to find the open source. Then I thought that Firefox is interested for presentation. And I think I would like to present the statistic of download of Firefox. However, I don't have data for compare with other browser , so would you please get the statistic of download of Firefox to me?.
Best Regard
Jay
It mesures a 'safe app' by checking the md5sum against it's allowed database. (big!)
It takes a day for someone to accept the new checksum, and add it to the database, and another day for all users to have downloaded the new database. that accounts for two or three days of blocked firefox after every update!
Indonesian population.
"Why are so many people in China using the English version of Firefox?"
In Indonesian, all users are using en-US version, mainly because it's just right on your face when you go to the download page. None are complaining, since we're already using with English Windows, and (most?) of them is pirated version.
Internet bandwidth is still expensive for the most of us, not counting limited International bandwidth, but I guess most of the 'techies' and bloggers are using Firefox. Recently, I asked the stats from a big forum and Firefox is already surpassed 50%. There is stats from a 6 month old Indonesian blogging service that speaks the same.
The stamina and professionalism of the Mozilla team deserves an award for perseverance and development of a superior product.
Bruce
Why not Korea? (Presumably you mean South Korea). Is it:
- because FF has NO market share there?
- because FF owns the market there?
- something else?
In South Korea, there are some government regulations & crypto techs that are used by the government that require ActiveX implementations, and so it's been very very hard for us to get any significant traction -- as a result , it's harder for us to spread the OpenWeb message.
I really like how they removed the option to disable automatic updates