The Open Road

With open-source software businesses, you have two options. Actually, three, but the third belongs to Red Hat, and it applies to roughly no one else.

The first option is to sell support for open-source software. This option is generally advocated by those who have never grown a business beyond $10 million. It's a terrible model unless your only aspiration in life is to run a services company.

Hence, the support model might be good for Accenture or systems integrator, if they want to take on the burden of support, but it's a poor model for Red Hat, MindTouch, Microsoft, or other software company.

The second option is to contribute heavily to open source but not build your revenue model around monetizing that software directly. This is what the New York Times points to in its Sunday expose of allegedly fizzling open-source business models.

Open source can drive adoption like little else. It's not, however, necessarily a great driver of revenue. For that, you need to be selling something beyond the source code, and that "something" is often going to be proprietary, be it hardware, software, or a service.

Google is the master of this model. It gets roundly criticized for its half-open, half-closed open-source efforts, but the reality is that Google's products--Chrome OS, Android, etc.--are open enough to facilitate adoption without giving away the keys to the car, which drives wherever Google wants it to go.

That's the way successful companies are run: they take ownership of what they ship. They are influenced by but not controlled by the mystical whims of The Community.

Even Red Hat, which piggybacks on a lot of Linux kernel development, increasingly includes more home-grown software in its distribution and takes great pains to certify its Red Hat Enterprise Linux will work in the most demanding environments before putting its brand on the label.

Some, including me, have wrongly concluded that Red Hat's business model would apply to other product markets beyond the operating system. It doesn't. It only applies where the moving parts in the product are complex, multitudinous, and frequently changing.

For everything else, there's Option 1 (if you want a business that doesn't scale well or possibly at all) or Option 2 (which is really no different from the old proprietary model except that it effectively uses open-source complements to lower engineering costs and possibly sales/marketing costs).

Even Option 2 won't work if you under-invest in marketing and development, as Symbian is learning to its hurt. It turns out that there is no free lunch, even in the land of free software. It always takes work. And money. Lots of both, actually.

Some of us take longer than others. Tim O'Reilly moved on years ago from talking about open-source licenses and instead focused on the importance of data to business success. In the open-source industry, we heard his words but clearly didn't understand them.

We kept selling software through our "awkward teenage years," even as Google, 37Signals, Facebook, and others gave it away.

Years later, as Google pays for mountains of open-source code by aggregating data and selling data-rich services, we're starting to grok O'Reilly's message. It's what makes companies like Path Intelligence so interesting.

Redmonk's Stephen O'Grady notes:

Much has been made of the lack of an obvious revenue model for properties like Twitter, and to a lesser extent, Facebook. But when looking at the organizations' balance sheets...it seems self-evident that the value of the data assets involved is seriously underreported...

The economic value being assigned to data helps to explain why, while being sympathetic to questions about Twitter business models, I've never been overwhelmingly concerned. Where the revenue model for the dot-com era "eyeballs" strategy was equal parts indistinct and aspirational, the Web 2.0 businesses are being built out in an era of customers increasingly predisposed to analytics and data driven decision making. In other words, there's a market for their most valuable asset.

As Microsoft's Windows, Office, Xbox, and SharePoint businesses demonstrate, the real money is in the platform business, which is, or which can be, a data business. The more businesses and developers that build upon your software, the more valuable that software becomes. Even systems like Twitter are being turned into platforms.

But how you build the platform is increasingly important. Microsoft is Platform 1.0. Open source is Platform 2.0. It's a more efficient way to build community around a core, which is why Google and other savvy companies increasingly turn to open source as a fundamental way to entice developers, which developers create more software which invites more adoption which yields more data...you get the picture.

It's also why I believe Google Android, in its platform battle with Apple's iPhone, will ultimately prevail, so long as it can work in peaceful coexistence with the developer community (which has not always been the case).

Unlike many open-source companies, however, Google et al. have the singular benefit that since their business is data, not software, they can shepherd open-source development without taking a heavy hand in community management. More open source leads to more adoption, which leads to more data, which leads to the Googles of the world being able to give away even more software for "less than free."

It's genius. And it's amazing that it took so many of us so long to heed the counsel O'Reilly offered years ago.

In sum, this isn't a suggestion that companies should forgo profits in exchange for mindless popularity contests, as 37Signals' Jason Fried rightly pillories.

Instead, it's a call to look for ways to fund open-source development with rich, data-driven businesses. Most open-source companies focus too much on software, and most Web 2.0 companies focus too much on data. It's the blend of the two that makes a company successful.

Just ask Google.

(As an end note, I think Gartner's Brian Prentice is on to something when he speculates that enterprise applications may increasingly be communally developed by IT end users, though perhaps coordinated by vendors. It's a very interesting prospect, one that will enable even more open-source development in an area where data may not fund it.)

Open source offers a fantastic way to reach developers and users of one's technology. Ironically, however, the very group most inclined to adopt open source is the least likely to pay for it.

Therefore, to make an open-source business thrive in enterprise software, vendors must learn to distinguish between developer-users and IT operations-buyers. As I'll explain, however, open-source companies may need to guard against becoming too successful in order to preserve their exit opportunities.

It is, of course, quite possible to make money in open source. Lots of it. Red Hat, for example, is approaching $1 billion in annual revenue. MySQL had generated more than $90 million in sales the year it was acquired by Sun Microsystems for $1 billion.

That's real money.

It doesn't, however, come from the developers that download open-source code. Developers, in former MySQL CEO Marten Mickos' words, "spend time to save money."

Hardly the ideal customer.

Developers download software, which a great way to initial a buying conversation but a terrible way to finish it. Open-source companies talk about selling support, but this is a losing proposition. Developers, after all, are highly likely to support themselves through online forums or other means. They don't pay for software, and they don't buy support. Not most of them, anyway.

This is one reason that pure-play support models simply don't scale in open source. They focus on the exact wrong audience.

Sure, there's a honeymoon period for new open-source companies that launch support offerings around established community-led projects. Some developers buy support, either through personal need or corporate requirements. After that initial rush for support, however, it's a tough slog selling support to developers. It's like selling ice to Eskimos.

This brings us back to a real dilemma in open-source companies: how to monetize popularity (i.e., downloads).

Developers are the most efficient way to spread adoption of one's product but perhaps the least efficient way to monetize it. To get paid, vendors must learn to separate IT developers from IT operations, and build offerings for both.

Red Hat is a classic example. People think that Red Hat sells support. It doesn't. Not really.

The primary reason enterprises buy a Red Hat Enterprise Linux (RHEL) subscription isn't for Linux support, and certainly isn't for the bits: you can get the bits free from CentOS, and support comes heavily discounted from Oracle.

No, the reason companies purchase a RHEL subscription comes down to certification that RHEL works with a wide variety of hardware and software, as well as with the Red Hat Network, which delivers updates to an enterprise's RHEL servers.

In other words, IT operations pay Red Hat to help manage their Linux servers in production. The money is in operations.

Red Hat isn't alone. Look at JBoss. The company started minting money, once it licensed Hyperic's software to build the JBoss Operations Network.

SpringSource took it one step further and actually bought Hyperic, the company, as the foundation for its Build-Run-Manage message, a message founded in selling to IT operations, not developers. (Rob Bearden, chief operating officer at SpringSource, was deeply involved in both decisions and remains one of the smartest people in the industry on building open-source businesses. If there's any wisdom in this post, it is his.)

For new open-source companies grappling with how to supercharge sales, the answer is operations. It may not be a systems-monitoring tool like Hyperic or Zenoss, but it likely is about systems management, as operations need and pays for it.

There you have it: the secret to your billion-dollar open-source opportunity. Except for one niggling fact: despite the value of IT operations to make sales, it's really developers who create the most company value, from an asset perspective. SpringSource's sales didn't justify its $420 million valuation. Its developer base did. Developers have strategic value, in terms of IT operations and creating tactical value.

In fact, SpringSource's valuation might well have gone down, had it been making more money, just as TechCrunch's Michael Arrington astutely argues could happen with Twitter. Sales provide a measurable, tangible valuation. Developer traction creates an amorphous, strategic value.

Hence, while IT operations is the crux of making sales in open source, it might well be that open-source companies should focus on community development and avoid making too much money so that they can maintain a healthy valuation. But not too healthy: there isn't an incredible amount of IT vendors that can swallow $1 billion acquisitions, the IPO era seems to be over.

Is this the new open-source entrepreneur's dilemma?

Follow me on Twitter @mjasay.

Each year InfoWorld sets out to rate the "best open source products" with its Bossie awards. Too bad it has decided to cloud the voting with open-source politics, as well.

The politics reveal themselves when InfoWorld tries to settle on a winner between Zenoss and OpenNMS. (Why Hyperic isn't also in that mix, or Reductive Labs' Puppet, I can't fathom, but...)

The editors write (note: the emphasis is mine):

Although Zenoss clearly has the more developed feature set, our Bossie goes to OpenNMS. The reason boils down to business models. OpenNMS is a purely open source software project, meaning that customers get the complete set of features available for free as open source. There is no "enterprise" version. OpenNMS makes its money strictly by selling support and training services.

Zenoss uses a common business model in the open source world: it provides an open source version of its software with a limited feature set for free, and it sells a more extensive "enterprise" version of the software with support through an annual subscription. So while Zenoss may be a good value compared to HP or IBM or CA, it's not a good value compared to OpenNMS.

If only enterprise IT could cavalierly discard superfluous things like "features" in favor of licensing ideology. But it can't, which is why Agilent, Telstra, Accenture, MySpace, and other companies that need enterprise-grade network management systems have been opting for Zenoss. They seem to need those pesky "features" that InfoWorld glosses over. They're buying a product, not a political platform.

Regardless, if we allow business model to be a valid factor in InfoWorld's decision criteria, how are we to explain its contradictory decision to judge Intalio the winner in the Business Process Management (BPM) category? The editors reason:

Intalio has been criticized regarding its open source claims, most likely because the company does not provide source code on its Web site (where binaries of the free community edition can be downloaded). However, Intalio's enterprise edition customers do get full access to source code, and the source code of community edition components -- which fall under Apache and Eclipse licenses -- are obtainable from their community-based repositories....

However, new beta features reflect enterprise needs, including a business rules engine, Ajax-driven forms for easier editing, and a more streamlined deployment interface. The full enterprise edition also includes BAM (business activity monitoring), a portal interface, ECM (enterprise content management) based on Alfresco, fail-over clustering, and support for application servers beyond Apache Geronimo.

I think Intalio is great, but I can't understand why Zenoss' business model is considered a demerit but for Intalio, which has the same model, it's a non-factor. Zenoss also provides source code to its enterprise customers, so why is Intalio right because it provides an enterprise-class experience with an Open Core model but Zenoss is wrong for doing the exact same thing?

Personally, I think awards should be given based on the merits that will most appeal to IT buyers, and such will have little to nothing to do with business model nuances and everything to do with solving business problems at a compelling price. If Zenoss is the better enterprise IT bet, shouldn't it get the Bossie, regardless of OpenNMS' licensing model?

InfoWorld set out to name the "top open source products." By deciding, instead, to name the top open-source products and business models, it has failed to serve its audience as well as it has in the past. The Bossies are still a good resource, but it's best to read the reasons behind some votes carefully, as they may have nothing to do with the products at all.

Follow me on Twitter @mjasay

Drupal is a fantastic Web publishing platform that derives much of its value from a disparate community of contributors, as Xconomy recently wrote. With more than 4,000 contributed modules from over 3,000 active contributors (741 of which contribute to Drupal Core), Drupal has something for everyone, which is both its greatest asset and biggest liability.

Choice is good. Too much choice, however, can be bad.

The same holds true for Red Hat, which charges a premium for its Red Hat Enterprise Linux distribution to enterprises that want to tap into Linux but don't want the bother of rolling their own version of Linux from Kernel.org.

The problem, however, is that such a business model depends upon the complexity of the underlying platform. If that complexity goes away, does the business model?

The Drupal-focused company Acquia is thriving because deploying Drupal, what with its myriad of choices, can be complex. Ditto for Red Hat. There are thousands of packages that comprise Linux, making it worthwhile to pay a trusted guide like Red Hat.

But Red Hat CTO Brian Stevens suggests that "you need product complexity" to make its model work. Does this mean that Red Hat and Acquia will be hitting the unemployment lines if they do their jobs too well, and Drupal and Linux become too easy to use?

I don't believe so, but that is certainly one way to read IDC data that shows unpaid Linux adoption is now outpacing commercial Linux adoption, which has the potential to disrupt Red Hat, Novell, Canonical, and other Linux vendors.

It's less an issue, however, for Acquia, which has been rolling out for-free software and services to augment Drupal, including enterprise-class search. Acquia's Drupal distribution teases the complexity out of Drupal, but it also extends beyond Drupal with enterprise-only features. It is therefore well-positioned to foster the Drupal community while simultaneously feeding its top and bottom lines.

For this reason, I believe Acquia's model has more staying power, though it has a long way to go before it generates Red Hat's nearly $1 billion in annual revenues. Acquia's model allows it to thrive even if Drupal becomes easy, and also affords the company greater latitude to enter new markets, like the application market, where product complexity is not as much of an issue.

Red Hat, on the other hand, must resolutely focus on core infrastructure, an area of the software stack that is naturally prone to such complexity, because of its adherence to its complexity-dependent business model.

Both models are generating impressive returns. But Acquia's open-source model seems to offer more potential in the long term than Red Hat's does.

Follow me on Twitter @mjasay.

For all the value that open-source development provides, Red Hat CEO Jim Whitehurst recently reminded the industry that open source may be more potent as a business model than as a development model:

Open-source development is great and all that, but I think more of the value of Red Hat comes from our open-source business model than from the development model.

The truth of this statement came home clearly to me in a conversation with Michael Mullany, vice president of marketing at Engine Yard, a cool company that provides on-demand deployment and management solutions for Ruby on Rails applications. Mullany walked me through some calculations he has been running to determine optimum business models for Engine Yard and others in the industry.

As Mullany tells it, despite all the chatter about free and freemium business models, the debates have yielded a paucity of valuable, actionable information. A mostly free business model like Craigslist rocks. A free but poorly-monetizable business model like YouTube does not. There's more to "free" than a price tag. But how do you determine when, where, or whether to use "free" as part of your business model?

After running a variety of models, Mullany filtered down the free (or freemium) vs. non-free debate to just four factors:

• The cost of marketing and selling to a user in a paid model
• The cost of serving a free user
• The cost of acquiring a free user
• How successful you are at converting users from free to paid (whether it's a cross-sell or a premium offer)

If these factors trend one way, in Mullany's math, free or freemium is a good idea. If the factors trend another way, then paid-only is a better idea. (Free trials are practically always a good idea--that's not what this debate is about.)

So how does this work? Here's where Mullany's mathematics background kicks in, so follow his formulae carefully.

Before we talk about free, we should talk about "paid," where the basic formula is: Sales - Costs = Profit. Or in long form:

Price Paid By User - Cost of Providing the Service - Cost of Selling & Marketing to Signup a User = Profit Per User.

Or in formula form:

PP - PC - PS&M = PPU

Free or freemium models are more complicated, however. Generally, you have a free offering that comes with restrictions and limitations, and a premium, paid offering that you sell as an upgrade. On one side, you spend money to get free users to sign up to a basic level of service (SEO and PR aren't free!), and then you spend more money to provide that free service.

On the other hand, you usually get to convert some percentage of those users to premium users with minimal additional costs in sales and marketing.

With this in mind, Mullany's equation for freemium businesses looks like this:

Price Paid by Premium User - Cost of Proving the Service to the Premium User - [ (1/Ratio of Premium:Free Users) * (Cost of Free Service Per User + Cost of Marketing to get a Free User) ] = Profit Per Premium User

Then in formula form (labeling the percentage of Premium to Total Users the Conversion Rate (CVR)):

PremP - PremC - 1/CVR * (FreeC + FreeS&M) = PremPU

So...to figure out which is the better model for your business, compare the two profits. Freemium is better than a paid model if:

PremP-PremC - 1/CVR * (FreeC+Free S&M) > PP-PC -PS&M

As an English major, I was asking Mullany for some simplification at this point. Fortunately, he provided it because, as it turns out, the price and cost to serve a premium versus a paid user will be fairly close, allowing us to cancel out on both sides.

This leaves the formula looking like:

1/CVR x (FreeC+Free S&M) > PS&M

With a little algebra you get a magic formula of sorts. Namely, freemium will be a better choice when:

Conversion Rate % > (Cost to Serve a Free User + Cost to Acquire a Free User)/Cost to Acquire a Paid User

From some informal surveys, it seems like best-in-class freemium conversion rates are anywhere from 2 to 8 percent (at the very high end). The implication of this, as Mullany noted, is that for free or freemium to make sense, your cost to serve and acquire a free user has to be from between one-twelfth and one-fiftieth the cost of acquiring a customer under an alternative paid model. (This is one reason why successful freemium businesses have a laser focus on conversion rates.)

Based on this (perhaps overly simple formula), it becomes clear why freemium models make sense when a business has either very strong word-of-mouth marketing, a very low cost to support a free user, or where the cost of sales and marketing to acquire a paid user is relatively very high. If your marketing costs are high, it can be either because it's an audience that's really hard to target or because it's a very new product category that requires a lot of education and assistance for a potential paying user to understand.

I really like the framework Mullany brings to open source, in particular, where "freemium" translates to Open Core and other business models.

His formula removes some of the dogma from open-source debates. The question is how to solve customer problems at the lowest possible cost/highest possible value, not whether one should join the Church of Free Software.

Sometimes "free" is absolutely the right way to provide customer value. And sometimes it is not. Mullany's formula helps to elucidate the decision-making process.

Follow me on Twitter @mjasay.

In 2008, Todd Barr left Red Hat to try his hand at an open-source start-up. Now, one and a half years later, Barr is doing just that, hoping to turn the telephony world upside down at Bandwidth.com, a company that operates a proprietary voice network, but is betting a great deal on the open-source FreePBX project.

This shouldn't be surprising, given the increasingly intertwined fates of proprietary and open-source software. But it's interesting to watch Bandwidth.com--a 10-year-old company that sells business voice and data services--incorporate open source into its strategy. After all, the company recently built its own, nationwide CLEC IP voice network. It has a lot riding on that proprietary network, including 4 billion minutes of voice traffic in 2009.

So why bother with FreePBX, an open-source graphical user interface for managing Asterisk, the leading open-source telephony engine? That's the question I took to Barr. His answers are instructive for anyone hoping to leverage open source into their business.

FreePBX is mostly invisible: not many people know about it but it just happens to be the basis for the interface that you see if you're using AsteriskNOW or Trixbox CE, and many other open-source PBX distributions. In total, FreePBX is running on over 300,000 phone systems (with 3 million downloads, to date)...with virtually no one knowing they're even using it.

This is an impressive level of adoption for an independent open-source project. Now Bandwidth.com, and the FreePBX community, are investing in the next generation of FreePBX.

Why? Why is Bandwidth.com spending so much money developing software that it has no interest in selling?

Bandwidth.com's interest is in selling network services, not software. But that software provides the basis for a larger community's interest in Bandwidth.com's network services. The FreePBX community, in turn, benefits from having access to Bandwidth.com's network features, but for the most part, it focuses on technology innovation that benefits other communities such as Asterisk and FreeSWITCH.

Or, to make an analogy to Linux, FreePBX is like Gnome or KDE, and the open-source telephony engines (Asterisk and FreeSWITCH) are like the kernel. While the kernel makes the operating system function, the user experience and key tools for the application development environment is really driven by the user interface, as with FreePBX.

With the preview release of FreePBX version 3 on Tuesday, FreePBX now supports FreeSWITCH (and soon Asterisk, as well), making it completely engine agnostic. Not only does this expand the footprint of open source in telephony, it also gives customers choices, so they can pick the right engine for their use case.

It also turns open-source telephony inside out, setting up FreePBX as the focal point for future innovation in open-source telephony. I suspect that FreePBX's new modular architecture and standards-based frameworks may inspire application developers to target new telephony applications to FreePBX to be able to run with any "engine".

All of which should help Bandwidth.com. Barr told me that the key to innovation is better linkage to network functionality. Because of its network, Bandwidth.com is in a position to drive innovation through FreePBX, faster. The more FreePBX adoption, the strategy goes, the more consumption of Bandwidth.com services.

Bandwidth.com launched a Developer Sandbox Program on Tuesday that aims to give developers enhanced access to its VoIP network features. The idea is that developers can now make use of open-source software and open access to network functionality and develop a new generation of applications that function seamlessly with the voice network.

As an example, Bandwidth.com already has a contextual "store" in beta that allows FreePBX users to turn-on dial-tone and get phone numbers right from the interface.

As Barr puts it, FreePBX, backed by Bandwidth.com, is about lowering the overall cost of telephony for customers and having better open-source technology that encourages new network-aware application development.

It's a classic razor/blade model, as Bandwidth.com heavily invests in open-source complements for its network services core. The model seems to be working for Bandwidth.com. Can 4 billion voice minutes be wrong?

Follow me on Twitter @mjasay.

For years, Microsoft had it easy. The two busiest groups within the software behemoth were the accountants, adding up all the billions in profits, and the CD/DVD burning team, which simply had to burn more copies of Windows and Office to keep up with demand.

Today, life is a little less rosy for Microsoft, as it calls out in its recent 10-K filing with the Securities and Exchange Commission. As TechFlash highlights, Google Android is now called out by Microsoft as a competitor, as are Apple, Opera, and Google in browsers, whereas only Mozilla was deemed worthy of Microsoft's competitive glare in 2008.

Even more interesting to me, however, is Microsoft's acknowledgment of its increasing competition with Red Hat, Canonical (Ubuntu), and other open-source companies given the different business models these companies use to undermine Microsoft. Yes, Red Hat has made Microsoft's hit list for years, and it has also called out open source as a threat, but there's a far greater variety of open-source threats for it to worry about now.

Accordingly, Microsoft's annual report for its fiscal year 2009 reads like open-source Whac-A-Mole.

Microsoft highlights its 'desktop' competitors...

Client faces strong competition from well-established companies with differing approaches to the PC market. Competing commercial software products, including variants of Unix, are supplied by competitors such as Apple, Canonical, and Red Hat. Apple takes an integrated approach to the PC experience and has made inroads in share, particularly in the U.S. and in the consumer segment. The Linux operating system, which is also derived from Unix and is available without payment under a General Public License, has gained some acceptance, especially in emerging markets, as competitive pressures lead OEMs to reduce costs and new, lower-price PC form-factors gain adoption. Partners such as Hewlett-Packard and Intel have been actively working with alternative Linux-based operating systems.

The Windows operating system also faces competition from alternative platforms and new devices that may reduce consumer demand for traditional PCs. Competitors such as Apple, Google, Mozilla, and Opera Software Company offer software that competes with the Internet Explorer Web browsing capabilities of Windows products. User and usage volumes on mobile devices are increasing around the world relative to the PC. OEMs have been working to make the Google Android mobile operating system more compatible with small form-factor PCs or netbooks.

...and server competitors...

Our server operating system products face intense competition from a wide variety of server operating systems and server applications, offered by companies with a variety of market approaches....Nearly all computer manufacturers offer server hardware for the Linux operating system and many contribute to Linux operating system development. The competitive position of Linux has also benefited from the large number of compatible applications now produced by many leading commercial and non-commercial software developers. A number of companies supply versions of Linux, including Novell and Red Hat....

Our Web application platform software competes with open source software such as Apache, Linux, MySQL, and PHP, and we compete against Java middleware such as Geronimo, JBoss, and Spring Framework.

...and even office productivity application competitors:

Competitors to the Microsoft Office system include many software application vendors such as Adobe, Apple, Corel, Google, IBM, Novell, Oracle, Red Hat, Zoho, and local application developers in Asia and Europe....The OpenOffice.org project provides a freely downloadable cross-platform application that also has been adapted by various commercial software vendors to sell under their brands, including IBM, Novell, Red Hat, and Sun Microsystems.

One should assume that Microsoft exaggerates the impact of competition in order to stave off further antitrust inquiries, but it's telling that it calls out so much open-source competition for its broad product portfolio.

Of course, open source is also increasingly a tool used by Microsoft to distribute and develop software. Microsoft must choose whether to be a friend or foe to open source, especially as it seeks to build its next billion-dollar businesses.

But given that Microsoft opts to call out the fact that "Certain 'open source' software business models challenge our license-based software model" and that "To the extent open source software gains increasing market acceptance, our sales, revenue, and operating margins may decline," it seems that Microsoft is preoccupied with open source as a threat, rather than as an opportunity. That's unfortunate.

Follow me on Twitter @mjasay.

Matthew Szulik, Red Hat's former CEO and current chairman, has been in semi-retirement for the past two years, but you'd never know it from listening to his interview with the BBC's Peter Day. Szulik, ever the revolutionary, talks up open source's opportunity to disrupt conventional software and promote social reform.

As he does so, however, he inadvertently describes Red Hat's winning open-source business model as directly parallel to the Web 2.0 business models deployed by Google and others. While this isn't surprising (I've written about it before), it was the first time I've heard Red Hat state it so baldly:

Think about [open source] like mining. There's all of these natural resources buried under the ground.

But unless you have a large aggregator to pull it out of the ground; to take the gold or the coal or the copper and turn it into something usable, then it's really not useful for mankind.

[Day interjects, "But Linux was usable before, wasn't it?"]

It was usable by the 14 people that are sitting in Cambridge or Oxford right now. It was really not a mainstream technical activity.

Red Hat develops some of the open-source software it distributes. It is the primary contributor to the Linux kernel, writing 12.3 percent of the Linux kernel, and spends, by Szulik's estimation, "over $100 million per year to advance Linux and open source [development]."

But Red Hat, like Google, Facebook, and other "Software 2.0" companies that heavily employ open-source software, gets more than it gives to open source. It is a net beneficiary.

Sure, as Szulik goes on to explain to Day, Red Hat "puts all of that [$100 million worth of] intellectual property into the public domain," which differs from Google and others that write a lot of proprietary software that is never shared with external developers, but the core model of building upon open-source software and wrapping it with proprietary services is very much the same model employed by every significant Web company today.

I'm also willing to bet that Google contributes more than $100 million worth of intellectual property as open-source software every year, making it an even bigger open-source company than Red Hat, at least in terms of lines of code contributed.

I used to complain that the Web companies pick and choose where to contribute back, but it all seems to work out in the end. Yahoo works on Hadoop while Red Hat releases KVM while Google releases Gears while...you get the picture. Different companies releasing and "free-riding" on different open-source code, with the entire industry growing and benefiting in the process.

Follow me on Twitter @mjasay.