Browsers the Killer App?

This post is in response to a blog post on ChrisKeane.net.

While I agree with every aspect of your argument, your premise is wrong. You’re limiting your mindset to your style of work and/or computer utilization. Web browsers do a great job of displaying information and transmitting it over the network. If your workflow involves displaying, creating or transmitting data then a web browser is all you need. However, the browser falls on it’s face when you examine alternative tasks.

Specifically the browser is not designed to perform processor intensive functions. For instance, calculating a massive chart on a spreadsheet is something the browser can emulate, but not do well. Sure, you can have a ’spreadsheet’ application via clever AJAX in a web browser, but a spreadsheet program is optimized for the task. Meaning, the code for the spreadsheet program is written to very specifically utilize the resources of the computer, whereas a browser plugin or the browser itself will never be able to leverage the machine’s resources as directly or efficiently - if only for the reason that they will always have the browser overhead on top of it.

Many people nowadays find a computer without a network a useless hunk of plastic. But there are also people who continue to utilize computers away from the resources of the internet. For these people and their workflow, there is no good reason to use the browser as a ‘launcher’ for their applications - whether those are calculative programs or even simple word processing. Utilizing a program directly coded for the task is always going to be faster and more efficient, if perhaps not as dynamic. Additionally, even desktop apps are becoming network aware so that they can leverage many of the advantages of a browser (auto updates, etc.).

As far as code compatibility goes I think Java pretty much answered that battle cry years ago. Java apps will, by and large, deploy in a platform independent way.

Of course, all of this argument hinges on how you use your computer. For most folks you could offer them usable applications via the browser. But for a sizeable minority you’re never going to be able to provide them a browser based application that will be suitable to their needs - witness the abysmal failure of the ‘thin client’ as evidence that not everyone wants remotely hosted applications.

The other factor you have to throw in is network reliability. If all your apps are delivered over the network via a web browser and the telephone pole outside your house falls over your computer truly does become useless. I know it’s hard to imagine to most people in urban areas, but broadband in the US actually has extremely limited penetration outside of cities. Additionally the network is fairly robust, but it is by no means bullet proof. Having your modem go on the fritz and killing every computer in your house is a scary thought.

Ultimately I think the browser is a killer app, but it isn’t the universal app you’re making it out to be. There are severe limitations to utilizing the browser as a type of virtual machine, especially if you’re launching code over the network. These, ultimately, provide compelling enough reasons that I don’t think people will migrate to an all browser desktop any time soon, and even when it does happen it won’t be everyone.