Why all the fuss about open source software? And what does "open" actually mean, anyway?
I was fascinated to read about Software Freedom Day last month, where open source enthusiasts took to the streets around the world to celebrate open software. We've had freedom marches in favour of land rights, troop withdrawal and industrial relations laws, but I never realized proprietary software was in the same category as country invasions and reductions in workers rights.
This prompted the question: What's all the fuss about open source (OS) software? Which immediately prompted a more important question: What does "open" actually mean? Whenever my department heads demand new open systems, they always mean Windows on Intel. This formed my initial theory . . .
Open = PopularOpen is whatever everyone is buying. It's Windows at the moment, but open in the 1970s must have been MVS on an IBM S/370, and PC DOS was therefore the open software of the 80s.
Since FreeDOS has just released PC DOS as an open source product, perhaps that should become our standard. Not only is it open, it would also save money by needing servers with only 2MB rather than 2GB of memory.
Supporting this definition is the Firefox browser, which is gaining market share over Internet Explorer, so it's more popular and is open. Then I discovered it only has 11 percent share versus 83 percent for IE, so it's less popular, and still open. I looked for a new definition.
Open = CheapThis is a very popular view unless one is trying to run a business rather than a university project. My first hint that this theory is flawed is that while Linux is freely downloadable, most people pay RedHat or SuSE to get the code.
Although I discovered an open slather of open source databases, groupware, applications and systems management, I also discovered how much time it took me to locate consistent versions, documentation and implementation scripts. To effectively implement cheap open source software requires experienced, skilled personnel, and the type of company looking to cut costs in IT spending by going OS has usually already cut costs in skilled staff. With this definition now modified to open = cheap if you don't count all the costs, I moved on.
Open = Non-ProprietaryIt's obvious that Open must mean not owned by anyone. Except all the companies with OS applications seem to be funded or owned by someone, and have a business plan to make money. Some of the most publicly active companies in OS include Sun, Yahoo, IBM and Microsoft - none of which are top of mind when thinking about not-for-profit organizations.
Sun has announced it will publish the first open source Java this month. This month? I thought Java was the essence of open, and now I find out the code is all owned by Sun and unpublished? Yahoo, that doyen of open (and free) software, stated it would never license all its code due to integrated links to secret functions that provide better performance.
IBM, having adopted Linux as a little brother, is now promoting open development in storage, blade, grid and application servers. Does that mean the old adage "You won't get fired for buying IBM", which reversed in the 1990s, now applies again?
Microsoft's OS Software Lab has offered to help Mozilla port Firefox to Vista. Some liken this to the spider inviting in the fly for a comfy seat on the web, but I prefer to see this as the corporation pitching in for the little guy.
Open = LinuxNow we're talking. This is undeniable. An operating system created by one guy, released to the world, added to, updated and fixed by anyone who wants to and available for free. But businesses don't run on an operating system any more than anyone drives to work on a car engine. Applications must be part of the equation, so while this might be an answer, it isn't the full answer. Furthermore, the benefit of Linux being updatable by anyone is also its weakness in that many variations abound. There are already five flavours of Linux just for mobile phones, all battling for supremacy. This leads to my next candidate.
Open = Fragmented and No controlThe previous open system was Unix, which split into System V and BSD, developed standards like OSF, but still evolved into different Unices (my plural) of AIX, Solaris, HP-UX and Tru64. The same has happened with Linux and potentially with every other OS software.
One of the issues for Sun and Yahoo publishing their code is the danger of their code getting fragmented, or forked. If a business strikes a serious bug that occurs in a less popular variant, so unlikely to fixed quickly, it's not only the software that'll be forked. However there are concerted efforts to bring some order to open, so perhaps . . .
Open = Consortium Approved
Anarchy only works with a small number of participants (ideally one), so putting a defining fence around the open spaces is a necessary evil. There is a mobile Linux consortium to specify a single MoPho Linux, and accepted definitions like Enterprise Web 2.0 all trying to build acceptable fences. The most comprehensive is the Open Source Initiative which has established licences such as GPL and CDDL under which OS code can be published. Currently there are 57 such licences, so they can't be accused of not working hard, but this large number leads me to my next theory.
Open = DIYOpen means separate, independently developed, differently licensed software components for each aspect of my business. All I have to do is identify each one, locate them, install and customize them, debug problems and bingo - I have a complete system realizing cost savings to my business and demonstrating my commitment to the open source community (in order of importance).
I recall the time I put in my DIY home theatre system, purchasing the components separately, making non-standard connectors to combine them, and, with some difficulty due to their different shapes, co-locating them into a wall unit. After two weeks of fiddling and replacing components, I paid a technician to try to repair my install. I've now got the only home theatre with Dolby 4.5:1 sound, since the fifth speaker only works sporadically.
Peter Levine of Xensoft, when asked about the issue of unfinished code in Xen virtualization software being incorporated into Linux, said "I don't know what Novell . . . or RedHat does with our stuff" and "Yes, depending on the environment, it's safe". Google asserts that Tesseract is the best open source OCR program available, though admitting it struggles with colour and grey-scale and lacks the accuracy of the commercial packages.
This doesn't give me a warm and fuzzy feeling inside. I can see why there are so many OS applications performing systems monitoring and management - there's such a huge need for it.
Open = a GameHaving unsuccessfully tried seven definitions for open, I'm wondering if the whole OS movement is just the latest technology game whose popularity will pass. In the early 1990s, we played the distributed servers game which we followed with the e-commerce game, played eagerly until its board crashed to the floor in 2000.
Looking at OS products such as Nexaweb studio, Zimbra and Scalix collaboration, Open-Xchange groupware, Qlusters system management, Zenoss monitoring, Zmanda backup, Xinit storage software, Xen and Mozilla Firefox, the name of the OS game suddenly becomes clear. The important factor in open source is to have a name spelled with a Q, X or Z - which are all essential letters for getting a high score in Scrabble.
Open = a ScrabbleIt seems the most complete definition of all.
Bruce Kirkham is a veteran IT satirist and professional speaker specializing in leading edge technologies and scepticism, who views the IT industry not so much as "dot com" as "dot comedy"
Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.