August 2008

So the bloggers over at ZDNet have once again proclaimed the end of the operating system. Larry Dignan says:

The operating system may be losing its luster. In fact, you could argue that the operating system–Linux, OS X and Windows–will become an application that just happens to boot first. And hardware vendors are on to the OS’s diminishing importance.

He goes on to say:

My working theory: The OS is being slowly downplayed as hardware vendors and Web developers grab more control over the user experience. The OS will never be totally irrelevant, but it will be increasingly less important. It’ll be plumbing. Simply put, the OS is being squeezed between hardware vendors that are cooking up their own applications to handle key tasks and the so-called Webtop, which will deliver programs through the browser.

I actually agree with much of what Larry says, even though I think the title and some of the points are too broad. Is the OS going away if people primarily use applications via a browser? Absolutely not. The OS remains: it’s only people’s legacy understanding of what an OS is that goes away. For instance, to my generation of computer users our experience with a computer was an experience with the OS. It was Windows, it was DOS, it was Apple’s, it was Linux. To my niece’s generation (age 14), their experience (except with gaming) is defined increasingly by the browser. Or by their cell phone.

As the traditional experience of the OS becomes less important, the value of a bloated OS with an incumbent advantage becomes less important. I have used both Linux and Windows and honestly when it comes to getting my work done, I find very little differences. Why? Because I use hosted applications via a browser. I use Word Press, Flickr, Google Apps, Gmail, online money management, online banking and so on. I don’t use native applications. The performance and experience of Linux in that case, is quite superior since it loads faster, performs better and gives me more flexibility. (It also doesn’t come pre-loaded with tons of crap-ware from AOL, security vendors and the like.)

Just because the OS becomes less visible, does that mean the OS goes away? How can it when software still needs to control hardware. You still need a kernel, you still need a scheduler. You can’t virtualize thin air. Plumbing is vitally important. (Just visit a third world country without it. Sanitation is the backbone of civilization.)

Is there a shift in the important of features required in an OS? Yes — see my points above. If the OS role has changed, I’m not so willing to pay the monopoly premium for Windows if native apps aren’t quite as important. (See the rise of the eeePC.) The ability for an OS to be flexible, to be customized by hardware vendors, by niche vendors who want to customize an OS for a specific audience: all of this becomes much more important. And all of this points to Linux. Desktops themselves are changing. They are becoming more like smart phones (or simply becoming phones). Those vendors want a customizable, modular OS they can brand with their own brand and not have to pay for the privilege to do so. This is why you are seeing the increase in Linux across mini-pcs, phones and embedded computing.

It’s funny to hear about the end of the OS when in fact, an OS or a component of it, like the Linux kernel, is in more and more places: it’s in your LCD TV, it’s in your car, it’s in your Tivo, it’s in your wristwatch. For something that is about to die, it sure is thriving.

There is a shift going on. There is a migration of margin in software sales, there is a migration of user attention from the OS to what you can do with the OS. But don’t forget there has to be an OS running all those apps people are accessing through a browser. See my list above — Google, Amazon, Ebay, Flickr, Facebook — all running Linux. In the history of evolution, those species most adaptable survive and thrive.

Popularity: 13% [?]

At the last Linux Foundation Collaboration Summit, the late Joe Barr wrote up this exchange on day one of the conference:

The summit’s first panel yesterday, a State of Linux Roundtable, was made up entirely of Linux kernel hackers. During the Q&A session that followed, a gentleman from Nortel introduced himself and told the panel that Nortel was running Linux on one of its switches, and it worked just fine, but the company had to make a number of patches to the kernel to get it to work. He wondered how Nortel could get its patches into the mainstream kernel.

While I was pleased the kernel panel helped him with his request, I know that approach doesn’t scale. Not every developer can attend our Summits face-to-face after all. This wasn’t the first time I had heard this call for help: in Japan, in Korea, in Taiwan, in the US and elsewhere, I frequently am asked: “How do I participate in the Linux community? What can I do to increase my chances for inclusion in mainline?”

This illustrates that for an open and thriving community, the kernel development process can still be daunting for many new participants, even highly technical and sophisticated ones. Although participation in Linux is growing steadily (over 1,000 developers from over 100 companies per kernel release) I knew we could do better.

So over beers that night, Jon Corbet, executive editor of LWN.net, and I sketched out a plan to change this. The result is a 30-page guide on exactly how to participate in the Linux kernel community. I think this material is a first of a kind and of the highest quality, not surprising since Jon wrote it. Jon is not only the Linux “chief meteorologist” he’s also its “Emily Post.” (Sorry Jon!) In this guide, he has chronicled exactly how the kernel development process works, why companies and developers would benefit from mainlining their code and common pitfalls along the way.

Why is this important to release now? As Linux use expands to new areas like mobile and sub-notebooks, it’s even more important to reach out to these new participants and make it as easy as possible to participate. The strength of Linux is its community. With this guide and the other help we can provide (like our Summits) we hope to encourage even more individuals and companies to participate. Please let me or Jon know if you have feedback on this work, or better yet, leave a comment on the LDN.

Popularity: 10% [?]

Deb Gage at the San Francisco Chronicle recently profiled a voting machine that will be given a tryout at a mock election at Linux World, opening today. Attendees of the conference will have the ability to cast their vote for one of the two candidates on the US presidential ticket. Besides obvious political fervor of many open source devotees, what’s the connection between this machine and Linux?

Dechert and a couple of colleagues founded the Open Voting Consortium, a nonprofit group dedicated to delivering “trustable and open voting systems.” In addition to lobbying against proprietary voting machines, they have spent the last several years working with scientists and engineers around the world to design and build a voting machine of their own.

Probably many of you have heard of the Diebold voting machines that were designed for use in elections in this country. They are closed and proprietary systems, and because of that, have serious security vulnerabilities that could result in election tampering. Besides security risks (as if there needs to be anything else), these closed systems are also extremely expensive, and just like with proprietary software, are designed to lock you in. Once election officials start using these machines they are beholden to that vendor unless they decide to phase them out. Because they are closed and opaque, no one but the vendors can work on the machines. Take in contrast, these new Linux-based machines:

At a price of about $400, the new voting machine is a tenth of the cost of proprietary machines - less if made in quantity, Dechert said - because it’s simply designed and based on free software. Its workings are transparent, he said, unlike some of the electronic voting machines that California decertified for security problems.

What would you rather your government spend your tax money on?

These Linux machines started me thinking about the ideals behind Linux (and open source) and how they could be put to work within the workings of government. The ideals I’m referring to are its process: transparency, accountability and meritocracy based on contributions. Last year I was quoted in an article about Open Legislation. What if our government officials drafted laws on a wiki (like the one powering Wikipedia or the Linux Foundation site). There you could see who actually wrote pieces of the legislation. You could see that a lobbyist from Chevron actually drafted part of the energy bill and then track that your representative “signed off” and then voted for it.
Or as Peter Leyden, director of the New Politics Institute, said in Open Legislation article:

“Laws go through all kinds of markups, changes and amendments,” Leyden said. “The process has evolved from making those changes on parchment to at least using word-processing documents, but it’s not that big a step to think of moving to the next generation of tools and crafting a whole piece of legislation on a wiki.”

The next likely step would be opening up the process further so that citizens could view and comment upon legislation in the works, or even — along the lines of California state Senator Joe Simitian’s concept in his “There Oughta Be a Law” contest — submit their own ideas, Leyden added.

We have the tools now for collaborative development: this has been proven in software and editorial content. Now let’s see if we can prove it with legislation. Collaborative, transparent development produces better software (and better voting machines). I think it can also produce better government.

Popularity: 14% [?]