A Little Perspective

I live in a world dominated by the unfeeling, unaccountable whims of browser vendors and so it is with a 50 lb bag of rock salt that I consume the optimistic projections of Firefox’s triumphal ascendence to market dominance.

As it turns out, my skepticism is absolutely justified. Now, this isn’t to say that I’m not an optimist – I am, and given what I do for a living, I need to be hopeful about the future. I would have succumbed to serious depression long ago if I weren’t deeply hopeful at some level. Clearly, though, big dreams are only rewarded with the depleted melancholy of a future forever deferred. We Ajax hackers color in the margins because we know all-too-well that Firefox hasn’t won. Yeah, it’s great for the 20% that have picked it up, but we can’t count on them. People, it seems, don’t easily change their habits regarding which button on the desktop to click to get to “the google”.

Lets put the last year’s market numbers for browsers in some perspective. From July ’07 to July ’08, IE 6 market share declined from ~45% to ~26%, or a drop of 19%. Given that we know that the Win2K market share numbers are now well below 5% on the public interwebs, this gives us serious reason to hope: if the remaining 25% could just be convinced to get with the century and run Windows Update (12 times or so), we’ll all be able to target the best in late-’90’s browser technology. In the same year-ish timeframe, IE 7 went from ~33% to ~47%, a gain of 14%. That leaves 5% of the churn on the table for non-MSFT browsers, most of which was picked up by Firefox. Since Win2K was still below 5% a year ago, who are these users who pick Firefox and not IE 7?

Part of the story about the uptake of IE 7 is the story of Vista. Vista was hovering near 5% in June of last year and a year later had taken a nearly 15% share. Now, it’s hard to say anything useful about potential correlations, but if there’s a 10% gain in IE 7-by-default installs of an OS, and if some large percentage of that number came at the expense of non-upgraded XP boxen (likely old computers), it wouldn’t be hard to spin a yarn about how Firefox is gaining market share primarily at the expense of IE 6. To get a concrete answer to this will likely mean paying $100 for access to “pro” version of the Net Applications numbers, but even without it we can say something very concrete about the power of OS bundling of browsers:

Competitor displacement of bundled browsers on the monopoly market-share OS has a demonstrated year-on-year market-share improvement rate of 4%/year.

That 4% a year is pretty consistent since 2004 as well. Ouch. By way of comparison remember that IE 7 is replacing IE 6 at a rate triple that. Obviously that’s not apples-to-apples since every browser’s “internal” version replacement rates are much higher than their competitor-displacement-rates, but it’s clear that most users aren’t making choices about browsers. Auto-upgrades are largely doing their thing and users are making choices about OSes and (mostly) living with whatever shows up on the desktop.

This isn’t to say that Firefox can’t do better and that it’s not having an effect. It appears that the adoption rate for Firefox is going to have improved this year versus last, perhaps significantly. Safari is also making inroads into IE’s market-share, both as a result of iTunes bundling and Mac market-share gains in the face of Vista. It’s also unlikely that IE 7 and IE 8 would be happening but for the competition that Firefox has brought. But the take-away here is as powerful as it is dispiriting: we may be able to abandon IE 6 in another year or two, but no matter who works to displace it, IE 7 is going to be with us a long, long time. Worse, the sustained rate of competitor-displacement in the browser market is now much, much lower than it was in the previous era of browser competition. In one sense, competition is working in that every browser vendor is creating new versions. But the bigger picture remains: Flash can get to “ubiquitous” across the entire web with new capabilities in roughly 18 months and the Open Web faces a best case replacement time-frame of 5 years.

Reducing that differential from 42 months to zero is now the defining challenge of the Open Web. HTML is back in the hunt. Time to see how fast we can teach it remember the new tricks we’re so eager to teach.


  1. Posted August 3, 2008 at 7:06 pm | Permalink


    As a total non-sequitor – thank you for putting some better visual indication of links into your blog! :p

  2. Posted August 4, 2008 at 5:44 am | Permalink

    ‘Flash can get to “ubiquitous” across the entire web with new capabilities in roughly 18 months and the Open Web faces a best case replacement time-frame of 5 years.’

    In the early days of the web time-to-ubiquity was the few months it took Netscape to replace Mosaic, because the web was 90% nerd back then. Now the web includes grandma and other such non-nerd categories I can accept that we’re _currently_ on an _average_ of 5 years.

    BUT that’s surely not the whole story: More and more, web updates are not driven by user request, with Firefox I’m fairly sure you are now on the latest version just by clicking on the default button all the time. Maybe this is naive, but I think that in a few years (maybe the 5 mentioned previously) the time-to-ubiquity for Flash and the Open Web will be the same: a month or so.

    Yeah optimism.

  3. Posted August 4, 2008 at 7:27 am | Permalink

    @Alex- I’m curious: what features do you think Flash is lacking (or going to get) in the next 18 months that will help it become “ubiqitous”?

    Is there anything that Flash can learn from the AJAX world, or vice versa, to help either or both really make quicker steps forward than that?

    Or do we have no choice but to just wait for old lame browsers to drop off?

  4. Posted August 4, 2008 at 8:35 am | Permalink


  5. Posted August 4, 2008 at 9:10 am | Permalink

    Great data. Even with ie6 going away just knowing that we are still going to have to deal with ie7 and ie8 the many issues is just sad. It wouldn’t be so bad if they were followed more of the standards.

  6. Posted August 4, 2008 at 10:59 am | Permalink


    So Flash is absolutely ubiquitous already. What I was pointing out is that it’s “internal” upgrade rates are better than 50% a year, so that whatever Flash introduces it will be everywhere (relatively speaking) in less than 2 years. Flash already has lots of features that HTML would kill for and Flash 10 is starting to try to target HTML’s sweet spot (text layout) as well.

    Can the Flash and HTML world’s lean from each other? Absolutely, although at this point I think we can make a pretty strong case that the Flashes and SIlverlights of the world are the ones learning fastest. From bytecode backgrounds, both have picked up markup-driven UI construction, some amount of CSS and text layout, and simple in-browser deployment. Then they add high-performance vector graphics, media and codec support, rational models for custom control development, and integrated tooling. HTML’s closest rejoinder are Safari and Firefox-specific CSS features. Oof.


  7. Joe Chung
    Posted August 4, 2008 at 12:48 pm | Permalink

    There was a five-year lag between the release of Windows XP and Windows Vista. Barring another Longhorn-esque calamity, I don’t think that we’ll be seeing the same lag between Vista and Windows 7. It’s likely that the default browser in 7 will be IE 8 and that you won’t be able to run IE 7 on Windows 7 so the churn from IE7->IE8 should be quicker than it was for IE6->IE7.

  8. Posted August 4, 2008 at 12:53 pm | Permalink

    Flash has always had better rendering consistency across platforms. However, HTML is much easier to build.

    I always wondered why someone (Adobe?) doesn’t build a HTML/CSS/JS to Flash bridge. Sort of like sIfr but for the whole page. I really thought AIR could be that bridge. Write once, and render anywhere. :)

    It’s been 14 years for me since I wrote my first html. When Netscape 1-2 was out, it was easy. No alternative engines. When IE6 was dominant, (95%+) it was easy for the same reason. Competition is good for innovation but terrible for us poor web developers.

  9. Posted August 4, 2008 at 1:11 pm | Permalink


    I’m not sure why we shouldn’t expect another Longhorn-esque calamity. WinXP was hugely delayed, Win2K was hugely delayed, and Vista apparently only shipped by sheer force of bill. I think what would be best for everyone here would be for MSFT to commit to a down-rev “re-qualified OEM” version of both XP and Vista that runs IE 8 by default. Better yet, they can (and should) un-bundle the browser.

    Neither a new OEM version nor unbundling would address installed base, but they could sure do something useful for folks not looking to fall on MSFT’s OS strategy sword as soon as Windows 7 is released. We’ve been burned by every promise of a Windows ship date this century. Why should we really expect Windows 7 to be any different?

    I think it’s time to demand a workable contingency plan.


  10. Posted August 5, 2008 at 3:17 pm | Permalink

    There aren’t “no choice”. Humans are humans, even Grandmas.

    Their was a time, in the early web, where almost any website was including in their web page something like “you’re browser is deprecated, you’re has been, switch to the XXXXXX browser, you’ll be happier, you won’t need viagra anymore and your wife will be beautiful” (ok it was more often “optimised for XXXX 12.0” or “best viewed with XXXXXX 42.5”).

    During this period, we were just looking to download this “brand new hyped browser which provides super cool new feature”. I remember a time when I even advised a lot of people to switch to IE4 because of its features, and how the web was way better with it.

    WE just have to make people feel their current browser is deprecated and does not work well anymore with the new open web. _This_ is the key.

    Don’t look at the market part : if you are a big company providing a good service, and if my grandmother browses your website and get something like “you’re using a bad old browser, if you want to get the latest technologies and get into this website, just move to one of these browsers, they are definitively cool” (don’t even think I would have suggested to move even on IE8, as a JS programmer, it’s just … IE6bis, and as a CSS developer it’s just FF 1.0 or maybe 1.5…), my grandmother will just phone me and say “my dear, please upgrade my browser, someone told me it’s a bad software”.

    With my father, it would be different : “a friend of mine has told me that with this browser you cannot go everywhere, it’s too old, please upgrade my browser, and my friend said that this oone is way better”.

    That’s just how humans work! Pressure them, use things like “Pushup the web” ( http://www.pushuptheweb.com/ ).

    Don’t waste time supporting IE : it will cost you about 30% of your webapp front budget, and you’ll just have what can support IE as it’s maximum for this price, not what every other can other do!

    This is really noticeable if you want something quite advanced, particularly using advanced Ajax, even using good frameworks(and I make one of them !). I think I know about 20 ways to just kill an IE getting on one of my webpage, and a hundred of bugs linked with rendering, memory leaks, css and html structure, loading problems.

    I often say I’ll start a blog to talk about all of these bugs, it’s just unbelievable… IE was a good platform 13 years ago, but it’s no more what we should support.

    We don’t have to fix IE on each website we make, Microsoft can fix its browser by itself, otherwise let’s get IE out of the landscape, it’s a real crap for developers:/

    Viva Open Web !

  11. Posted August 6, 2008 at 4:19 pm | Permalink

    “Flash can get to ‘ubiquitous’ across the entire web with new capabilities in roughly 18 months and the Open Web faces a best case replacement time-frame of 5 years. Reducing that differential from 42 months to zero is now the defining challenge of the Open Web.”

    Hi, interesting essay. I don’t have the answer, but I do have some datapoints which may be of use:

    o Consumer adoption rate of Adobe Flash Player is now actually quite a bit faster than that… Player 9 (with Just-In-Time compilation etc) was released end of June 2006 and was audited at 97% consumer support twelve months later. Its Update 3 (with H.264 video) was released in December 2006 and reached 82% consumer support within six months.
    Neither of these used the auto-update mechanism, which is reserved for security releases. It was the range of content out there which drove the adoption rates.
    To have over 80% overall consumer support, across browser brands, within six months is indeed a big deal.

    o Download size has been an issue in the past. I’m not sure what effect it will have in the future, particularly across device types. In the early days Macromedia did studies adding null kilobytes to Player downloads and measuring the dropoff rate in completed installations. The more time people have to hit that “Cancel Download” button, the more will do so. And browsers are bigger than plugins….

    o I haven’t seen measurements of it, but one factor I’ve long suspected for slower browser adoption is that it risks change to each person’s daily visual/tactile routine. Browsers have chrome, menu items, new preferences to investigate. Adobe Flash Player is just invisible. There are no habits to change, no risk of breaking daily concentration. Visible newness can slow adoption.
    Rephrased, changing your browser chrome involves a conscious choice, the assumption of risk of habit change. But invisibly improving a background capability is an easier choice to make.

    o Needless to say, changing browser brands imposes a much higher cost on endusers than just updating within the same brand. If all browser creators don’t agree on providing a capability, then it will always be costly for sitemakers to employ that capability. This is intrinsic to the spec-first, implement-later model, and is one reason why cross-browser plugins are such a valuable complement.

    o Firefox does have a healthy update rate. But it accomplishes this by daily phoning home to check for new updates. Adobe Flash Player defaults to checking for updates every 30 days, and such updates are published at a less frequent rate than for Firefox. I’m not sure of the optimum notification setting or criteria for the majority of the web, but suspect it’s more at the Player end of the scale than the Firefox scale. Open question.

    o What’s the fastest way to use new abilities? I suspect it’s opening up “The Open Web” and not casting aside technologies just because of branding. Folks at Sitepen already have an open-minded attitude towards web technologies. But there’s also some needless antagonism out there too.
    By embracing cross-browser technologies, instead of retreating into a browser-brand niche, wider audiences will be able to use newer abilities more quickly.

    My apologies for any poor phrasing on my part… typed quickly. These are some of the ideas I thought of while reading Alex’s essay and subsequent comments. Thanks for the read!


  12. pawel
    Posted August 8, 2008 at 12:16 pm | Permalink

    Here are stats from Poland. The market is quite different than global average or the USA, currently 41% of users use Firefox.

  13. David Reese
    Posted August 8, 2008 at 11:00 pm | Permalink

    So interesting that John Dowdell says “Adobe Flash Player is just invisible. There are no habits to change, no risk of breaking daily concentration. Visible newness can slow adoption.” That’s true, but Flash is only transparent in this way from the point of view of the *interface to the interface*. The actual interface that we interact with most of the time is web pages and applications.

    Flash is kind of insidious that way, because as transparent as it is from a meta-perspective (no browser-chrome, etc), the actual pages/interfaces built with Flash almost invariably make the user, as John says, change their habits and break their concentration: *How do i use this widget? Where do i click? Why is that animating, I’m trying to READ.*

    Prominent exceptions include the use of flash to substitute for functionality that should darn well be in the browser anyway, like video and audio embedding, file upload progress indicators, and, um, vector graphics. I don’t think it’s a stretch to blame flash for the lack of openweb progress on these fronts.

    Of course, this is tangential to Alex’s point, which is, of course, CALL YOUR GRANDMA AND TELL HER ABOUT YOUR FAVORITE BROWSER.

  14. Posted August 19, 2008 at 2:01 pm | Permalink

    @Glen ….

    “Flash has always had better rendering consistency across platforms. However, HTML is much easier to build.”

    “I always wondered why someone (Adobe?) doesn’t build a HTML/CSS/JS to Flash bridge. Sort of like sIfr but for the whole page. I really thought AIR could be that bridge. Write once, and render anywhere…”

    FYi; The Adobe Air runtime implements the WebKit layout engine. While the WebKit document model is backwards compatible with much of the legacy Web, they are pushing the envelope when it comes to HTML5-CSS3-JavaScript-DOM, but in a very structured way.

    It will be interesting to see how this works out. The Web 3.0 and the emergence of client/WebStack-Cloud/server RiA applications changes everything. Especially the legacy Web document model.

    Does the Web document model gradually morph into graphical RiA interfaces?

    Or, will RiA efforts simply wrap Web documents into content-data-streaming media pockets, and never look back?

    IMHO, there are three RiA contenders to consider: Adobe AiR-Flex-Flash, Webkit-SproutCore, and XAML-Silverlight. Maybe Google can make a play with GWT, but i think even there much depends on how far they will push WebKit.

    A new generation of highly graphical applications and services are on their way. An associate of mine phrased the change as similar to that when DOS was replaced by the Windows GUI. These RiA apps have stunning potential to deliver on demand volumes of infogrid spawned information in easy to understand graphical interfaces. The Web document is not going to go away, but it is going to be delivered in a mashed context where the visually clarifying capabilities of RiA are certain to surprise.

    What is certain to change and perhaps go away is the legacy model of desktop client/server applications. Microsoft knows this, and they have successfully stalled open interop with the MSOffice desktop until such time as their client/ WebStack-Cloud /server pieces are in place. Now they are ready, with the XAML-Silverlight cutout of WPF-.NET developers framework ready to rip. The great transition of MSOffice bound client/server apps to th MS WebStack-Cloud computing model is underway.

    One of the more important pieces demonstrating how Microsoft was to control and direct the great transition was first revealed in December 2007 MSOffice SDK beta. It’s an easy to implement conversion between MSOffice OOXML documents and something called the XAML “fixed/flow” document model.

    The idea behind this converter seems to be that Microsoft has found a way of transitioning MSOffice business documents to a web ready format that they control. Yes, OOXML is an open XML format (ISO 29500). But the conversion to Web readiness and the new XAML-Silverlight RiA developers model is totally proprietary.

    The beauty of the OOXML-XAML converter is that tf anyone complains about the proprietary aspects of WPF, (most of which are clear replacements-alternatives to Open Web formats, protocols and interfaces), then they are free to try to convert OOXML to a web redy format. OOXML is after all “OPEN”.

    There is no doubt in my mind that WebKit HTML5-CSS3-JavaScript is capable of handling the complex and compound MSOffice business process documents. Since so much of today’s business processes are bound to MSOffice workgroups, workflows and client/server systems, this conversion is essential to an easy transition.

    Many fault Adobe and Apple for pushing the envelope far beyond the W3C, ISO and Ecma consortia agreements. I would argue that they have to push beyond the mess that passes for standards if the Open Web is to be saved. For sure Microsoft will break the Web with XAML-Silverlight and the transition of the desktop client/server monopoly to a MS WebStack-Cloud computing model; even as Redmond cooperates at the lower legacy web standards – browser level. Kind of like implementing ODF 1.1 in MSOffice when the entire ODF community has moved to an ODF 1.2 version that terminally breaks backwards compatibility with ISO 26300. The interop most expect from standards becomes a standards sanctioned impossibility.

    One has to wonder if Microsoft’s participation in the W3C-WHATWG, Ecma, and ISO is designed to stall HTML5, CSS3, JavaScript and SVG? For sure they have no interest in PDF, XHTML, RDF and SPARQL. (WPF includes XPS, XAML, Smart Tags and LINQ proprietary alternatives).

    Apple, Adobe and Google are definitively outside this lock Microsoft has on MSOffice bound business processes. Are we looking at a Web future where the consumer side of the Web is Open? And the business side proprietary? All the more reason to insist that Apple, Adobe and Google get their act together. If they individually take on Microsoft, we can kiss the Open Web good-bye. The reason is that this isn’t so much about open standards and open web technologies as it’s about a monopolist having figured out how to successfully transition much of the business world to a proprietary version of the Web, and do so within the constraints of existing anti-trust concerns.

    In short, Microsoft has figured out how to manipulate the standards process and still maintain control of fundamental formats, protocols and interfaces.

    During the Combs vs. Microsoft anti-trust trial an important eMail from Chairman Bill to the MSOffice group was discovered. The eMail was titled, “Office Rendering”, and it contained this concern/demand:

    “One thing we have got to change is our strategy — allowing Office documents
    to be rendered very well by OTHER PEOPLES BROWSERS is one of the most
    destructive things we could do to the company.

    We have to stop putting any effort into this and make sure that Office
    documents very well depends on PROPRIETARY IE capabilities.

    Anything else is suicide for our platform. This is a case where Office has
    to to destroy Windows….”

    Looks to me like they’ve succeeded.


  15. Posted August 19, 2008 at 4:22 pm | Permalink


    Even if I agreed with everything you said, I’d have a hard time figuring that out. Instead, I decided to play a game:

    • … Web 3.0 …
    • … changes everything …
    • … web ready …
    • … WebKit-SproutCore …
    • … the great transition …

    (Bullshit) Bingo!


One Trackback

  1. […] is a key comment in Alex’s perspective where he discusses his analysis on the browser market, and how it is changing… or isn’t […]