Open Source with few Open Benefits

Several months ago I was attending a conference on open source when a senior executive from IBM who was participating on a panel declared:

"Approximately 0% or our customers that use Linux ever modify any source code."

This statement is probably a bit of a commentary on IBM's customer base (they are IBM customers because IBM tends to serve companies that want IBM to do everything for them), but I think it is also a commentary on the current status of commercial open source software products. For as much as open source has shocked the software industry, the software market is essentially the same as before open source, but with a slight change in the players on the sell side of the equation. The open source companies are using open source as a signal to the market that their software is low cost (no pricing leverage through proprietary rights) and high quality (anyone can inspect the code), but they are not signaling to the market that open source is a way for the customer to create the unique innovations that matter to the customer. Generally speaking, commercial open source software is still "one size fits all" and de facto proprietary. Let's take the example of a couple of poster children in the open source space - Red Hat, JBoss (now Red Hat), and MySQL.

Red Hat is the great citizen of open source. The amount of R&D spending that Red Hat pours into Linux and other related projects to which it exercises no rights of ownership is extraordinary. However, if you are a Red Hat customer and you want commercial support for some small tweak to Red Hat Enterprise Linux, forget about it (unless of course you are one of a handful of $million plus annual customers, then the policy is similar to gays in the military - "don't ask, don't tell").

Then there is JBoss, "the professional open source company." When I first met Marc, he took great pride in explaining to me that the only contributors to JBoss products were the JBoss employees, which put JBoss a commercial cut-above most open source companies who rely on developers not on the payroll for large portions of the code. I never really understood why JBoss was open source. If you are not building a community of developers to create leverage in R&D, why not just give away the binaries and sell the product for a low price? Also, the tagline "professional open source" implies by contrast that every other company in open source is an amateur hack (which Marc supported with his frequent attacks on other open source projects and companies, especially Red Hat).

As for MySQL, I was on a panel with Marten Mickos when Robert Leftkowitz stood up in the audience with a red clown nose on his face (which he also sported during a contract negotiation with Red Hat while working for Merrill Lynch) and asked about commercial support for "bozos that want to modify the code." Marten responded, cleverly, "we have a special bozo contract for those customers." Again, if you want to be a bozo and modify the code, we will treat you like a bozo in our commercial relationship. Perhaps Marten was just being clever for the panel, but generally speaking open source software tracks very closely to proprietary software as it relates to business model assumptions.

I mean no disrespect to the commercial success of any of these companies, but they have really changed very little about the software industry (except perhaps the pricing expectations of the customers) through their success. There is so much more work that needs to be done to make software truly "soft" such that it can literally be molded to the requirements of each customer. Current software business models are still in a phase of maturity that is reminiscent of Henry Ford's commentary about customers that wanted colorful autos, "You can have any color you like so long as it is black." Open source is a great step towards "mass customization" for software, but there is still much more work to be done before the industry can truly claim that a customer centric transformation has been achieved.

Small Percentages make a Big Difference

I had the good fortune of having a very entertaining conversation about the economics of the software industry with a bright senior software executive this week. We were talking about the economics of a subscription model as compared to the economics of a perpetual license model, and she had all of the fundamentals down cold:

- R&D is cut in half (from 16% to 8% of revenue, on average) when you can bring all customers along with every bit of maintenance so there are no legacy platform or release issues.

- Renewal rates are higher when customers see the software vendor continuously delivering innovation that makes their application more valuable

- It is difficult to transition to a subscription model when you are so dependent on perpetual licenses for quarterly revenue

She explained to me that her company historically had a very high maintenance renewal rate of 90% compared to the industry average of 85%. I thought about those numbers and how they contrast with Red Hat's most recently reported renewal rates of nearly 100% for the top 100 customers, which I wrote about in a previous blog post. I decided that I wanted to do a little test of the power of subscription renewals for topline performance of a software company.

Let's take a hypothetical case of a company that books $100 in new subscription bookings each year, with some percentage of the prior years new bookings renewing each subsequent year. In year 2 and year 3, for example, the company bookings would be:

- Year 2 Bookings = $100 + (Year 1 Bookings)*X%

- Year 3 Bookings = $100 + (Year 2 Bookings)*X%

and so on for each subsequent year. Over a ten year period, how much more annual bookings would a company have if they are able to improve their subscription renewal rate from the average of 85% to 90%? to 95%? to 98%?

Here are the numbers and the percentage improvements for each case relative to the industry average of 85%:

- 85% renewal yields $555 in Year 10 Bookings, base case

- 90% renewal yields $686 in Year 10 Bookings, 24% improvement

- 95% renewal yields $862 in Year 10 Bookings, 55% improvement

- 98% renewal yields $996 in Year 10 Bookings, 79% improvement

I was shocked to discover that a mere improvement of 13 points on renewals yields a topline improvement of nearly 80% over 10 years. Then again, I think it was Albert Einstein who claimed "The most powerful force in the universe is compound interest." Why should I be suprised when Einstein is proven correct yet again?

Sun's Gift to Linux

According to a recent article on ZDNet, Sun plans to remove the distribution restrictions from its Java Runtime Environment. Now developers and solution providers that use Java programs can provide the golden standard of Sun's JRE to their customers without the customer burden of annoying licensing and download routines. Without these cumbersome restrictions, Linux developers that were formerly cautious on Java might find themselves making a different decision in light of Sun's JRE beneficence.

As with the many gifts that Sun has provided to Linux in the past (OpenOffice, sponsoring GNOME development, buying Cobalt for $2B, making SPARC expensive and slow so customers adopt Linux instead), I have not quite figured out how this latest gift benefits Sun. I suppose that fewer restrictions on Java resulting in more Java programs in the market makes for a bigger market overall for Sun's wares, but I don't know if Java growth still correlates to Sun growth in the current environment.

I shared this sentiment with a friend and colleague of mine, Nathan Thomas, and he asked me if I had ever heard of the Underpants Gnomes from the Comedy Central show Southpark. I had not, so Nathan pointed me to wikipedia to fill in the details. It seems that the Gnomes steal underpants as part of a three step business plan:

1. Collect Underpants

2. ???

3. Profit!

Rather than looking a gift horse in the mouth by questioning the motives of the giver, let's just thank Sun for helping advance the cause of Linux one more time and hope the market trend lines associated with Linux get yet another boost from Sun's actions. Perhaps the questions regarding motivation and business models that perplex all of us that watch Sun from a distance are more clear to those inside the Sun organization.