This is another datapoint that generational attitudes towards the value of software have shifted.
In the early days of software, it was an afterthought: what made it possible for IBM and others to sell the more valuable hardware asset.
Microsoft (founded 1975) perceived the economic opportunity attached to volume distribution of user facing machines, and helped usher in an age where the value shifted from the hardware to the software.
Google (1998), by contrast, derived effectively zero revenue from software directly. Instead, they leveraged open source and commodity hardware to control their costs and were able to scale more effectively than existing search alternatives. While they didn't sell software, however, they continued to perceive it as a competitive advantage, and thus a minority of their overall development was shared via public repositories.
For Facebook (2004), Twitter (2006) and GitHub (2008), like Google, software is the means to an end rather than the end itself. Unlike Google, however, most do not behave as if code itself is a competitive advantage. You can argue about what their actual product is - a critical mass of users, the data they generate or both - but you can't build the case that it is primarily the software.
And if the software doesn't represent a competitive advantage, the follow on benefits of releasing it as open source software - whether it's general goodwill, amortizing development costs, greater efficiencies in hiring and recruitment, etc - are believed to outweigh the costs.
Hence Cassandra, FlockDB, Hip Hop, Hive, Jekyll, Resque, Storm, Thrift and so on.
None of these entities open source the entirety of their infrastructure, but it is increasingly common to see open source as the default rather than the exception.
Software has and will continue to have value. But in an increasing number of cases it will not represent a competitive advantage, and therefore assuming the burden of maintaining it internally will become less attractive.
> Instead, they leveraged open source and commodity hardware to control their costs and were able to scale more effectively than existing search alternatives.
I would propose a caveat that Google's initial strength wasn't merely that it scaled better but that it used a search algorithm that produced better results. As Tom Preston-Werner argues, Google jealously guards their PageRank algorithm because it "represents core business value".
I believe you are missing the point of the argument. Substitute "PageRank algorithm" with "whatever method(s) they are currently using to rank pages" as it is always changing in response to people who are gaming the system.
Very insightful. Reminds me of a talk I saw Tim O'Reilly give at OSCON back in the day, basically about this same concept. He described it as "moving up the stack", with each successive phase commoditizing the thing beneath it (hardware at the bottom, then software, then services). Here's a link to his writeup of the concept (although it's kind of dry -- I remember the presentation itself having more pictures and being more engaging):
http://tim.oreilly.com/articles/paradigmshift_0504.html
I wonder what's going to sit above "services" in the next phase? Maybe the services will be commoditized because there will be so many of them that do similar things and they all talk to each other via RESTful interfaces... so then the thing that's valuable is whatever sits above that. Identity systems? Yahoo Pipes? I dunno... predicting the future is hard, let's go shopping!
"I wonder what's going to sit above "services" in the next phase?"
IMO, and I suspect Tim's as well given his "data is the Intel inside" comments, data will be the next major revenue stream. Consider that open source software generally excels at distribution, but is poor at customer conversion.
Open source vendors would be wise to gradually de-emphasize customer conversion mechanisms such as open core or hybrid source in favor of maximizing usage and thus the size of the dataset they may generate. This data can then be analyzed and sold to customers as analytics or to third parties interested in broader trend data.
See Spiceworks, Sonatype Insights etc as indications of where this market will go.
Brilliant! This makes a lot of sense. Interestingly, though, I see data as working in tandem with services (as the services are the ones that would be able to make use of the data) as opposed to the data commoditizing the services. I guess that's okay -- this theory is just a model, not immutable laws of physics :)
EDIT: Okay, I just looked at those two companies, and I totally get it -- they commoditize the services by making the service free (which they can do because they're making money off the data they gather from the free users). Consider my mind blown.
In college I did some App Engine development. At a previous job, we used open source Google Java libraries like Guice and Guava. I recently started at Google and walked in the door largely proficient in using their stack. In my opinion, they did everyone a favour.
They've done good by other companies by providing them with solid libraries. They did right by me -- knowledge of their libraries presumably made me a stronger candidate. Finally, they did right by themselves by reducing the time it took me to get up to speed.
In the early days of software, it was an afterthought: what made it possible for IBM and others to sell the more valuable hardware asset.
Microsoft (founded 1975) perceived the economic opportunity attached to volume distribution of user facing machines, and helped usher in an age where the value shifted from the hardware to the software.
Google (1998), by contrast, derived effectively zero revenue from software directly. Instead, they leveraged open source and commodity hardware to control their costs and were able to scale more effectively than existing search alternatives. While they didn't sell software, however, they continued to perceive it as a competitive advantage, and thus a minority of their overall development was shared via public repositories.
For Facebook (2004), Twitter (2006) and GitHub (2008), like Google, software is the means to an end rather than the end itself. Unlike Google, however, most do not behave as if code itself is a competitive advantage. You can argue about what their actual product is - a critical mass of users, the data they generate or both - but you can't build the case that it is primarily the software.
And if the software doesn't represent a competitive advantage, the follow on benefits of releasing it as open source software - whether it's general goodwill, amortizing development costs, greater efficiencies in hiring and recruitment, etc - are believed to outweigh the costs.
Hence Cassandra, FlockDB, Hip Hop, Hive, Jekyll, Resque, Storm, Thrift and so on.
None of these entities open source the entirety of their infrastructure, but it is increasingly common to see open source as the default rather than the exception.
Software has and will continue to have value. But in an increasing number of cases it will not represent a competitive advantage, and therefore assuming the burden of maintaining it internally will become less attractive.