Web browser versions are dead – automatic updates is the future
A somewhat provocative article title, I know, but I think it’s really important to discuss where we are right know in regards to web browser versions.
The reactions I have gotten to HTML5 hurdles: what is missing and web browser update rate problems and from constantly meeting/getting in touch a lot of web developers in very varying contexts from all over the world, by far the biggest problem is the rate of end users updating their web browser.
What do you mean no web browser version?
Internally, or behind the scenes, if you will, naturally you’ll have a web browser version number. It will be needed for development and support cases. But with the advent of IE10, Firefox is the only web browser who haven’t reached a two-digit version, and also with the frantic version update rate of Google Chrome, version numbers are really losing the point.
Right now it seems to be only for branding and getting attention, but I think it would be much better for everyone if that focus was spent on why people should use a certain web browser (features, performance, integrity etc).
Does the end user really need to know?
I’d say no. End users really really don’t care about the version number. They just want things to work. We have so passed the point (or I wish we have) of demanding a certain web browser, resolution etc to be able to see a web site.
The web is the port to the world
Given how much time users spend on the web when using a computer – communicating with others, finding information, leisure reading, gaming, watching weird videos – it is vital to make this as seamless as possible for them, no matter which web browser they use.
Automatic updates is the future
I’ve been wary of jumping upon the “we must have automatic updates”-bandwagon, and I think mostly that it has been for two reasons:
- I’ve felt that end users should get the option/notification of updates being available, respecting their choice to update if they want to.
- Given the backwards compatibility problems with Internet Explorer and a lot of major corporations who have built web sites specifically targeted at that, automatic updates would break all of that.
I still think those two are valid points and something to consider, but I’ve started to sway towards automatic updates being the best choice. When it comes to the first point, I think many many end users just decline updates when they see the update notification – they don’t dare to install things, they don’t feel computer savvy enough and, most importantly, they don’t want to break things that work.
Regarding the second point, the problem is still definitely there, but with IE9, upcoming IE10 and so on, I think Microsoft has reached a point where it will actually be adding on more and more support without breaking/redefining old functionality in it.
With other web browsers, I don’t think this has been an issue in a long time. Sure, of course there might be something somewhere, no matter the web browser, that might change an existing implementation in a previous web browser version. But overall, I think there’s is a lot of respect towards backwards compatibility and making things work.
It’s also about cost. With having already 4 different concurrent versions of Internet Explorer in the market (IE6-9), and IE10 probably out there within a year, and a vast discrepancy between them, we will reach a point where you can test in every version – there’s just not always time and money for that. Having automatic updates would remove a lot of those hassles, costs and time loss.
So, my suggestion is that all web browsers become version-less and all have silent automatic updates. For corporations, there should be tools to be able to control this centrally and decide if automatic updates should be enabled or when it should be pushed out to the organization.
I like the idea of automatic updates except for 2 issues that I can’t resolve in my conscience:
1. If the new browser version breaks something it will do so for every client. There will always be a factor of human error involved there including edge cases that (probably inadvertantly) did something that worked but stopped working…. to keep backward compatability through this process forever would bloat the browser so that’s probably a hiccup to resolve.
2. More importantly… many people (here at least) are still on rural landline modem connections OR are paying through their nose like my step-son for exploitative data rates on broadband (especially those USB sticks). So pushing data at people can be somewhat presumptive.
But, as I said, it looks like the best way to go in the long term when we can resolve those two hiccups. Our biggest problem with versions is probably not on the desktop at all but in the fractured mobile landscape.
With the increasing use of the browser as an application platform, it becomes ever more important to ensure the stability of the platform. Automatic updates work against this, no matter how careful the browser vendors are. There are environments where it is inexcusable for stuff to suddenly stop working – updates need to be qualified and tested before they’re rolled out.
The risk is if older browser versions are not maintained. This is less of an issue with IE but it can certainly be an issue with browsers such as Chrome. Once 12 is out how long is 8 maintained for? If you had standardised on 8 and a security hole appears in it your choices are limited.
The Linux vendors faced this style of problem when Mozilla stopped supporting 3.0.x and users had to be shifted to later versions of the browsers ( http://lwn.net/Articles/432360/ )
Ha, something similar was one of my first blog posts: http://leaverou.me/2009/02/silent-automatic-updates-are-the-way-to-go/
Except that standardization needs rather stable versions to test implementations against specificiations’ test suites. So a way to “stabilize” the browser we run is mandatory for the standardization process. Live specs and live browsers are a huge problem for standardization bodies…
I completely agree, as long as:
* You’d have a way of opting out. At the end of the day, automatic updates are a little bit creepy. I want a way to opt out from automatic updates of some vendors (*cough*Apple*cough*) that push their bloatware along.
* We’d keep a version number in the UA string so server-side heuristics can be used to complement feature detection and speed it up (as discussed by Alex Russel and by myself)
* In an ideal world, web devs would have an option to install several versions of the same browser side-by-side. But now I’m just pushing it 🙂
With the way that the two largest browser vendors (you know who you are) let their UX people redefine the UI with every new generation, the future looks very bleak for most users of browsers: non-tech-savvy people, just trying to get on the web.
But, I am sure you’ve considered this, and have a solution to this?
If we start getting automagic updates, then I might no longer be able to access my bank on the web, because the bank was unable to anticipate bugs rolled out globally in a new version of my favourite browser … but that’s ok, it’s not like I need to be able to pay my bills, right?
And, when I complain to the bank, I can only tell them that I am using BrowserX, and that the problems have started after monday …
While I agree, that having several years between major releases like IE and FF teams do it, is problematic, going to the other extreme is going to be even more problematic.
Surely, there must be some middle ground … Google and Apple seem to be able to push out quality updates quite frequently
I agree that there will always be risk of breaking things, which is where testing and focus on not making major changes (unless absolutely necessary) to existing APIs is very important.
Slower connections is a very good point. I guess it would be nice to be able to detect the connection speed versus the size of the update, and at a certain ratio, check with the end user if it’s ok.
And I do agree that mobile will most likely be bigger challenge.
Absolutely, but I believe that’s where more testing and quality assurance will come into the picture.
Right, but why would you ever standardize against a certain web browser version? Better to focus on general web standards and using feature detection for checking support.
The Firefox 3.0 thing is also the risk with separate distribution channels and embedding things, as opposed to just one central source.
Ha! 🙂 Nice!
It’s a good point, but for that specific reason, I think you could have certain web browser versions available. At the same time, if there’s an incomplete/buggy API in a released web browser, isn’t it better for anyone to push out an update as soon as possible?
The possibility to opt out has to be available, version numbers has to be available somewhere and you should be able to install multiple versions of a web browser in the same operating system.
I’d say the UX is a separate thing, but I’d say it’s a delicate balance between trying to be consistent and improving something that most end users will gain from.
With updates and possible bugs (hopefully ironed out by proper testing), where’s the difference if the user gets a notification and then upgrades? The user will never know anyway if there’s a potential bug in there or if the bank’s web site has been properly coded or not.
And like you say, so far, Google Chrome has been quite successful with this. Sure, their dev channel releases can be shaky, but the final pushed versions have been in a very good condition.
I already sort of responded to a tweet of you about this. Teasing you about Mozilla ( and Microsoft ) being a bit panicky at the moment when it comes to the version number their browsers are at. I think they’re all just trying to please the end user with their ‘progress’ but as you’ve perfectly pointed out it should be about performance, (standards compliant) features etc.. Great article!
There’s a flip side to the “platform stability” coin: A fractured platform is never stable. Today, it’s simply not feasible for many developers to test rigorously on every single combination of OS and browser out there, even if just counting the major combinations.
Contrast this with having a single version of the five major browsers to test for.
Also, any breaks will be quickly found, since all users will get the update at about the same time. If browser vendors let developers test new versions a set time before going to auto-update, and users can opt-out, with the dire warning that they should not expect any website outside their own intranet to work with legacy browsers, I really can’t see a problem here.
Absolutely, I think it’s time to shift focus on other things. 🙂
Great input! I definitely agree!
I’m a windows guy, so I might be wrong, but isn’t Safari also still in the single digits?
You are absolutely right, my bad. Interesting that no one else has pointed it out yet. 🙂
As far I know the next IE10 will run in Windows 7 Only, not even in Vista. So, I guess there might be issues with user’s PC’s configuration and OS version. One update might not work for all specially for someone who is using a browser 2 or 3 version earlier than the latest.
Who has an interest in version numbers? Developers!
Who doesn’t care at all? The user!
Who won’t allow automatic updates? server admins in bigger companies and in the civil service
We are lucky if a normal users doesn’t use “Windows” or “Google” for surfing but really knows that there is a browser with a name. And usually he won’t care which number it has. Browsers are not cars or hifi-racks.
For you and me version numbers will be of interest.
And all the server admins won’t agree automatic updates being good because they want and must control everything.
I don’t think that our problem with outdated versions derives from personal usage. It derives from corporate usage where Win XP is still state of the art and the admin feels like Hulk when updating to IE7.
And I fear that this will not change in the coming decade.
You are so right! Nobody, really nobody needs browser versions, and really nobody should be forced to download a browser anymore. Many people have almost instant access the Internet 244/7 nowadays, so why should new browsers not only be downloaded immediately when they are released? That’s the future and furtunately more and more browser vendors have already recognized that.
I think automatic updates would be fine for the general browser, if we could syphon off early IE specific web apps into some sort of app container tool.
If Microsoft released something like Mozilla Prism (a browser engine that launches only one website without the usual browser buttons or ability to browse the web) then all the old IE6 and IE7 only apps could be accessed in this manner and we could move forward more quickly with the public facing website technology.
The other alternative would be to have VMWare or similar setup just to allow certain apps to run, whilst having a more modern browser for other needs.
It’s a good point, and I can see where that would present issues. I think for automatic updates to work, the updates needs to at least be compatible with the previous (as in, not latest) operating system version too.
I acknowledge that problem, but with some at least, I hope they will welcome automatic updates to keep their computers safe without manual processes.
Glad that you agree!
Interesting ideas! I would love for Microsoft to release a stand-alone version for that kind of purposes.
I would say automatic updates by deafult, with an option to disable them (for users with slower internet connection, corporate users and everyone who doesn’t want to or can’t update for some reason).
Sounds like a good approach!
Chrome already does that, and I think it’s awesome. Completely seamless, you don’t even realize that the browser has updated unless you go to the About menu and see a “checking new version” message.
I hope other vendors adopt that strategy. I use Firefox most of the time, and already find annoying the update notifications. But even worse is Microsoft taking years between new IE releases, and people still not updating it.
That would probably be the end of hacks for old browsers (I’m looking at you, IE6), at least this problem would be greatly reduced.
It does work well in Google Chrome, and it will be interesting to see how other web browser vendors will approach this.
It will be hard with IE, though, and the operating system fragmentation they have going for them now.
Possible solutions could include removing UX layers from rendering engines, so when new updated engines and underlying technologies are updated, they could be more automatic (or silent), but the users are presented the option to upgrade to new interfaces as they become available. This may reduce the users being put off all important upgrades simply because they don’t like change…
Also why not have browsers marketed as Generation numbers, like Firefox Gen 4, so minor updates and security / performance features can be improved silently without user confirmations all the time, and compatibilities being maintained in each generation, possibly only updated annually?
Just some ideas that would hopefully note become too much of a fragmentation issue. I must say I do appreciate the what Chrome handles it’s updates 🙂
Interesting idea about the UX layer. and could be a possible approach. Or at least notifying the user that the UI has changed.
Generation numbers could be ok, but I wonder if even they are really necessary.
I have to agree here… this is the way that Chrome has been from the start… and more companies need to be more open to that way of thinking here. The security flaws in leaving un-patched cruft of everyone’s desktop when that’s the #1 attack vector needs to die. Firefox is pretty close to this as well, though they make their updates optional, and rarely offer the next full version as an automatic update.
Glad you think the same thing!
Firefox does eventually push major updates, AFAIK, but not directly.
I’m fine with automatic updates, so long it doesn’t break stuff. Thus, I’m fine with Chrome auto-updating, as I know by experience that it will keep working. I’m less excited when ze Fox wants to update to a new version. Will my theme, or some extension break?
And then there’s IE. I’ve honestly not paid attention to what’s been happening in IE land lately, but if you look back 10-15 years, IE upgrades used to be a torment in a certain dirty cavity. You knew for a fact that half of the OS shell would be affected and there was no way back. The OS integration is decreased since then, but the I suppose the reputation lives on.
With Firefox, I’m thinking that the APIs offered there must be backwards compatible (just like it works with Google Chrome).
With IE the history haven’t been that nice, no. I hope that will be better.
I hope to live to the day when Microsoft will embrace this principle.
As for now – I’m advising everyone to use Goole Chrome, Firefox or Opera.
I really hope Microsoft will, but it doesn’t look that promising.