VenChar

Performance Oversupply and Linux on the desktop

During my two week vacation in India I got to read Clayton Christensen’s ‘The Innovator’s Dilemma’. One of the key concepts I picked up from the book is the idea of performance oversupply: In their efforts to provide better products than their competitors, suppliers often ‘overshoot’ their market: they give customers more than they need or ultimately are willing to pay for. Christensen argues that when performance oversupply occurs, it creates an opportunity for a disruptive technology to emerge and subsequently to invade established markets from below (by offering less functionality at a lower price). He also argues that performance oversupply tends to trigger a change in the basis of competition within a particular market.

Christensen argues that existing technologies or products have a tendency to migrate in a “north-easterly” direction (on the performance vs time graph) and steadily acquire more features and functionality in an effort to cater to the higher value (higher margin) customers. In doing so, they will tend to create performance oversupply scenarios that are ripe for exploitation by cheaper, simpler substitutes that have a reduced functionality set and can profitably cater to the lower end of the market. Complementary (and similar) to this notion of performance oversupply is the idea or principle of "good enough", that has been written about elsewhere recently.

An interesting example, discussed in detail in the book, is Intuit’s introduction of an accounting software package targeted at the SMB segment. Scott Cook, the founder of Intuit, made the following 3 observations about the state of the accounting software market for SMB: (1) previously available small business accounting packages had been created under the close guidance of CPAs and required users to have a basic knowledge of accounting (debits/credits, assets/liabilities, etc.) and double-entry bookkeeping. (2) most existing packages offered a comprehensive and sophisticated array of reports and analyses, an array that grew ever more complicated and specialized with each new release as developers sought to differentiate their product by offering greater functionality. (3) 85% of all companies in the US were too small to employ an accountant. Books were kept by proprietors or family members who had no need for or understanding of most of the entries and reports available from the mainstream accounting software. Cook decided that the makers of accounting software for SMB had overshot the functionality required by that market, thus creating the opportunity for a disruptive software technology that provided adequate, not superior, functionality and was simple and more convenient to use. Intuit’s disruptive Quickbooks changed the basis of product competition from functionality to convenience and captured 70% of the market within 2 years of introduction.

Christensen speculates that if one examined how people use Microsoft Excel, one might find that this is a case of performance oversupply, where the degree of functionality and the number of features provided far exceed what the average user (or even the power user) utilizes. He suggests that an opportunity might exist for a disruptive technology to invade this market from below. I would go further and suggest that there are broad classes of commercial software packages that are bloated, have way more features and functionality than the average user cares about or uses and that there exist open source alternatives that would qualify as “good enough” for large classes of users. One example that is beginning to get talked about is Linux-on-the-desktop as a good-enough substitute for Windows.

Now, performance oversupply can only be diagnosed relative to a specific class of users or market segment: what might be overkill from the perspective of one class of users might be minimal functional requirements from the point of view of another group of users or market segment. Which begs the question, what are the classes of users for whom the Linux desktop is already a good-enough substitute? This is one of the questions addressed by the excellent report Desktop Linux Technology and Market Overview that I read recently.

The report examines the state of the Linux desktop in terms off maturity, existence of core applications, developer support, etc.. In particular, it identifies a few classes of users for whom the Linux desktop is already a compelling alternative to Windows: (1) transactional workers, (2) small- and medium-sized businesses, and (3) public sector organizations internationally. For instance, the requirements for transactional workers (customer support staff, call center employees, bank tellers, sales reps, admin staff) are very different from those for knowledge workers in that (1) they spend most of their time using just a few apps, (2) they don't need the ability to run arbitrary third-party apps on their desktop, (3) costs savings (from using Linux as well as from other open source apps) are important. The primary weaknesses of Linux as a desktop environment namely: (1) broad application support, (2) ability to natively run office productivity apps like Office, etc. are not as important for this class of users and Linux, possibly in conjunction with thin clients such as Citrix, would be a good fit for this segment.

Christensen also argues that performance oversupply tends to trigger a change in the basis of competition - once the market’s demands along one axis of performance have been sated, other product attributes, whose performance have not yet satisfied market demand, come to be more highly valued and to constitute the dimensions along which vendors differentiate their products. As noted above, the disruptive introduction of Quickbooks into the SMB segment changed the basis of product competition from functionality to convenience. How is the Linux desktop going to redefine the desktop playing field? What are the new product attributes or measures of performance that will constitute the bases of product competition?

One tentative answer is "openness of source and formats". For instance, the Oregon bill that would require the state government to consider using open source software when doing procurements states: "It is necessary to the functioning of the state that computer data owned by the state be permanently available to the state throughout its useful life. To guarantee the succession and permanence of public data, it is necessary that the state's accessibility to that data be independent of the goodwill of the state's computer systems suppliers and the monopoly conditions imposed by these suppliers" (this is an interesting point that is also made in passing in the Desktop Linux Overview and it seems to refer to the perpetuity licenses of open source software as well as to open, standards-based formats for data storage that break the dependence on software from a single vendor)

Interview with Jim Ready of MontaVista Linux

Jim Ready of MontaVista Linux offers an interesting set of perspectives on embedded systems and the drivers for Linux adoption in that segment.

Ready makes the point that one of the distinguishing characteristics of the embedded market is the diversity of underlying microprocessor architectures - PowerPC, MIPS, ARM, etc. - compared to the desktop where the Intel architecture dominates. On the surface this presents a challenge to a vendor like MontaVista in terms of developing versions of the OS for each of these architectures. However, he argues that Linux is flexible enough to be easily adapted to all these environments.

An interesting observation, that resonates with some of my recent thinking, is that there is a trend towards highly networked devices that signals a move away from the traditional RTOSes that did one or two things really efficiently towards a more general purpose, powerful OS like Linux with a lot of networking capability built in. For instance, Nikon recently announced a digital camera with WiFi enabled. What this implies is that the device now needs to have (minimally) a WiFi driver and a TCP/IP stack - enter Linux. Quote: "A combination of factors has moved the center of gravity and design techniques toward a more sophisticated, more modern OS, such as Linux, and away from the traditional, dedicated RTOS-style architecture. It’s just that the average design is now more complex and has a higher networking requirement. That’s what’s shaping the market...It’s not that realtime requirements have completely gone away but that, in general, the trend is toward more high performance in silicon and softer realtime requirements for the application and embedded OS."

Sony, Matsushita and others recently announced the creation of the Consumer Electronics Linux Forum aimed at driving the advancement of Linux as an open source platform for embedded devices.