Business Strategies for Open Source Players in Enterprise Computing
Strategic Computing and Communications Technology
MBA290C/EECS 201/IS224/E298A
Aaron Brick
Arindam Chakrabarti
Nicole Kim
Kedar Shah
1. Introduction
We intend to describe the primary incentives and strategies surrounding software openness. We analyze the decision-making for both suppliers and adopters, cast in the context of the current trends towards openness and expansion of IT markets. We will illustrate how the open source movement actively commoditizes sectors of the software market, forcing proprietary vendors to defend their market share through differentiation. This increase in competition will increase the value of software for customers and users.
2. Openness Incentives
2.1 The Open Source Spectrum
In proceeding to analyze software license decision-making, it is useful to consider existing and potential licenses along an axis, as an openness spectrum. Rather than being defined in terms of one specific criterion of these licenses, the measure is a general reflection ofoverall openness.
The most closed kind of software cannot be stored on a user’s machine; in these cases the closedness is so severe that the supplier manages the software's actual execution and exposes only an interface to its clients. Many online service providers follow this model.
Slightly more open and much more common is the case of the binary-only distribution; here users are given a copy of the software, in a largely unintelligible format. This is the model adhered to by Microsoft, Adobe, and othercorporate software suppliers.
In some cases suppliers will openly reveal APIs to their otherwise closed software and allow it to be “glued” together with other applications on the demand side. Some graphics driver software is released like this.
In uncommon cases, source code is distributed in a nondisclosable, hands-off manner. Microsoft was recently forced to do this with Windows for some of its most important customers. They were not allowed to modify or mention its contents, but their visibility made certain peripheral tasks easier.
From this point on, licenses are regarded as “open source.” Most common, and restrictive, among those is the GNU General Public License (GPL). It requires source distribution, credit where due, and for derived works to carry the same license (leading to the charge of “viral” licensing). Linux is probably the most significant example of software under the GPL.
The Berkeley Systems Distribution (BSD) license is so open that it allows redistribution for profit. The BSD OS kernel, having forked several times, was incorporated into Apple's OS X, which includes closed components and is sold. This would not be legally feasible to do with Linux.
The most open of all are works in the public domain (PD). They can be used under the least onerous license terms imaginable: none. However, limited amounts of software are available in the PD.
2.2 Trends
It was observed[1] that “the more liberal you make the terms under which customers can have access to your product, the more valuable it is to them.” Thus, all else being equal, customers will prefer more open systems. Exploitation of this fact by movement towards openness has accelerated in recent years: Closed software used to be the norm, but now almost all applications are available in an open form. This movement is closely related with the advent of the public Internet: it has lowered transaction costs so much that distributed collaboration is much easier and adoption does not require any special infrastructure.
Software opening is a form of commoditization; it only takes one supplier whose goal is complementary to a given system to commoditize it in order to broaden access to their profitable product or service. (An intriguing corollary is the supplier industry's panic over piracy; they are battling its commoditization with predictably little success.)
2.3 Incentives towards closedness for suppliers
The flipside of customers' preference for openness is the supplier preference for closedness. Most important for them is to retain their customer base, by ensuring that clients be locked in by high switching costs. This affords the vendor more control over the product's pricing.
Closedness is faster than openness because of the latter's high overhead costs. Suppliers can act unilaterally in the market with much greater agility than they can orchestrate standards agreements. This quickness and flexibility, together with marketing dollars, can potentially help them capture the first-mover advantage.
If a supplier has truly unique software, which cannot easily be re-implemented, it can charge a premium for it. Part of this premium usually comes in the form of closed licensing terms, restricting the transferability and transparency of the product. Attempting to take over a market niche with such a product is dramatic, difficult, and potentially very rewarding: this is the classic “performance play.”
Suppliers with unique software, marketing budgets, and locked-in users tend to be prestigious, established companies like Microsoft, Adobe, or specialty vendors. They are driven by profit and tend to be hierarchically organized.
2.4 Incentives towards openness for suppliers
Along with preferring openness, consumers are known to be fond of that which is free. Adoption can be meteoric when the product is available for no charge. Such products can potentially capture the first-mover advantage. Interested programmers can now hack together a program, upload and announce it, and receive feedback and participation on the order of days.
Several benefits are available to suppliers with large networks of users, especially those where trust is established. Users can constitute an important resource for quality assurance of products in development, because while less disciplined than the supplier's own professional testers, their numbers are so much greater. Plaudits and improved reputation can also be earned by suppliers seen as good citizens, contributors to the public good; marketing and advertising can even be colored by these aspects.
As previously mentioned, the commoditization of a niche of the software market happens when open alternatives to traditional, expensive software become available. Coopetition is a main manner of achieving the process: Groups collaborate and share the burden of reimplementation, then compete in the complements market.
Community-oriented commoditizers tend to be volunteer organizations or academic projects, meritocratic and driven by personal interest and altruism.
2.5 Incentives towards closedness for adopters
It was long assumed that good software was expensive and restricted, so this was users' choice by default. With the recent trend towards openness and much greater availability of open software, this supposition has been losing ground. That said, most organizations use some combination of open and closed software. This is unlikely to change overall, because proprietary suppliers will always dominate uncommoditized niche markets, and many users are in some niche or another.
One advantage of adopting closed software is that tight control generally corresponds to professional service. Buyers can be confident that versions, upgrades, and personal support will be available. It is also easy to identify the software's source, which helps in situations with rigorous responsibility tracking.
If a closed product is the market standard, then users may be coerced into adopting it. If one relies on direct network effects – trading files with other users – who use the dominant application, then there is a high barrier to adopting an incompatible solution. In other words, the storage and transfer of these files is not commoditized, as it is tightly controlled and developed to promote lock-in.
Proprietary, closed software has a deserved reputation for quality in certain areas. It frequently is more polished and caters better to the user than open products. The vendor might be known for the excellence of their products. After all, if people will buy it for a hundred or a thousand dollars, isn't it worth that much?
2.6 Incentives towards openness for adopters
This is the section with the most recent growth. The following incentives were not well-known previously because there was almost no software with which to observe it. The largest change enabling distributed development and release is, of course, the Internet. With its help, the market for software is much more liquid – and more sectors commoditized – than was previously the case.
As observed a priori, all else equal, customer value correlates with openness. Adopting open source software (OSS) means no exploitation by entrenched suppliers; lower lock-in by not sinking any purchase costs; and the assurance that document formats won't prevent migration. Being available for free, specifically, is a powerful motivator. David Wheeler recently estimated[2] the hypothetical cost of redevelopment of Linux at $612 million; it's a significant public good, easily sharable.
Source code is a program's complement. It enhances users' ability to debug, integrate, and improve the software. It will certainly be the case that an organization's own programmers are more responsive to its needs than those of a software supplier. This makes the product versionable on the demand side – users can themselves tailor the application for their needs. In the case of large deployments, this may very well be a better choice than building from the ground up or buying a supplier-versioned product. And if the improver isn't concerned about extracting competitive advantage from that slice of code, it might be a good unilateral contribution to the public good.
There are reasons to believe that open software is of high quality as well. Some programs have been in use for decades and are generally recognized as reliable. OSS features security through transparency, in which potential exploits are noticed, publicized, and fixed in short order. Active projects tend to have developer networks big enough to guarantee attention in the short term.
3. Openness Strategies
3.1 The Internet’s Impact
The introduction of the Internet has acted as a key enabler for the dissemination of OSS in general. It has made a significant impact particularly with the development process for the open development process, facilitating user accessibility and deriving complementary reactions in the market. As the very essence of OSS is based on collaborative efforts, the Internet has made it possible for users to easily contribute valuable input, improving the quality of a program at a much more rapid pace than proprietary software is developed. The Internet has also made distribution of OSS possible; for instance, it would certainly have been inconceivable for Linus Torvalds’ original attempts to have been distributed through retail chains and other such distribution avenues available to proprietary software. The Internet also helps crystallize demand-side network effects, allowing Linux users to collaborate with each other through bulletin boards and mailing lists to create a support network to replace support services offered by proprietary software companies for users of their software.
The steady growth of OSShas followed the lead of the explosive growth of the Internet. Studies show the percentage of Linux shipments has been consistently increasing since 1997[3]. By May 2000, 36% of public websites available on the Internet were run on Linux, demonstrating its growing popularity and acceptance by users.[4]Sales of Linux servers increased 200% from 1997 to 1998, 92% from 1998 to 1999, 63% from 2001 to 2002, and 54.6% from 2003 to 2003.[5]
3.2 Open Source Option Changes the Game
Proprietary software like Microsoft products have forced users and ISVs to become locked-in to their proprietary interfaces and format, confining users and ISVs capabilities to the functionality offered by a singular company. Microsoft has profited greatly by making itself the standard in B2B environments, exacerbating compatibility issues between business users not on the same platform. Yet with OSS, users are free from this confinement, enjoying benefits like low switching costs and customization options.
The use of open source programs frees users from some limitations that proprietary software places on users. Aspects such as licensing agreements, affordability, security, and reliability are not hindered by the restrictive nature of proprietary material.
3.3 Analysis of strategies being pursued by Linux vendors
The actions of the major Linux vendors, like Novell and Red Hat, exhibit interesting examples of coopetition. Red Hat, Novell, and SuSE (prior to being bought by Novell) were direct competitors. All three of them were selling Linux distributions: proprietary application software, device drivers, and middleware bundled around the GNU-controlled Linux kernel. However, all three of them were moving in divergent directions, and there was considerable pressure on each of them to differentiate in order to appear to be the most technologically advanced of all the Linux distributions, though, of course, all three used the same kernel as the core of their offering. This pressure to differentiate was leading to differences in interfaces, contributing to chaos, and hurting Linux adoption among mainstream, pragmatic customers.
In 2003, Novell bought a competitor, SuSE, for $210M. Currently, Novell, valued at $3B, and Red Hat, valued at $4.1B, are well-poised to cooperate to maximize the benefits they can achieve from demand-side network effects, and to minimize the lock-in and network effects of their competitors, primarily Microsoft and to a lesser extent, UNIX systems.
Red Hat and Novell are cooperating in the expected ways: they continue to support the Free Standards Group’s Linux Standards Base Project. Recently they both reaffirmed their commitment to LSB by supporting the LSB 2.0.1 specification released in September 2004. Interestingly, neither Novell nor Red Hat will benefit significantly from LSB, as the major application vendors for Linux-family operating systems already inter-operate with both Red Hat and Novell, the two biggest Linux vendors. The LSB specification will help smaller Linux vendors much more: smaller Linux-family OS versions like MandrakeSoft SA, Gentoo, and even a few larger ones like Debian and Cobalt (both small compared to Red Hat), which otherwise would have had a lot more trouble achieving inter-operability with Linux-based application vendors.
Thus, clearly Red Hat and Novell are giving up on differentiation on basics in favor of standardization of interfaces, even at the cost of helping their smaller competitors, and being forced to give up on technological innovations that would have helped them differentiate themselves against each other. This is clearly an example of coopetition, where competitors are finding it profitable to cooperate to help fight a much bigger and more powerful rival: Microsoft.
Novell, as one-time owner of the UNIX System V patents, trademarks, and copyrights, is in a unique position to defend Linux against legal challenges from SCO. In 1995, SCO’s predecessor, the Santa Cruz Operation, bought “all rights and ownership of Unix and Unixware” (according to the contract) from Novell; however, the corresponding asset purchase agreement filed with the SEC, specifically excludes “all copyrights” and “all patents” from the purchase.[6]The lawyers who drafted the SCO-Novell agreement in 1995 added a further paragraph that, when stripped of legalese, appears to assign to SCO the right to enforce Unix copyrights and patents, while it says nothing about whether this intellectual property is vested in Novell or SCO.
While SCO’s legal battles are, at least for now, primarily against IBM (IBM is accused, in a $1 B lawsuit, of allowing Unix code to get into Linux) and IBM customers using Linux (accused of using Linux, an alleged derivative of allegedly SCO’s Unix technology without permission from SCO), Novell has emerged at the forefront of the defense against SCO by repeatedly pointing out that it never sold any Unix copyrights or patents to SCO in 1995 (a claim SCO contests as we discussed above). Further, Novell declared in October 2004 that it would not use any of its Unix patents against Linux vendors (IBM has declared a similar policy). This is another example of coopetition, because IBM and Novell are direct competitors in the business of selling services for Linux platforms and applications.
3.4 Strategies of software industry to leverage OSS community
The Microsoft Office Suite, consisting of extremely useful and widely used programs such a Microsoft Word, Microsoft Powerpoint, Microsoft Excel, etc. is one of the major reasons for many businesses to use Microsoft platforms. All these programs use secret and proprietary data formats, and demand-side indirect network effects can make it very hard for a customer of enterprise computing to shift away from a Microsoft platform. The open source community has made numerous attempts to commoditize the Windows platform with projects such as WINE (Windows Emulator), but has not been successful. The Wine has been under development since 1993, but 11 years later in 2004, not seasoned enough yet to be a serious rival to Windows.
In 1999, however, Sun Microsystems noticed a company called StarDivision that had made a product very similar to Microsoft Office. Sun quickly realized that this product had great potential in reducing the customer lock-in and demand-side network effects of Microsoft Office. Sun acquired StarDivision in 1999 and invested significant effort on the project over a year, and came up with Sun StarOffice. In 2000, it forked the code base and donated one half to the open source community. However, the license terms under which Sun made it available to the open source community were extremely innovative. Sun is allowed to continue to collaborate with the open source community (represented by OpenOffice.org, the entity that manages OpenOffice, “OpenOffice.org” is an entity, not just a website address), and make its own contributions to the OpenOffice codebase, and it is also allowed to take snapshots of the code base to convert into proprietary StarOffice versions.