The Increasing Tail

If you've worked in the startup world for a long time, you inevitably have to ask yourself one question, "Why do certain companies become successful and others do not?" While it seems like the answer is a tangled morass of variables, every once and a while a theory emerges that can help untangle the hairball. This piece is about an old theory that was revolutionary (increasing returns), a new theory that has a lot to tell us (the long tail), and how we can glue the two together to predict some winners in the near future. Consider this my ode to IT 2.0.

The Increasing Tail

Or how Microsoft might be loosing at their own game

For many, Santa Fe’s beauty is derived from its simplicity. It’s a city widely known for many things – beautiful architecture, Native-American art, and the second highest per capita rate of massage therapists in the United States. However, just off of Hyde Park Road outside of downtown Santa Fe sits the most complex place you will ever find. Housing a continuously rotating lineup of 35 visiting professors at any time, the Santa Fe Institute (and the academics that take temporary residence there) focus on one thing - “The Science of Complexity”. Apparently, mere mortals and quantum physicists need not apply (although I do think they sneak a few physicists in the back door to keep the economists straight). What’s fascinating about the research done there is that complexity is really just a more academic way of saying “the real world”. And oddly enough, despite its focus on complexity, every once and a while a pretty simple concept pops out of the place.

In 1996 working out of the Santa Fe Institute, Brian Arthur published a seminal paper on the dynamics of the technology economy titled “Increasing Returns and the New World of Business”. Shattering the classic “Decreasing Returns” models that had been the foundation of economic analysis since the end of the nineteenth century, he essentially proposed a viable model to answer the $60,000 dollar question in the technology business, “why the hell has Microsoft been so damn successful”.

Technology Encrusted Sea Bass

Around the same time I learned about Brain’s theory, I took a career “diversion” (let’s call it) into the restaurant and bar business. Without digressing very far into my experiences with the restaurant industry, let me just tell you that Dante’s 9^th level of hell is pretty much reserved for restaurateurs.

Anyway, the concept of Increasing Returns (IR) is quite simple when contrasted to Decreasing Returns (DR). DR economic theory basically states that over time, it just ain’t that profitable trying to make more of the same thing. While Brian’s used the example of a coffee plantation as the classic DR industry in his paper, let me translate this into something I painfully know about: restaurants and bars. As an owner of a restaurant or bar, you have a fixed number of seats to fill. You spend a certain amount of money to build out your space to get these seats. To support these seats you need to hire a wait staff, bar staff, and kitchen crew. You need storage for enough food and booze to handle the potential nightly demand of your space. But as your place becomes popular, at some point in time you are going to stress the system. In high end restaurants, a qualified wait-person can handle four, maybe five, tables at the same time. A bartender can only crack open Heinekens at a certain rate, and your kitchen crew only has six to eight burners on the stove with which to concurrently sear foie gras and encrust sea bass with almonds. To boot, the physical restaurant is constrained by seats, space, and bathrooms. You want more seats, you’ve got to build more bathrooms. Seriously, in most city building codes, when you whittle down all of the architectural variables that matter, it almost always comes down to how many bathrooms you need.

This inherent physical limitation of the “plant” makes it hard to continuously expand to meet increased demand. While an initial investment in build-out nets you a certain amount of seats, you have to re-up that investment all over again to legally allow one more seat past your original capacity. And because of the labor component, the cost of serving one plate of food is about the same as serving the second. This decreasing return on continued investment leads to a competitive market where a number of restaurants open, all eventually meeting the same constraints of their physical plant. This competitive market approaches a sort of détente equilibrium between a few popular players all fighting over the same annoying customers who want to argue whether the appropriate color of a medium rare filet is “pink”, “salmon”, or “light coral” (ahem, I digress). In this market, no one can get that much ahead of each other because no one can break the inherent DR barriers of their physical plant. As Brian points out, we see the same dynamics in other hard goods markets like retailing, steel production, real estate, and oil.

IR theory on the other hand emerged when for the first time, the production of the second, third, fourth and eventually millionth item cost vastly less to produce than the first. Software is clearly the obvious example. It takes a lot of money to develop a piece of software (think of this like the build out of a restaurant - Brian quotes the 50MM it cost to build the first copy of Windows) but the second, third, fourth, and millionth copy cost only $3 dollars for the disk.

But Brian’s argument is not to state the obvious: that bits are cheaper than atoms to reproduce (I think Nicholas Negroponte has cornered the market on this one). If that were merely it, then anyone who started a software company would be sipping Bellini’s with Gates at the Redmond country club. The important point is that in an IR economy there can be one runaway leader as opposed to a few big players tending towards a multi-player market equilibrium. The real fun is in understanding why there can be runaway leaders in an IR economy. And to go further, if you can actually understand this phenomenon, could you artificially cause it to happen?

 Runaway Leaders

There is clearly more to this iceberg than the tip I have shown so far. Brian hypothesizes three key aspects of the IR economy that allow this runaway leader phenomenon:

Up-front Cost versus Unit Production Cost

High-tech products (hardware, software, telecommunications equipment, etc..) are by definition complicated to design and to deliver to the market place. This has two effects. First, it limits the number of players at any given table in the “Casino of Technology” (as Brian likes to put it). Second, these products typically have R&D costs that are extremely large relative to their unit production costs. Compare again the $50MM to develop the first disk of Windows and the ongoing cost of $3 for each subsequent unit.

Network Effects

Many high-tech products need to be compatible with a network of users or an ecosystem of technology. The more prevalent a technology, the more likely it will emerge as a standard causing other technologies to integrate into it. This creates a positive feedback loop of adoption.

Customer Groove-In

High tech products are typically difficult to use and learn - they require training. Once users invest in this training (e.g. learning to use Windows versus a Mac) they merely need to incrementally update these skills to migrate to subsequent or peripheral versions or products. This allows IR players to constantly expand their market footprint at a small cost to their users (I would assert this is similar in nature to the “beachhead” concept that Geoffrey Moore immortalized in “Crossing the Chasm”).

 
Splitting Software Hairs

One distinction I will make here is between products and platforms. I firmly believe that the runaway leader phenomenon describes the adoption of platforms not point products. What Brian is really talking about is how companies can lock customers into a platform, causing daunting costs for their users to switch to an alternative. Through this lens, I consider point products as a strategy to drive further lock-in into that platform. That is by no means to say that point products are not important. In fact, I’ll argue in a little bit that Brian’s IR theory misses a key component critical to the success of the every runaway leaders he talks about. And surprisingly, it has to do with products not platforms.

With this distinction in hand, let me flatten the foundational components of Brian’s theory into a timeline of how these events play out. I’d propose that the path to becoming a runaway leader according to IR theory is a two phased approach. The first phase is to get massive distribution of your platform. The second is to drive adoption of the platform by luring users to it by delivering killer applications (point products) on top of it. Use of these products (and hence the platform) drives demand for other technologies to integrate into the platform (and products), and thus the positive feedback loop starts spinning. As I have alluded to above though, I believe there is still something critical missing here. It’s a required third phase of the game plan. But IR theory isn’t going to get us there; we’re going to have to use another emerging theory called “The Long Tail” to explain its importance.

 
Why Two Mouse Buttons are Better Than One

Let’s take a look at the IR gold standard to see how this all played out. You see, even though it took them 10 years, Microsoft gets a gold star for executing the first near-perfect two-phase IR lock-in strategy (apparently I am experimenting with hyphens in my writing style here). Their first phase success was clearly due to their mastery of physical distribution. Microsoft’s early deal with IBM to bundle MS-DOS enabled them to distribute their platform far, wide, and in rapid fashion. With wide distribution under their belt, Microsoft slowly migrated their existing customers to newer and newer versions of MS-DOS, and then eventually to Windows. In their second phase, which could only be catalyzed by their success in the first phase, they drove user lock-in by delivering the right killer applications on top of their platform. As time has told us, users rarely stick with an operating system because it looks nicer or uses two mouse buttons as opposed to one. Users almost always stick around for the applications. Microsoft figured this out very early on. While it took them some time, in a world where MS DOS was still one of many operating systems (remember DR-DOS, CP/M, or one of the 17 flavors of Unix available, etc..), Microsoft levered platform lock-in by using application lock-in with the Office suite.

Ironically, the first job I ever had in the high tech industry (circa 1989) was actually at Digital Research (makers of DR-DOS). Without knowing it, I was watching this whole IR pageant unfold from the inside. I remember that at the time, DR-DOS was a vastly superior product to MS-DOS in terms of functionality. But Digital Research’s killer application was not a word processor. It was “Gem Draw”, a graphics program! As a related side note, Novell, still a major competitor to Microsoft at the time did try and follow the two phase strategy. Unfortunately it was about 4 years to late. Having succeeded in the platform distribution phase with a very large NetWare user base, Novel attempted further lock-in of their users by first buying Digital Research, and then finally purchasing WordPerfect to copy Microsoft’s applications move. Unfortunately this was a Hail Mary pass much to late in the game; Microsoft was already working on the third phase I’ll talk about in a second.

Technology history is littered with examples of players who only got to play the first phase of the IR game. Take the Mac for example. Using a Mac is a pleasurable experience. It is sort of like shopping at Whole Foods but without the hippies. At the end of the day though, the Mac never had the key applications that brought users to the table and locked them in. The folks at Apple delivered applications based on the Mac’s key strengths, not on the needs of a mass market user base. Graphics and multi-media editors, games and desktop publishing just didn’t make a market in the 80s. Combine that misstep with a proprietary distribution model which limited access to the base platform (OS) and the fate of Apple was sealed long before anyone realized it. It’s not hard to name other examples that have gone the same route. Consider NeXT (you think Steve would have learned this lesson), BeOS, the GO tablet PC, the Newton (three strikes Apple!), and the always schizophrenic SGI. The list is painfully long and the stories are strikingly similar.

 
Cultural Coasters

One of the assumptions in IR theory is that the up front cost and distribution of technology is neither easy nor cheap. To even get into the game, a company has to have the capital to build something and then the right partnerships to conquer the distribution channel in a massive way very early on. Clearly Microsoft’s early partnership with IBM in contrast to the Mac’s proprietary approach to the market is a good comparison. Another fantastic example would be America Online and the story of their near rise to a runaway leadership position in the online computing segment.

If we all remember, there were many online communities 10 years ago. Apple had eTown, there was Prodigy, and CompuServe attempted to play the game as well. The tipping point for America Online was directly related to how they out-distributed their competitors. Who of us has not received an America Online CD in the mail? A few tons of plastic and one hell of a postage budget later, AOL had mastered the distribution game and blew everyone else out of the water. This approach to distribution allowed AOL to trump their competition and start locking in their user base by moving on to the second phase of the IR game. And this they did by constantly delivering a steady stream of killer online apps in the early 90s. Email, IM, games, financial information, etc.. this way exactly what their majority user base wanted from an online platform. AOL accounts were growing through the roof. But then something went wrong. As I mentioned, AOL was almost a runaway leader.

All of a sudden the Internet came along. And with it, a zero cost mechanism to deliver the consumer a competing online platform. Mosaic popped up, and then Spyglass, and Netscape, and eventually IE. At the end of the day, AOL’s “platform” was really just an early version of a web browser. If you look carefully at AOL, it’s really just a simple text, graphic, and form layout engine that can dynamically handle whatever content AOL sends it. Sounds a lot like a web browser, does it? But zero cost distribution for the browser should not have been enough to topple AOL. Yes, it was inexpensive to install Netscape or IE, but the users should have been locked into AOL because of their applications, right? Well, clearly the answer was no.

The key here is that the internet not only brought a shift towards zero-cost access to alternative platforms, it inducted the market into a culture of infinite choice. This was not something that IR theory had contemplated. In the 80s, choice of software was limited to what you could buy at Egghead or copy from a friend. The “Casino Of Technology” was really a high-rollers-only room. In the 90s, the internet fundamentally shifted the concept of accessibility to choice and changed the way IR could work forever.

The near miss for AOL came from the fact that they never opened their platform up to allow niche content and application development. While they could attract users to their platform by providing the apps that a small majority of users wanted, they stumbled when competing platforms (Netscape, IE) started to provide access to all the other niche applications the rest of the users wanted. And the irony is that neither Netscape (the company) nor Microsoft actually built any of it themselves. Hordes of HTML and Web developers did it for them!

 
Increasing Returns Meets the Long Tail

To understand more about the power of this phenomenon, we need to look to a more contemporary market theory called “The Long Tail.” For those of you not yet exposed to the Long Tail, let me give you a brief primer. Originally it was an analysis of content buying habits in the offline world versus the digital world. Authored by Chris Anderson (chief editor of Wired), and first published in Wired itself, the Long Tail was the first widely acknowledge explanation of the impact of infinite choice in a digital marketplace. The initial analysis was based on the following approach. Start with the music, movie, and book markets. Then count the total number of SKUs in the biggest brick and mortar stores for each of these markets (Tower Records, Blockbuster, and Barnes and Noble to be specific). Now take the online equivalent (Rhapsody, NetFlix, and Amazon) and rank every item in terms of its popularity (number of units sold). Split each online company’s sales into two buckets: the highest ranked (most popular) item to the Nth ranked item (where N is the total number of SKUs in the brick and mortal equivalent), and take the N+1st ranked item to the lowest ranked (least popular) item in the online store. Compare.

The results were surprising and eerily consistent: online stores derive 50% over their sales from the
1^st to the Nth ranked items and the 50% from the N+1^st to least popular item. In other words if B&N has 130,000 books in any given store and Amazon has 2M, Amazon will derive half their sales from the top 130,000 ranked items on Amazon and half from the 130,001^st-2,000,000^th ranked item. In actuality the buying curve looks like this:

(thanks to Michael R. Nelson Design for the well designed graphic)

Wow! What a revelation. If consumers aren’t constrained by sellers’ distribution and warehouse costs, infinite choice really does play a significant role in consumption.

Well, it turns out that this curve is actually called a Zipf power distribution curve. Zipf distributions have two forms, logarithmic (which looks simpler but is harder to understand) and linear which is the more common form above that people are used to seeing in relation to the Long Tail. Zipf distributions have been used to explain anything that comes even close to the proverbial 80/20 rule. Geographic population distributions, social behavior, fads, complex systems, etc.. It appears to be some sort of intrinsic law of nature, just like the golden mean, that given a frictionless market of infinite choice you end up with consumption that looks like the Long Tail curve.

As a side note, since the Long Tail’s original analysis of the three primary consumer content sources (books, music, movies), its long shadow has grown to cover an explanation of almost everything under the Internet sun. The use of programming languages, blogging, internet usage statistics, politics, TV programming, web searches, and advertising revenues have all been analyzed using the Long Tail.

 
Economic String Theory

One of my few frustrations with the Long Tail is that we’re only a few years into fully understanding it. One of the blind spots right now is that there just isn’t much visibility into how it can help us predict winners and losers. It sort of says “Here is the data. If we give people choice, here are the buying habits you will see”. While I am sure Chris will take issue with this oversimplification, it is a fantastic starting point and argues heavily that there is good business in aggregation of niche content markets. But we still have few data points to show how one Long Tail company (Amazon, EBay, iTunes, BrightCove, YouTube, etc..) becomes the runaway leader in the space.

It dawned on me that the answer to my frustration was lying right there with our old friend, IR theory. Being a runaway leader (in software or in content) has always been about lock-in. IR says lock-in has two barriers, heavy distribution cost and customer lock-in through killer applications that drive a positive feedback loop of adoption. But the Long Tail says just the opposite: distribution is cheap (or free), and that killer applications (content) will only win you half the market. Winning half the market only leads to market equilibrium. How do we unify these two concepts?

Well, the conclusion I have come to is that both theories are right, but they failed to recognize the most critical factor in a runaway leader’s success: how the leader influences the tail of the market. Becoming a runaway leader is not really about how you win the first 50% of the market, but how you conquer the second 50%. First, you have to have a platform. Second, you have to drive adoption of the first 50% of the market by seeding the market with the most popular content (or applications). But in the end, the reason users stay locked in is directly proportional to their access to content (and applications) in the tail. As I’ll hopefully show, real runaway leaders do everything they can to drive the creation of content (applications) in the tail. With much respect for both Brian and Chris’ work, I’ll call this economic string theory of sorts “The Increasing Tail”.

 
Developers, Developers, Developers

Many of us have probably seen the now infamous Steve Balmer video master piece colloquially known as “The Monkey Dance”. For those unfamiliar with this small screen gem, Balmer goes bouncing around a stage (apparently looking like a monkey in the process) screaming “Developers, Developers, Developers” thunderously to a stadium size crowd. The irony here is that most people remember the monkey part and not the “Developers!” part. Consider for a second what Balmer was really saying, “Developers are the key”. And he’s not just talking about Microsoft employees – he’s talking about all developers on the Microsoft platform. Think a little about why Microsoft has been so successful and you’ll soon see that Balmer might as well have been bouncing around the stage screaming, “Secret to our success, secret to our success, secret to our success”.

You see, the problem with the first two phases of the IR game plan is that, at best, you can lock up the hearts and minds of a small majority of the user base. The Long Tail tells us this looks like about 50% of the market, but the niche stuff is important to. I’d argue that Microsoft understood this (perhaps in different words) a long time ago. What they needed for true platform lock-in was to drive the creation of applications for the tail of the market. And this could only be accomplished through the development of thousands of niche applications. This was clearly not a strategy they could fund on their own. So they turned to their users.

Whether you like MFC, VB, Visual Studio, Source Safe, WinForms or any of the other development tools Microsoft delivers to its user base, you simply cannot deny that Microsoft works harder than anyone else at making it easy to produce and deliver additional applications on their platform. Contrast them with Apple, who as a company did very little in terms of supporting their developers. The most popular development applications for the Macintosh were not even made by Apple. Having been a Mac developer in my past life, you quickly learned to depend on third party developer applications like MetroWorks, BBEdit, Butler SQL (I’m dating myself here), etc.. to get done what you needed. This turned out to be a major mistake for Apple. Not only had Apple lost their beachhead through a flawed philosophy on distribution and targeting of non-mass market applications, their lack of focus on development tools meant they were going loose the battle over the tail of the market as well.

To extend the Increasing Tail concept into the traditional content-oriented Long Tail world, let’s take the comparison between Zen//Rhapsody and iPod/iTunes. One can make all sorts of claims about the success of the iPod due to its wonderful form and design. But lets all remember that MP3 players had been around for a long time (e.g. the Rio). I think the critical success of the iPod to date, is related to the fact that Apple delivered the platform and the majority market content at the same time. The fact that if I bought into the iPod platform, I could also download Coldplay’s newest album was a winning combination. While P2P networks are the love of starving college students, they are cumbersome to use in reality and still the domain of a relatively small segment of users. But Zen and Rhapsody (and Napster) were quickly on the heels of Apple’s iTunes. In reality they all have pretty much the exact same 1.5M track media library available for download. So why hasn’t a lower priced Zen or a cheaper priced subscription model from Napster or Rhapsody beaten out the iPod. I’d argue that the reason for this had nothing to do with anything Apple did, and everything to do with an ex-MTV DJ named Adam Curry. Adam Curry, our generation’s “voice of the people”, invented (or was the poster child for) the concept of PodCasting. The explosive growth of PodCasting and the relative ease with which podcast content can be developed is a first pass at the fulfillment of the Increasing Tail in the MP3 market. Apple has a world wide group of users, constantly developing new tail content for their platform and increasing the tail. With every new podcast, new potential value is being added to the iPod platform. Because of this phenomenon, companies like Creative (who make the Zen MP3 players) have become an after thought in the market, left to disappoint many children on Christmas morning when they rip the gift wrap off the box and find their parents didn’t know the difference between a Zen and an iPod. It isn’t after all called ZenCasting. However, podcasting is really not a technology specific to Apple’s iPod platform. And Apple isn’t doing a whole lot to help the general population create podcasts and increase their tail. It’s my assertion here that slowly, as more tools become available to create audio and video content for any portable MP3 player, the iPod market share will slowly dwindle as customers don’t have any true lock-in. You heard it here first, start shorting that Apple stock.

All joking aside, the point is this; if you look at the Microsoft’s past success driving applications into the tail and you look at the future as it appears to be forming with content, a few things become clear. You can’t always count on up front costs to be a barrier, and you can’t rest your laurels on taking over only 50% of the market. Lock-in is a three stage process of platform distribution, winning the majority market, and then driving as hard as possible the development of niche markets on your platform. In other words, winning in the technology economy is still about one thing: developers, developers, developers. That’s the Increasing Tail philosophy.

 
Okay, I Thought You Were Going to Talk About the End of Microsoft?

Oh yes, lets go back to beginning. Thanks for indulging me there on quite a long digression. Well, the idea is pretty simple: for the first time in two decades, according to my Increasing Tail theory, Microsoft may be losing their grip on their most prized asset: developers. And if you believe anything I’ve said, this can only lead to a slow loosening of their grip on the operating system as well. The key here is that over the last ten years, the platform that developers “built on” has been migrating from the OS to something at a higher level. This has been a many stepped process that is only now coming to a head with the recent moves of Oracle, SAP and Salesforce.com. And with Oracles 16 Billion of investments in 2005, it’s coming to a head extremely quickly.

The problem for Microsoft is that developers have progressively been intermediated from working directly with the operating system. The first nail in that coffin was the browser. Even though Microsoft masterfully crushed Netscape’s 80% browser market share in less than 5 years, they actually pushed developers further away from their own operating system in the process. For many, the browser replaced the operating system as the primary application development platform. The applications and content delivered through the browser aren’t dependent on developers starting with Microsoft’s OS. Microsoft tried very hard to re-establish developer lock-in on top of the browser platform with DHTML, ASP, VB Script, and Active-X. Unfortunately they were outflanked by the massive early adoption of JavaScript and Java. While Active-X was arguably not a direct competitor to Java (around 1998 there must have been 1000 articles a day on that subject), it was clearly an attempt to preserve developer loyalty on the Microsoft platform. The problem for Microsoft was that JavaScript had incredibly rapid distribution when Netscape was still a contender and Microsoft needed to support it to keep in the game. Their battle with Java was similar in nature. In many ways, Sun used the browser as a Trojan horse to get JVMs distributed to millions of users. I don’t think the game plan was ever to have Java be used only for applet development. Sun clearly had their eyes on shifting developers away from the Microsoft OS and back towards Solaris, their own operating system. If they had to allow for some developers to use Java on HPUX, Linux, AIX, etc..that was a Pyrrhic victory they were willing to fight for.

To make things worse for Microsoft, the shift from client server to three tier architecture came like lighting. While Microsoft was still fighting client/server battles against Novell and others, all of a sudden the application server emerged as a significant part of the enterprise developer stack. And this is where Microsoft critically fumbled the ball on their own 1 yard line. They simply missed the boat on the application server. BizTalk? ASP.NET? Seriously! Not until Windows 2003 (5 years after the likes of BEA, iPlanet, WebSphere, SilverStream, etc..) have they even attempted to put a real application server into their enterprise server lineup. Perhaps this final goal has been part of the .NET game plan all along. In an interesting interview with Anders Hejlsberg, the chief architect of C#, one of the points he made was that C# (and all the .NET languages) finally adopted a component-oriented architecture. This might be the same component oriented approach that made Java such a ripe candidate for the component oriented development approach (e.g EJBs) of most application servers?

But even though Microsoft is making some strides with Windows 2003, .NET, SharePoint, etc.. they are quite far behind the other players. And to make things worse, in the last 3 years, both SAP and Oracle have stolen a major play from the Microsoft playbook and are hard at work winning the hearts and minds of developers. I have talked a lot about the evolution of the packaged application architecture from a bundle of applications to an application operating system. SAP and Oracle are fighting hard to become the next generation platform for development in most enterprise IT shops. Their stacks are as much about application delivery as they are about development.

Oracle calls it Fusion and SAP calls it NetWeaver. Whatever you call it, it’s a threat to Microsoft’s hegemony. What’s important about the way Oracle and SAP got to this point is that they have already executed the first two phases or IR strategy and now they are working on the Increasing Tail phase. Over the last 10 years, Oracle and SAP have delivered both the platform and the majority applications that have locked users into those platforms. CRM, ERP, SCM, HRMS, they are all killer applications that attracted users to their platforms. And for all the right IR reasons, they locked customers into their platforms in the process. Most shops have integrated their legacy systems into these platforms and spent millions training their users how to use Oracle Financials or SAP Manufacturing. Read most any report on ERP systems and they will tell you it’s just too damn hard (and expensive) to switch to another vendor.

After their success at the first two steps, they are clearly now moving onto the third: getting developers to write all the niche applications they themselves can not. All Oracle and SAP want to talk to you about today is composite applications. What’s a composite application? It’s an application built on top of the SAP or Oracle application operating system that uses some pre-built pieces from the vendor and some pieces you build yourself from scratch. SAP calls their pre-built pieces xApps and Oracle has yet to name them. But what’s bares a striking resemblance to Microsoft’s history is that both Oracle and SAP have spent a ton of time and money on the development tools that enable developers to build their own applications on top of their stacks. Oracle’s JDeveloper is a very advanced platform for developing on top of Oracle 11i, Forms and eventually the Fusion toolset. SAP’s latest release of Web Dyn Pro is an extremely sophisticated tool that lets you pretty much glue together all the services of NetWeaver in drag and drop fashion. And guess what? All of these development environments are primarily pushing Java as the core language of the future. While PeopleSoft PeopleCode and SAP ABAP will surely be supported in some form for the next 10 years, it is clear that Oracle and SAP are pushing developers as far down the Java path as possible. Oracle claims 4 million JDeveloper downloads. That’s 4 million! SAP’s Developer Network (SDN) claims something like 2 Million members now. And as both companies extend their development tools to the Eclipse platform as plug-ins, I imagine the number of users who would count themselves as potential Oracle or SAP composite application developers is growing at a staggering rate.

But Microsoft is no 40 watt light bulb. They are making some serious moves to stem the tide of this assault. Recently, there was an interesting article in C|Net where Gates opined about the Oracle/Siebel acquisition. Personally, as I mentioned in another blarticle, I was convinced that Microsoft would buy Siebel to take advantage of Nexus (Siebel’s very own composite application architecture). This however was not in Microsoft’s plans. But there are two interesting and directional things to note in the article that are actually pretty easy to miss if you’re just skimming. First, Microsoft is delaying the next release of Dynamics, their business application platform, until 2007/2008 while they “integrate them into one code base”. I’d hazard a guess that the reason for this delay is to move in the same direction as Oracle and SAP. Microsoft needs to provide a similar next generation application development platform as Fusion and NetWeaver, and I am guessing it is all going to be based on SharePoint. The second interesting comment was that Gates actually is quoted using the phrase “composite applications”. I suppose we can consider “composite applications” common terminology now that Gates is writing memos about it.

So why should Microsoft be so scared? Well, every second that goes buy, some developer somewhere is making a decision to build an application from the ground up using Microsoft tools or start on top of Fusion or NetWeaver. And every time the decision swings in the wrong direction, Microsoft is loosing its most valuable asset. To be fair, they have already put a lot of the pieces in place to fight this battle. For example, one of the most underrated aspects of the .NET language architecture is language neutrality. That means that any language can be relatively easily extended to compile into the Common Language Runtime and run in the .NET framework. And because there is a Common Language Runtime, half the code can be written in one language and the other half in another. This allows Microsoft to leverage its huge installed base of developers to fight this new battle for them without requiring developers to learn a whole new language (such as Java). If you already know VB, then great, just learn a few VB.NET extensions and you’re ready for the future. If you liked C++ or Java, the conversion to using C# is really a pretty a small step. And Microsoft is actively expanding this language footprint through the ECMA standards body and other initiatives. A little know fact of .NET is that currently over 40 languages are supported on top of the CLR. It looks like the Microsoft gang is preparing for a good ol’ street fight and they’ve called up all their friends.

 
Benioff, the Sleeping Giant?

However, this blarticle would not be complete without saying a little something about Salesforce.com. SF.com has become one of my favorite new companies. Not only because we use them at Newmerix, but because they completely understand both IR, the Long Tail, and my Increasing Tail philosophy. While it took Oracle and SAP 10 years to get to their platform, applications, and development features in place, Salesforce has done it all in less than 6 years. While everyone sloughed them off as building yet another CRM app, what they were really building was a platform with zero-cost distribution, ready for developers to extend it at will. To seed it, they delivered a killer application, CRM. Like Bezos picking books to sell over the internet, CRM was the right killer application to bring users to the platform. Once the platform was widely used, it created a marketplace for developers to create their own applications and extend the value of the Salesforce.com platform. While I think there is still room for Salesforce to tie up even more of the market with their own applications (I’d put money on the fact that they deliver an HR suite in 2006 – especially with Dave Duffield’s new venture nearly upon us), Salesforce just announced that they are extending their platform to any developer with AppExchange. Benioff, you crafty guy you! How did you sneak this one by us? I think he was even quoted as saying he wanted SF.com to be the Microsoft for middle market business applications. And honestly after writing this blarticle, I think he’s got a shot. He’s totally executed every phase correctly, all the time faking left and going right. All this without a developer ever having to touch an operating system!

 
To Conclude

First, let me thank you for taking this journey with me. Clearly, I love Brian Arthur’s Increasing Returns work and I am a huge fan of Chris Anderson’s Long Tail. If I can add even a little to the discussion with my Increasing Tail theory, then this tome will have been worth the typing. And in a brief moment of vanity, it appears to me I might be on to something. No matter how you slice it, the battle is right back to “Developers, Developers, Developers.” While I think Microsoft is on the back foot right now, they have always had an uncanny knack of bouncing back. Maybe Ray Ozzie will drive some new focus into the company. But unless they realize that their past should guide their future, we may actually see a slow decline of the Redmond Empire.

Comments

[herbertwilliams]

This is a really good read for me, Must admit that you are one of the best bloggers I ever saw.Thanks for posting this informative article.replica rolex day date

[supra vaider]

Love those! I enjoy following your posts on facebook and rss!

[professional wordpress themes]

The people who get on in this world are the people who get up and look for circumstances they want, and if they cannot find them, make them.so come on everyone.

[рыбалка]

Nice post admin.. Keep it right.
I added it to my bookmarks

[Generic Cipro]

It provides you valuable information and exciting idea..

[ugg sale]

I appreciate the time you positioned on this concentrate on or on this post. even in spite of the reality that have fascinating ideas.

[air yeezys]

The thing that oneself like to grasping, strive for, no regrets!

[supra shoes]

Our destiny offers not the cup of despair, but the chalice of opportunity. So let us seize it, not in fear, but in gladness. The people who get on in this world are the people who get up and look for circumstances they want, and if they cannot find them, make them.so come on everyone.

[air yeezy]

I'm for your article, I can not help but check for new updates to your web page. Your page is very interesting!

[Corbin]

I cannot say for sure if Ive ever seen so many pretty news at one place. Thanks for wonderful post!

[YVETTEEstes]

Set your own life more simple take the loans and everything you need.

[Volker]

It probably will need to read it again a few times before understanding all, but your blarticle is absolutely worth the investment:-)

Question: how does the OS/2 story looks under the light of your Increasing Tail idea?

[Jim Plamondon]

Neil,

I read with interest your blarticle of 8th February, “The Increasing Tail,” which seeks to combine the notions of “increasing returns” and “the long tail.” Overall, I agree whole-heartedly with its conclusions. Please allow me to comment on some of its points.

First, my bona fides. I started as a software developer in Silicon Valley, focusing primarily on the Mac platform. When Windows 3.0 caught fire, I realized that no matter how good my code, it would fail commercially if it targeted the wrong platform – so I became very interested in how platforms gained market dominance. Because Microsoft clearly understood how to make its platforms “win,” in 1992 I joined its Developer Relations Group (DRG). There, I designed and executed strategies that established Microsoft’s new platform technologies as de facto industry standards, and gave internal courses on the theory and practice of doing so. I retired from Microsoft in 2001, after a brief stint at Microsoft Research.

Here are some comments on Microsoft that are relevant to your discussion.

Microsoft’s understanding of the importance of independent software vendors (ISVs) to Windows’ success can be traced to three men. First among these is certainly Bill Gates, whose experience selling programming language tools and DOS provided him with invaluable lessons in the practical use of network effects. The other key people were the Myhrvold brothers, Cameron and Nathan, who both joined Microsoft in 1986. Among Nathan’s many degrees is a master’s in mathematical economics from Princeton, while Cameron’s degree is in rhetoric – “the art of persuasion” – from Berkeley.

Combining Bill’s experience, Nathan’s analysis, and Cameron’s execution as founding head of DRG, Microsoft created an ISV-focused “platform marketing” effort, which eventually became a 320-member team with an annual budget of $65 million. This team was focused primarily on gaining widespread ISV support for Microsoft’s new technologies before they reached the market, by involving leading ISVs in the technology design process, helping them implement Microsoft’s new technologies in their own products, and providing significant co-marketing benefits on commercial release. Until the late 1990’s, Microsoft executed this platform marketing model in a near-vacuum of competition, because other platform vendors simply did not understand how platform marketing worked. Indeed, the Department of Justice cited DRG’s “First Wave” model of platform marketing – which I co-invented – as giving Microsoft a nearly-insurmountable advantage. Now, THAT’s effective marketing!

Microsoft also created an important advantage for itself by having the Holy Trinity in-house: a platform for new features (Windows), tools to help ISVs implement those features (Visual Studio), and an application which could give any new platform feature instant critical mass (Office). The Trinity – leveraged aggressively by DRG – blessed Microsoft’s new platform features, and cursed its competitors’. Eventually, ISVs perceived that non-Microsoft features were doomed to failure. This perception became reality, ensuring the development of Windows-only niche applications, thereby locking the users of those Long Tail applications into Windows.

Contrast this with Apple, which actually cut back on its platform marketing staff in 1991, just as Windows 3.0 was gaining momentum. Michael Spindler, Apple’s CEO, derided independent Mac developers as “parasites” who were making a living at Apple’s expense. Apple started nickel-and-diming its ISVs around then, charging ISVs for each new Mac OS feature’s Software Developer’s Kit. This was criminally stupid, because Apple had the opportunity to learn from the Apple II experience – as Microsoft learned from the language tools and DOS experience – that ISVs are the key to success for any platform.

Microsoft has never forgotten that lesson – although, lately, it looks like it has. Its tools and MSDN subscriptions are ridiculously expensive, giving free open-source tools a compelling cost-advantage. Further, bad planning has significantly delayed the implementation of Microsoft’s next generation application development platform. After the Tech Wreck, Microsoft drank the offshore development Kool-Aid, placing a hold on domestic expansion (in effect) while planning to expand in India and China instead. Those plans have not worked out, and it’s clear from Microsoft’s recent King County land acquisitions and building approvals that it’s about to go on a huge building binge – and therefore, presumably, a hiring binge, too. What are all of these new employees going to work on? The “Live!” versions of Windows and Office, certainly, but I would not be at all surprised to see much of that growth be devoted to the development and marketing of Microsoft’s next generation application development platform. I expect that once Microsoft has this new platform in place, it will make developing, deploying, and maintaining powerful Web 2.0 applications a simple pleasure, allowing Microsoft to once again give ISVs the opportunity to do more, faster, and easier, than on any other platform.

Neil, in your blarticle you wrote that “a little know fact of .NET is that currently over 40 languages are supported on top of the CLR.” Therein lies a tale. I had to fight internal groups right up to Microsoft’s #3 guy, Paul Maritz (Vice President of the Platforms Strategy and Developer Group), to get an independent budget to make this happen. With that budget – and Paul’s air cover – I designed and executed “Project 7” (1998-2000), which helped the developers of the leading academic and commercial programming languages develop versions that targeted the CLR and integrated into Visual Studio. Project 7 was opposed by a surprisingly influential group of Microsofties who – stunned by the rise of Java – thought that Microsoft ought to suppress all new programming languages, rather than support them as Project 7 was doing. It is a testament to Microsoft’s collective wisdom (at the time) that this short-sighted reasoning was itself suppressed.

Project 7 delivered on all of its major objectives: (1) it forestalled Sun from suing Microsoft with claims that the CLR was just a rip-off of the JVM, by providing Microsoft with third-party expert witnesses – the participating language developers – who could credibly testify to the contrary; (2) it made Java “just another language” as far as Microsoft’s technical infrastructure was concerned; and (3) it delivered a suite of languages that proved Microsoft’s claim that the CLR (and Visual Studio) really was language-neutral (well, so long as your language didn’t need to crawl the stack, anyway, which was just too big a security risk to allow).

Since then, I left Microsoft and made what appear to be some breakthroughs in the display and control of musical information, including a new musical instrument – the Thummer™ – which I am commercializing through a new company, Thumtronics Ltd (www.thummer.com).

While on the one hand launching a new musical instrument might seem risky, I would argue that the strategic issues raised in your blarticle make it likely to achieve lasting success. As Clayton Christensen wrote in "The Innovator’s Solution" regarding Sony's early management team, “they looked for ways that they might help a larger population of less-skilled and less-affluent people to accomplish, more conveniently and at less expense, the jobs that they were already trying to get done through awkward, unsatisfactory means.” The Thummer does exactly this, and should reach the market in time for Christmas 2006.

Thanks for your excellent blarticle.

Jim Plamondon
CEO, Thumtronics Ltd
The New Shape of Music™
www.thummer.com

[Rishi Khaitan]

I suppose those in the Web 2.0 meme would call "how the leader influences the tail of the market" edge-feeding. Content is increasingly being created at the "edge", so as a platform, you want to encourage publishers to create content around your service. i.e. Bloggers that integrate Flickr photos or YouTube videos into their posts.

Thanks for taking the time to compose this post. I like your style of analysis.

[Cote']

YUH!

[Peter Cranstone]

Simply brilliant. I wonder how many people will not only take the time to read it, but also the time to comprehend it.

It's all about the tools and developers who use those tools.

Great job! and thanks for writing this.