September Google Zeitgeist

[jgraham]

For the purposes of this discussion, I took it from the Google Zeitgeist graph and, in truth, Gecko has less than 5%.

In that case, I'm confused as to what your 5% means. I assumed "5% variance" meant that the factors mentioned wouldn't affect any of the numbers by more than 5% of their value. Here, you seem to be stating that that none of the factors are enough to push 'true' gecko usage to above 5%. I'm more interested in the first figure - how reliable are the google results? If we don't know that (at least approximatley) there's no use in trying to work from them.

Now, we can argue whether these stats are accurate based on tabbed browsing or other factors ... but the metrics are what they are and until there is an industry acceptable alternative, these are the ones that stand.

My point is entirely that the google stats should <strong>not</strong> to be considered "industry acceptable", if "industry acceptable" is also to imply accurate and useful. The google stats are not, because they are a single number coming from a closed box - we have no information about how those numbers are generated and what we can guess suggests they are very prone to misinterpretation.

Fine, I'll even accept that but the point I've been trying to make all along is that the non-IE browser-share has NOT increased to a point to where their plea for "standards compliance" has any weight and for the Evangelism folks to dellude themselves into professing that they are close to being there is a problem in itself.

I don't follow. Are you saying that 5% market share (if that is accurate) is not enough for companies to bother catering for alternative browsers. That certianly should be high enough - after all very few businesses will arbitarily turn away 1 in 20 customers. Or, are you saying that the people stating Gecko will have a very large market share sometime in the near future are wrong? That sounds right - unless switching browsers can be made more compelling (or much easier e.g. it is easy to switch from IE/Mac to Safari simply by buying a new Mac or by buying OSX 10.3) then IE will retain a marketshare that is, by all accounts several factors that of the competition.

Ignoring the google stats, there is positive news: recently there have been high-profile sites switch from IE/NS4 only browser policies to more liberal policies which include Mozilla. For example http://www.argos.co.uk and http://www.marksandspencer.com/ both now accept gecko browsers having rejected them for years. I don't claim this represents conclsive proof of anything (you could probably pull some stats from bugzilla if you think that would help), but it does provide some evidence that companies, even those who are slow to adapt to a changing browser environment, still percieve alternative browsers as worth supporting, whatever their exact marketshare.

[Sailfish]

Here, you seem to be stating that that none of the factors are enough to push 'true' gecko usage to above 5%.

Yes.

I'm more interested in the first figure - how reliable are the google results? If we don't know that (at least approximatley) there's no use in trying to work from them.

As I stated, the stats are what they are and unless someone can point to another set of stats that are used by the industry, these are what we have to go by.

My point is entirely that the google stats should <strong>not</strong> to be considered "industry acceptable", if "industry acceptable" is also to imply accurate and useful. The google stats are not, because they are a single number coming from a closed box - we have no information about how those numbers are generated and what we can guess suggests they are very prone to misinterpretation.

While they may not be as accurate as they can be, my guess is that a whole lot of people in the industry use them to assess browser-penetration and, in that case, the perception they convey becomes the reality.

I don't follow. Are you saying that 5% market share (if that is accurate) is not enough for companies to bother catering for alternative browsers.

One, I don't believe the aggregate 5% of "other" browsers are all "standards compliant."

That certianly should be high enough - after all very few businesses will arbitarily turn away 1 in 20 customers.

As I stated before, most folks on "other" browsers tend to be on the geeky side. While this may influence hi-tech sites, its probally not enough for the broader comsumer companies to worry about. Actually, I'm not sure how much technology-related companies worry about it, considering that CNET, for example, doesn't even include a DOCTYPE statement?

Or, are you saying that the people stating Gecko will have a very large market share sometime in the near future are wrong? That sounds right - unless switching browsers can be made more compelling (or much easier e.g. it is easy to switch from IE/Mac to Safari simply by buying a new Mac or by buying OSX 10.3) then IE will retain a marketshare that is, by all accounts several factors that of the competition.

This, too.

The real thrust of my argument is that if the evangelists continue to fool themselves that the "stats" are wrong and that gecko browser-share is growing at ever increasing rates then it will continue to detract them from looking seriously at what needs to be done to actually make those predictions come true.

Ignoring the google stats, there is positive news: recently there have been high-profile sites switch from IE/NS4 only browser policies to more liberal policies which include Mozilla. For example http://www.argos.co.uk and http://www.marksandspencer.com/ both now accept gecko browsers having rejected them for years. I don't claim this represents conclsive proof of anything (you could probably pull some stats from bugzilla if you think that would help), but it does provide some evidence that companies, even those who are slow to adapt to a changing browser environment, still percieve alternative browsers as worth supporting, whatever their exact marketshare.

And not to sound overly negative, I agree that incremental movement towards standards-compliance is happening and will continue to do so. I just don't believe that this has yet turned into a ground-swell movement.

[jgraham]

As I stated, the stats are what they are and unless someone can point to another set of stats that are used by the industry, these are what we have to go by.

Many and perhaps even most people, quite reasonably, use the stats for their own site. That's a far more reasonable meausre, since it at least narrows the demographic down to 'people who actually visit the site'. Of course it doesn't mean that the stats are accurate in the sense that they represent the number of people using each browser, but that's a very difficult number to measure.

While they may not be as accurate as they can be, my guess is that a whole lot of people in the industry use them to assess browser-penetration and, in that case, the perception they convey becomes the reality.

Then the idea that google stats are unreliable needs to be spread. At least, if someone quotes a figure for browser useage at you (e.g. "We won't support Mozilla because it's only used by 1% of people") then ask where the figure comes from.

One, I don't believe the aggregate 5% of "other" browsers are all "standards compliant."

Yeah, but they're not IE either. From the point of view of preventing lockin, cross browser compatibility in any form should be enough - precise standards compliance is not necessary (although many people acheive something close to both).

That certianly should be high enough - after all very few businesses will arbitarily turn away 1 in 20 customers.

As I stated before, most folks on "other" browsers tend to be on the geeky side. While this may influence hi-tech sites, its probally not enough for the broader comsumer companies to worry about.

So? 1 in 20 is still 1 in 20 whether or not all of those people are 'geeks'. I expect that many consumer companies (Amazon, say) have a very wide visitor demographic ranging from geeks to grandmas.

The real thrust of my argument is that if the evangelists continue to fool themselves that the "stats" are wrong

The stats are only wrong in the sense that it is impossible to draw right conclusions from them. It may be in fact that all the uncertianties cancel the numbers are a perfectly accurate profile of actual web users. However, this seems very unlikely. Without knowing that systematic problems in the data collection have been accounted for, it is impossible to use the data with a reasonable expectation of reliability.

and that gecko browser-share is growing at ever increasing rates then it will continue to detract them from looking seriously at what needs to be done to actually make those predictions come true.

I don't believe that the marketshare is exponentially increasing is true because I haven't seen anyone producing convincing evidence that it is true. So yes, anyone who believes that gecko share is surging either has data that I haven't seen (in which case, they should present it) or is deluded. As for making these predictions come true - I think many people have ideas about what needs to be done, but people have limited resources for turning ideas into reality.

[Sailfish]

Many and perhaps even most people, quite reasonably, use the stats for their own site. That's a far more reasonable meausre, since it at least narrows the demographic down to 'people who actually visit the site'. Of course it doesn't mean that the stats are accurate in the sense that they represent the number of people using each browser, but that's a very difficult number to measure.

While those are useful, most companies would be foolish to use them as their primary metric. Why? Because they are mostly seeing stats from existing clients/customers. If they are interested in growing their market share, they must use/rely on an industry stat for making site expenditure decisions, imo.

Then the idea that google stats are unreliable needs to be spread. At least, if someone quotes a figure for browser useage at you (e.g. "We won't support Mozilla because it's only used by 1% of people") then ask where the figure comes from.

Good luck, Don Quixote .. er .. jgraham. Just make sure you can discern windmills from Google stats

Yeah, but they're not IE either. From the point of view of preventing lockin, cross browser compatibility in any form should be enough - precise standards compliance is not necessary (although many people acheive something close to both).

The point I was making is that only the "standards compliant" share will have any effect in moving the industry. For example, my guess is that even today, Communicator occupies a very significant portion of the dubious 5%.

So? 1 in 20 is still 1 in 20 whether or not all of those people are 'geeks'. I expect that many consumer companies (Amazon, say) have a very wide visitor demographic ranging from geeks to grandmas.

1 in 20 lookyloos is not the same as 1 in 20 customers. Also, as stated above, even if one accepts the 5% "standards-compliant" figure (I don't), the fact remains that 99% of those folks have IE around for when they need to access those IE-dependent sites.

The stats are only wrong in the sense that it is impossible to draw right conclusions from them. It may be in fact that all the uncertianties cancel the numbers are a perfectly accurate profile of actual web users. However, this seems very unlikely. Without knowing that systematic problems in the data collection have been accounted for, it is impossible to use the data with a reasonable expectation of reliability.

Yet, industry does draw conclusions from them, no? It's the reality vs. perception phenomenon I discussed earlier.

I don't believe that the marketshare is exponentially increasing is true because I haven't seen anyone producing convincing evidence that it is true. So yes, anyone who believes that gecko share is surging either has data that I haven't seen (in which case, they should present it) or is deluded. As for making these predictions come true - I think many people have ideas about what needs to be done, but people have limited resources for turning ideas into reality.

Yep, and many of us assist where we can with donations and eye candy (jelly beans, Life Savers and Babe Ruth bars) AND not wearing rose colored glasses.

[jgraham]

While those are useful, most companies would be foolish to use them as their primary metric. Why? Because they are mostly seeing stats from existing clients/customers. If they are interested in growing their market share, they must use/rely on an industry stat for making site expenditure decisions, imo.

To a certan extent yes. But there are real differences between the stats for different sites. In Germany, for example, I believe that Mozilla usage is very much higher than in many other parts of the world. Hopefully the people making these decisions understand concepts like 'demographics' and so will take these things into account rather than blindly trusting data.

The point I was making is that only the "standards compliant" share will have any effect in moving the industry. For example, my guess is that even today, Communicator occupies a very significant portion of the dubious 5%.

Maybe. Out of interest, have you tried browsing the web with communicator recently? Using Netscape 4 is painful, specifically because <em>many</em> sites only semi-support it. A much larger percentage of websites (from my non-random, limited sample) work in modern browsers than in NN4. So my perception is that the industry is moving away from the web as it was in 1998 and toward the web as it is in 2003.

1 in 20 lookyloos is not the same as 1 in 20 customers. Also, as stated above, even if one accepts the 5% "standards-compliant" figure (I don't), the fact remains that 99% of those folks have IE around for when they need to access those IE-dependent sites.

What's a lookyloo?
I accept that more people have access to IE than actually use it. At the same time, how many goods or services can be provided by a single site only? At least some people will find it easier to choose a different site than to change browsers (I have done this, for example, but then I don't have IE)*. So yes, the potentail number of visitors to a IE only site is greater (but how much?) than the IE market share.

You did make me realise, a more interesting statistic would be percentage of web sales by browser, both for individual sites and generally. That would be much more meaningful, since regardless of what factors contribute, it focuses on the single number (money made) that can justify expenditude. Any suggestion on where to find this data?

Yet, industry does draw conclusions from them, no? It's the reality vs. perception phenomenon I discussed earlier.

But they are still wrong to do so. That is my point. <em>Anyone who tries to draw wide reaching conclusions about browser market share from Google statistics is drawing incorrect conclusions</em>. This fact applies just as much to the people here as it does to designers trying to decide which browsers to support. In spite of this, I think we both agree that the higher Gecko browsers appear on any market share statistic, the better.

Yep, and many of us assist where we can with donations and eye candy (jelly beans, Life Savers and Babe Ruth bars) AND not wearing rose colored glasses.

I don't deny that you've done lots of good work, particularly producing nice themes. However, my point was that the Mozilla foundation has no marketing budget, a limited budget to employ programmers and limited ways of making money. That means most of the code and marketing effort has to come from volunteers in their spare time, which severley limits the possiblities.

Incidentally, in what way does trying to look at a set of data being discussed (the google statistics) with some rigour (not assuming they are correct just because they are on a pretty graph from a well known site) amount to wearing 'rose coloured glasses'. I don't believe I have ever made the connection that because the google stats are unreliable they must underestimate the Mozilla marketshare, although admittedly I did suggest one mechanism (tabbed browsing) through which such an underestimate might occur.

*I wonder how disability laws affect this - business premesis in the UK are required to provide full disabled access within the next year. If this applied to online retailers (it probably doesn't but I haven't read the law) then it would force business sites to cater to a range of browsers. But that's just idle speculation.

[Sailfish]

Maybe. Out of interest, have you tried browsing the web with communicator recently? Using Netscape 4 is painful, specifically because <em>many</em> sites only semi-support it. A much larger percentage of websites (from my non-random, limited sample) work in modern browsers than in NN4. So my perception is that the industry is moving away from the web as it was in 1998 and toward the web as it is in 2003.

I have it installed but very seldom use it; although, I still find a lot of folks one the newsgroups who steadyfastly refuse to move away from it for a variety of reasons.

What's a lookyloo?

In the States its a term that is often used in real estate. Specifically, when one puts their house up for sale, the real estate agents tend to flood the seller with "potential" buyers. The reality is that most of these buyers have no desire in actually "buying" the property but go for a slew of other reasons, e.g., comparing the seller's home to theirs or just plain nosey people who have nothing better to do with their time. <reminder>Hmm, need to find a way to recruit these folks for mozilla hacking</reminder> :_)

I accept that more people have access to IE than actually use it. At the same time, how many goods or services can be provided by a single site only? At least some people will find it easier to choose a different site than to change browsers (I have done this, for example, but then I don't have IE)*. So yes, the potentail number of visitors to a IE only site is greater (but how much?) than the IE market share.

Perhaps, but not that many, methinks. For example, I just keep this bookmarklet handy...

You did make me realise, a more interesting statistic would be percentage of web sales by browser, both for individual sites and generally. That would be much more meaningful, since regardless of what factors contribute, it focuses on the single number (money made) that can justify expenditude. Any suggestion on where to find this data?

Did you try googling?

Seriously, I'm not close to this issue to even offer a suggestion as to where to look.

I don't deny that you've done lots of good work, particularly producing nice themes. However, my point was that the Mozilla foundation has no marketing budget, a limited budget to employ programmers and limited ways of making money. That means most of the code and marketing effort has to come from volunteers in their spare time, which severley limits the possiblities.

I only provided those links so that you (others) would realized that I'm really a Mozilla bigot and not just your garden variety troll. Also, fwiw, I prefer the term "porting" to "producing" since 99% of the design rightfully belongs to the original authors of those themes.

Incidentally, in what way does trying to look at a set of data being discussed (the google statistics) with some rigour (not assuming they are correct just because they are on a pretty graph from a well known site) amount to wearing 'rose coloured glasses'. I don't believe I have ever made the connection that because the google stats are unreliable they must underestimate the Mozilla marketshare, although admittedly I did suggest one mechanism (tabbed browsing) through which such an underestimate might occur.

Casting a scrutinizing eye on the stats is fine, even commendable, as long as one doesn't scrutinize them so finely as to claim the see the image of a diety in them.

*I wonder how disability laws affect this - business premesis in the UK are required to provide full disabled access within the next year. If this applied to online retailers (it probably doesn't but I haven't read the law) then it would force business sites to cater to a range of browsers. But that's just idle speculation.

So far they haven't worked their way down to businesses in the US; although, there is some gubmit edict regarding their sites where they must become accessible. Sorry, I don't recall the initiative's name?

[d_ralphie]

It's certainly more of an IE clone than Mozilla. Hell the user agent is the same.

what, because it identifies as msie by default it is a CLONE? oh please. i am talking about the fact that both mozilla and opera support msie extensions, so it doesn't make sense to talk about opera being more of an msie clone.

That means it's very hard to distinguish stats, and so Opera usage does very little for awareness on the part of web developers.

no, it is extremely easy. opera always has 'opera' in the useragent string.

[d_ralphie]

I already explained before. Opera supports huge parts of IE's proprietary web extensions, including nearly all of the IE proprietary DOM,

wrong. mozilla tries to support as much as, if not more than opera.

by default uses filename extensions instead of MIME types to determine a file's type, sends a user agent string that mimics IE's,

nonsense again. msie does content sniffing. opera doesn't.

and even has cloned IE's colored scrollbars.

and mozilla supports marquee, and supported it AGES before opera.

Mozilla supports very little of IE's proprietary DOM, always uses the MIME type instead of extension, and sends a user agent string very different from IE's. And don't even think about colored scrollbars!

oh come on! you can't be serious. no one can be as unenlightened as you... what the hell does colored scrollbars have to do with anything? it's a minor thing and isn't even supported in standards/strict mode in opera, only quirks.

mozilla also supported msie's innerhtml ages before opera.

so don't give me that crap about emulating ie.

and the useragent spoof is there not to emulate ie but to trick stupid scripts that only know about netscape and msie.

The truth is, Opera does its best to cope with sites that were written and tested only for IE. Mozilla has carefully chosen a small subset of proprietary extensions and says "TO HELL!" to IE-only sites.

you are an ignorant fanboy. mozilla does its best to copy with msie specific sites too. the huge amount of proprietary dom supported by mozilla is hardly a 'small subset'.

FACT: you are trashing opera for EXACTLY what mozilla is doing.

please tell me you are trolling and/or joking! no one can be THIS stupid and ignorant!

come on mozilla people, can't someone at least set this moron straight and show that the mozilla community isn't full of complete morons?

[WildBreeder]

how exactly does opera try its best to be an ie clone?

you are making a fool of yourself, fanboy.

I already explained before. Opera supports huge parts of IE's proprietary web extensions, including nearly all of the IE proprietary DOM, by default uses filename extensions instead of MIME types to determine a file's type, sends a user agent string that mimics IE's, and even has cloned IE's colored scrollbars. Mozilla supports very little of IE's proprietary DOM, always uses the MIME type instead of extension, and sends a user agent string very different from IE's. And don't even think about colored scrollbars!

The truth is, Opera does its best to cope with sites that were written and tested only for IE. Mozilla has carefully chosen a small subset of proprietary extensions and says "TO HELL!" to IE-only sites.

So what you are saying is that Opera actually makes an effort to work with the web as it exists today, while Mozilla developers live in some dream world and think only W3C exists? Maybe this is why Mozilla/Firebird has problems with so many pages? I sure hope this isn't the case, but you seem to be convinced of this...

A geek friend recommended Firebird to me, but I had some problems, and I can see why. I didn't want to pay for Opera when you can get other browsers for free, but it seems that you get what you pay for...

Thanks schapel, you saved me the effort of learning this new browser which it seems would ultimately lead to frustration!

[shadytrees]

<li>What is counted. Is it homepage hits? Is it total searches? Does it filter by IP address?</li>

Probably people who hit Google sites (Google, Google News, Froogle, etc.) and probably by IP.

<li>Whether searches from the various search bars are counted</li>

Probably unless Google has added a way to specifically identify referers from the Google Toolbar, which it may or may not have done. But that's about it. There's no way that Google can tell if the search came from a Firebird search bar or an Opera search bar.

<li>How the count is affected by the IE-only google toolbar, which is known to communicate with the google servers for every page visited</li>

Again, it's by IP, so I don't think a communication would count as one more, although the first one would.

<li>What effect tabbed browsing has on the statistics. Tabs mean less navigating back and forth so maybe less google hits</li>

Again, it's by IP.

<li>How sophisticated the browser detection employed is</li>

IP is probably as sophisticated you can go. Google might also use cookies and other methods that I'm not familiar with.

<li>Google is used by a representative sample of the web using population*</li>

Google gets enough hits per day so that the data is fairly accurate and has a small margin of error. Remember, Google has connections with other sites too.

I'm sure there are many more reasons why this data may be less than reliable. So although the trend is probably correct, I wouldn't put too much empahsis on either the absolute value or on the gradient at any one time, even though google is clearly one of the most used sites on the web.

Agreed. Also, not everybody uses Google although most do.

[MoNkaholic]

It's great to question stats, but seriously... it's not like other numbers have shown this to be false. I can't remember an impartial statistic where the other browsers made up anything more than some 5-8% in total... in a long long time.

I trust Google's numbers mostly because it's everywhere. Everyone uses Google, it's not like Linux users don't use Google, or Mac users don't use Google, or Windows users don't use Google.

I have to assume that Google cares more about the numbers they get than any of us, which is something a lot of people here seem to miss. They have no reason to skew the numbers one way or another, it's in their best interest to find a way to count things like Google toolbar searches for instance. The better their numbers, the better they know their users and whom to target, and I have faith that at the very minimum the pursuit of profit is enough to get them to produce results that are as accurate as they can be.

[IGAU]

So what you are saying is that Opera actually makes an effort to work with the web as it exists today, while Mozilla developers live in some dream world and think only W3C exists? Maybe this is why Mozilla/Firebird has problems with so many pages? I sure hope this isn't the case, but you seem to be convinced of this...

I have encountered no problems on pages since I started with Firebird 0.6.1. Now I'm on 0.7, and still, no problems. The only sites that have "rejected" me were using stupid UserAgent scripts, and I got around that problem in about 5 minutes by installing and using an extension. Furthermore, the site is actually being redeveloped to support standards (and this is an Internet Banking site where they are very overly worried about security - and rightly so), and therefore will shortly cease to be an issue at all.

A geek friend recommended Firebird to me, but I had some problems, and I can see why. I didn't want to pay for Opera when you can get other browsers for free, but it seems that you get what you pay for...
Thanks schapel, you saved me the effort of learning this new browser which it seems would ultimately lead to frustration!

I fail to see what there is to learn really. Right click, open in new tab. There... That was difficult. I've encountered no frustration at all from Firebird 0.6.1, 0.7 and the Mozilla suite itself. If you had some problems, and presumably have the ability to post on forums, you could choose to help by telling Evangelists about it or posting a bugzilla report. Granted that most people would rather not bother, but if nobody is willing to do this then we're heading for a Microsoft based proprietry Internet and not a free standards compliant community where innovation by the masses is possible.

As for statistics, you can quote all you like, but ultimately none of the stats posted so far are worth anything. Someone suggested that it'd be good to see browser stats for online purchases; simple enough if large companies are willing to make their stats available. If someone like amazon or ebay could be persuaded, all they'd have to do is run a stats query for the "Thanks for your order" pages and bingo, you've got real stats that people actually care about. The only difficulty lies in getting people to do it.

[MoNkaholic]

As for statistics, you can quote all you like, but ultimately none of the stats posted so far are worth anything. Someone suggested that it'd be good to see browser stats for online purchases; simple enough if large companies are willing to make their stats available. If someone like amazon or ebay could be persuaded, all they'd have to do is run a stats query for the "Thanks for your order" pages and bingo, you've got real stats that people actually care about. The only difficulty lies in getting people to do it.

You expect people to make that kind of stuff available? I'm already surprised that Google does. Statistics like that are actually worth something. And I'm sure they exist already, just not publicly available.

[IGAU]

You expect people to make that kind of stuff available? I'm already surprised that Google does. Statistics like that are actually worth something. And I'm sure they exist already, just not publicly available.

No. Hence: "The only difficulty lies in getting people to do it."

[MoNkaholic]

No. Hence: "The only difficulty lies in getting people to do it."

My point was, people probably have already done it, it would just cost you money in order to get it.