2006-12-26

Characteristics of the winning WebOS

I'm going to predict the future for you here. Hopefully, like any good prognosticator, I'll leave my definitions vague enough and my time-line open enough that I'm almost guaranteed to be correct. What I'm going to tell you is the look of the web operating system, or WOS, that will ultimately "win".

The primary purpose of a WOS is to convert web services into a commodity. This is really just argument by assertion, but here goes anyway.

DOS is to disk as WOS is to ___?
The point of a computer operating system is to abstract all of the "interesting" parts of a computer in a way that allows application programmers to make use of them. Think back. Disk Operating System, or DOS, existed to read applications off of disks and to allow those applications to read and write to disks. Without a disk operating system, application providers would need to partner up with both computer and disk drive providers to deliver an application suite. Some amount of application code would need to exist in the ROM of the computer just to read the rest of the application suite off the disk.

This is largely the state of the web today. A complete web application utilizes a wide variety of web services for distribution, authentication, session management, search, etc. To deliver a web application today, you need to deliver an entire web server, or fleet of web servers. You can partner with other businesses for site hosting, back-end databases, and storage. You can utilize relatively common components, such as with LAMP. Nevertheless, as it stands, it is impossible to define a complete and relatively complex web application that can be freely moved from one operating environment to another.

Many of the critical services needed by a web application can be hosted by the local desktop. I argue a WOS must include a reference server, though I don't know if that is the only solution. What I do know is that one desktop server doesn't scale well to applications that could have users across the entire web. It is therefore critical the application just as easily makes use of commodity web services. That is the benefit of using a WOS. It provides the abstraction to make an web application that can scale without modification.

The interactions we can have with the web are much more interesting than the interactions we can have with disks, but I think the analogy holds. There are even some important interactions with disks that current desktop operating systems don't support, but that is a another story. I even go so far as to say that using a WOS should allow you move your applications to an entirely different web, off of the Internet and onto a private LAN. It is only some subset of the web that can be truly abstracted in this way and many web applications will have dependencies that don't allow them to work independent of the Internet.

That subset of services that can be made independent of the Internet is what is important to a WOS.

Growing the market for web services
I don't mean to say that a WOS developer should focus only on the server-side. A WOS must provide an environment where those web services can be consumed by a sufficiently large market. In this context, it can be critically important to provide a rich set of client services and template applications to simplify application development. Some people seem to get this secondary objective of a WOS confused, such as the blogger quoted below.

But what is a WebOS (not to be confused with another definition of the term, see here), or a Webtop, anyway ? Here’s a simple definition: WebOS is a virtual operating system that runs in your web browser. More precisely, it’s a set of applications running in a web browser that together mimic, replace or largely supplement a desktop OS environment.
The "webtop" functionality is important in the simplicity it provides in creating web application, but I think Wikipedia's definition is closer to correct. I struggle, however, to see where a webtop alone will significantly help generate the next killer web application. Web browsers already support tabbing and there are plenty of AJAX frameworks for producing rich interfaces. While a webtop and set of example applications seems necessary to be recognized as a WebOS, do any of these webtops offer sufficient benefit to application developers to lock themselves into one of them?

How a WOS will generate cash
Consumers want everything to be free. That is, they want it to cost them so little effort or resource that they don't even know they are paying. This is where a WOS has some real potential. A WOS can turn all of those profitless web application companies into differentiated web service providers. How?--by putting a line between what is scarce and what is abundant.

Software is abundant. Given enough time, someone is going to write this WOS of the future and make it open source, so there is little hope of making money off of selling one in the long term. Though there is little money-making potential for someone making just the software for a WOS, there is significant value for the service, content, and hardware providers. The existence of a WOS can increase the number of their consumers and provide the infrastructure they need for revenue, such as micropayments. The most successful WOS will turn the scarce service and content resources into commodities, maximizing their availability.

At first, nervous service and content providers may be reluctant to feed such an ecosystem where they are on level ground with their competitors. They will wonder if they will be able to sustain their value. Eventually, they will learn that there is still value in their brand, their services, and their content. Why? Brands are scarce, because attention is finite and it takes attention to build confidence. Services require resources and expertise, and are therefore scarce. Despite the seeming abundance of user-generated content and the high-availability of copies of any particular content, there is a finite amount of content available to sustain the value of any given brand. Though new markets can be generated on top of an existing consumer base, the ability to generate content that is suitable to an existing target market is limited to a finite set of content providers. In other words, people will get bored watching home videos of cats climbing the curtains.

With the scarcity of services and content, the abundant WOS will simply be the catalyst for revenue. The hardware is a story for another day.

The predictions
To conclude, here are the top 10 characteristics I see for the winning WOS:
  1. It will be a standard, not a single implementation.
  2. It will be implemented with open source software at least once.
  3. It will be packaged with a web browser, like Flock.
  4. It will make transparent use of free (community-based), fee-based, subscription-based, or advertising-based web services when local services aren't sufficient.
  5. It will make use of open identity services, like OpenID.
  6. It will provide namespace and tunneling services, like Paper Airplane.
  7. It will provide media playback, publishing, and subscription services, like Democracy and Broadcast Machine.
  8. It will work on any network, including a network of a single computer.
  9. It will be accessible, isolating the application presentation such that the interface can be implemented independent from the core application.
  10. It will provide bandwidth aggregation, like BitTorrent.
(Update Jan 6: link to my other posts on web operating systems)

3 comments:

  1. It's not a webos unless it makes it easy for various apps to tie into eachother.
    Each application in a webos should by default be able to provide an open api for other apps, and also share basic characteristics with them.

    ReplyDelete
  2. That's just an excellent goal for applications in general, not something specifically new to using a WebOS. In the case of the DOS analogy, DOS apps are only guaranteed to be able to read the files off of the same disk, not necessarily interpret files generated by other apps. Hopefully we've gotten a bit smarter, leveraging a more common set of data formats (through microformatted XHTML) and other characteristics. I'd just happen to rank this as a general application goal.

    ReplyDelete
  3. Interesting thoughts.

    The whole new movement to bring the web and the desktop closer would bring similar platforms.

    ReplyDelete