I'll start off with a caveat, I am an open-source advocate, a long standing committee member and current president of Linux Users of Victoria (http://luv.asn.au). My overall approach is that computer programs are scientific problems, and therefore there is an inherent tendency towards a better solution when the source code can be freely analysed, distributed and developed; release early, release often and with many eyes all bugs are shallow (to paraphrase Eric Raymond's famous essay 'The Cathedral and the Bazaar'). I should also note Eric Raymond's remarks from the leaked Halloween Documents, where Microsoft admitted that GNU/Linux was a serious threat to their corporate model and outlined their tactics against the operating system. Raymond claims that "peaceful co-existence" between the free and open-source software and proprietary models is, in the long-run, impossible.

The question explored here however is not an scientific one, it is a business one, although the latter does follow from the former. Is it possible for a business to gain from a switch to open-source software solutions? For the purposes of this discussion I'm doing a comparison between MS-Windows and Linux.

Firstly, there is the accounting problem of the software. There is absolutely Byzantine methods used to determine the cost per unit of proprietary software for a standard desktop system with little more than the operating system plus the office application suite. By itself it the professional edition of MS-Windows 7 is AU$471.99 (figures from Microsoft), and MS-Office 2010 Professional is AU$871.99. So that sounds like $1344 per desk. Yes, there are cheaper, stripped down, versions of the application, but if you're serious about desktop work you probably want a database application of sorts. Yes, there are also upgrade options instead of buying a new system. But given that the lifespan of the software is roughly equivalent of the lifespan of a desktop system in an office environment (circa 3 years) this can be ignored as well.

These costs are typically hidden, a pre-sale costs incorporated in the consumer purchase price of hardware. OEM (original equipment manufacturer) is where most pre-sales come from, and of course the cost is bundled in the consumer product, albeit at the equivalent of a volume discount. For example, with a little fishing around on the Dell website, we could ascertain that one could upgrade from MS Vista Home basic (nominally new cost of c$85) to MS-Windows 7 Professional for an additional $326.70. So you're paying $410 for the operating system, a saving of about $70 (c15%). So perhaps a more normal cost per seat is $1370, or even as low as $1240 if such discounts are applied to MS-Office.

This is of course not the only proprietary software that's installed. To be sure, the free Microsoft Security Essentials for anti-virus and spy-ware is fairly good for home use, but even then it's worth spending a handful of pennies and getting something like ESET NOD32 Antivirus (c$60 per seat for the period); that's relatively small crumbs. More importantly is the additional software required. There's financial software (MYOB, Quickbooks, Attaché) which is an absolute minimum $250 per seat; Attaché, for example, requires that you have a spare MS-Windows system to act a server. The creative individuals look towards Adobe's InDesign at a similar price. If anyone needs developer tools, well, a decent (read "Premium") edition of Visual Studio comes in at a cool $9,500 AUD.

Now let's talk about servers. Any business of reasonable size will need a file server, a mail server, a database server. They may even maintain their own webserver; which will, of course, need to be in its DMZ - you really don't want your business SQL database on the same machine as your webserver, really. Just to make even more fun, servers have their own licensing scheme based on a server license and client access licenses (CALs, a "damaged good" in economic terms) for users or devices. Your MS-Windows 2008 Standard Server will put you back a neat $1000, the CALs are $800 per 20 CALs ($40 per user and device), and your web server cost another $450. And for MS-SQL, assuming you can't get by with the free Express edition? Try $13,000 for R2 2008 Enterprise edition 25 CALs. So let's be generous and assume no MS-SQL server (except for Express) and no Visual Studio. Just a plain-server, with around fifty CALs bringing the cost to another $40 per seat. Oh, and one more on webservers.. Can anyone explain why websites with MS-Windows webserver applications (IIS) are cracked or defaced much more often than Linux ones (Apache, Nginx, Lighttpd), despite the fact that IIS runs less than a third the number of sites?

No matter how I look it at, it's about $1500 per seat for proprietory software, including one non-MS application, unless you have a friend that has a bunch of DVDs and codes that fell off the back of a truck. That should last at three years, so a relatively modest sum, you would think. But how about support? In reality, your average office staff member is not quite a sysadmin, and nor should they be. A very interesting survey was conducted by the Robert Frances Group for server admins way back in 2003. Despite it's age the principles still hold. Linux sysadmins are slightly more expensive, by about 4%. However, due to various management features in the respective operating systems, the Linux admins handled more than four times as many servers. Translated to desktop units, one will need at least one MS sysadmin per thirty seats; you would receive equivalent service from one Linux sysadmin per fifty seats, a figure which interestingly correlates in proportion with server downtime (hours per annum). Using our per seat basis, that's an additional $2350 per annum per seat for the requisite one and a half MS admins, or $1450 per year per seat for the Linux sysadmin. To put it in a nutshell from a support perspective, a Linux system just doesn't break as often, and when it does break it's a lot easier to fix.

Now, what about functionality? At these prices it would look at bit like a no brainer. But is a Linux operating system and applications at least as functional as a MS Windows? Is it as user-friendly? Or do you real need to be a propeller-head and operate with a text-console with green text on a black screen to really use it? Well, the first test, being the desktop environment suggests a high degree of equality in user focus in both operating systems. Of course a significant difference is that MS-Windows changes the desktop environment with each major release (consider the changes between Windows 2000 to XP to Vista to 7) whereas with Linux it changes between the numerous different managers (GNOME, KDE, Compiz, Xfce etc). The practical upshot being however is that with the latter the environment is subject to a high degree of choice - including styles which mimic existing environment familiarity (see below for examples that have the appearance of Windows XP, for example). Again, studies indicate that MS-Windows XP and GNOME (one of the Linux desktop managers) are roughly equivalent in terms of usability,

Moving from the operating system to applications a similar lesson applies. For the equivalent of Word, Excel, PowerPoint etc the most well-known alternative suite is OpenOffice, which combines a word processor, spreadsheet, presentation, graphics and database applications into a single suite and in a style that most administrative staff are familiar, unlike the cryptic new ribbon format of MS-Office which apparently causes more pain than pleasure. OpenOffice supports the ISO/IEC OpenDocument Format for data compatibility, including Microsoft Office formats, which makes an international and approved standard. There are some minor differences in the parsing of presentations, so if your Powerpoint Presentation has a whistles and bells it will almost certainly need some reformatting. But overwhelmingly it is very similar in style and functionality. A significant advantage is that it's completely cross-platform and excellent PDF export and even editing tools.

On a related note of ubiquitous office software it absolutely astounds me that people still use Outlook as their email client and Internet Explorer as their browser. IE is typically far behind web standards for basic web functions; it only became xhtml compliant with v9, has only partial support for level 2 stylesheets and image parsing. Both IE and Outlook are notorious for a vast array of security vulnerabilities, spy-ware, adware and viruses which are simply not a problem with other open-source applications, coming down to serious flaws in the security architecture. There are numerous free and open source alternatives for browsers (e.g., Firefox, Chrome, Mozilla, Opera) and email clients/personal information managers (e.g., Evolution, Kontact, Thunderbird and ) which are feature-rich, mature, tested and in active development.

For desktop publishers and photoeditors Scribus and GIMP make excellent alternatives and indeed improvements. For example GIMP provides capture, complex animation and video treatment utilities which Photoshop cannot. For finances and resource management, certainly GnuCash and GnuEnterprises are examples of alternatives to MYOB, Attache etc, which are feature-rich, mature and with a large and active support and development team.

Despite all this, there are some incompatibilities. Overwhelmingly, Microsoft office formats can be largely read by Linux applications (the reverse certainly doesn't apply). As a result I *do* recommend at least *one* complete MS system (operating system, applications, other closed-source software) per office to deal with other businesses that decide to send data in a common, but closed-source, format. This is not however a lasting solution; closed formats have an inherent tendency towards incompatibilities.

There is probably an assumption that staff will need training for such a change. But it must be admitted that recent changes in either MS-Windows or MS-Office these are usually significant enough to also require re-training. The changes from MS-Windows XP to MS-Windows Vista/7 are notable enough, as are the changes between MS-Office 2003 and MS-Office 2007/2010. More so, they are non-optional. It is a case of either accept the interface or live with an older technology. This is not the case in open-source operating systems. With familiar styling in the operating system (as recommended), training time can be further minimised, perhaps even as little as 10 hours per seat i.e., about the same as any other operating system/application suite upgrade. In other words, there really is no difference on a per-seat basis.

For similar reasons, transitions are best applied when upgrades are deemed necessary. Many advocates of open-source software argue, and quite correctly, that a migration from a closed-source to and open-source solution has an initial significant cost, followed by a decline and then lower long-term operating costs. The argument does address the long-term benefit of an open-source solution; but in the interests of mitigating the initial hurdle, adoption at upgrade points is superior.

At the beginning of this discussion, I emphasised that the provision of computer operating systems and applications is inherently better when it is open-source rather than closed-source from an engineering and scientific perspective. Problems can be fixed quicker, features can be added freely. A different angle is taken here by way of conclusion; the provision of open-source operating systems and applications makes better business sense. There is a common misconception that free and open-source is anti-commercial. This is incorrect; it is against vendor lock-in, an extremely expensive and dangerous form of "partnership" which entraps a company into a system or file format that they become increasingly dependent on ("addicted to" would perhaps be a better phrase).

Adoption of an open-source alternative should follow the path of less resistance. There will be, of course, resistance; this is inevitable with change. Perhaps one of the greatest difficulties is that from past experience is the greatest level of resistance to adopting open-source comes not from technicians or even administrative staff, but from higher levels of management. Part of this is from a relentless campaign of "fear, uncertainty and doubt" (aka FUD). One of the worst examples was Microsoft claiming that Linux had a higher total cost of ownership than Linux; the comparison was actually between a Microsoft products on a desktop-level virtual PC and Linux on an expensive IBM mainframe. This, combined with the extraordinary dead-weight of sheer habit, a perception of opportunity costs ("my time is too important to learn this!") means that the most influential people in a company are often the last to pick up on the technical benefits and business improvements by adopting an open-source solution.

To them, I present a bottom line. Total cost of ownership on the desktop, not including hardware, not including lost productivity due to outage times:

MS-Windows $2950, Linux, $1550. Per annum, per seat.

Halloween Documents
http://www.catb.org/~esr/halloween/

Practical Comparison
http://en.wikipedia.org/wiki/Comparison_of_Windows_and_Linux
http://en.wikipedia.org/wiki/Comparison_of_web_browsers (in particular look at unpatched vulnerabilities)

Linux vs MS Labour Costs etc
http://www.zdnet.com/news/linux-tco-edge-lower-labor-costs/297046
http://www.netc.org/openoptions/images/gif/migration.gif

Interface Usability
http://www.computerworld.com/s/article/83708/Study_Linux_nears_Windows_X...

Server OS Reliability
http://www.iaps.com/2008-server-reliability-survey.html

Functional Comparison between OpenOffice and MS-Office
http://www.techsoup.org/learningcenter/software/page4765.cfm
http://www.pcmag.com/article2/0,2817,2332502,00.asp
http://wiki.services.openoffice.org/wiki/Documentation/OOoAuthors_User_M...

Linux styled to a similar manner to XP
http://mediakey.dk/~cc/download-windows-xp-skins/

Scribus, GIMP vs InDesign and Photoshop (in Spanish)
http://www.scribd.com/doc/11449813/Scribus-Gimp-vs-Photoshop-InDisign

Emulation
http://www.codeweavers.com/products/crossover/stdvspro/
http://www.winehq.org

Fear, Uncertainity and Doubt
http://en.wikipedia.org/wiki/Fear,_uncertainty_and_doubt
http://attrition.org/~jericho/works/security/fud.html
http://www.silicon.com/technology/hardware/2004/08/25/linux-10-times-mor...
http://www.theinquirer.net/inquirer/news/1027930/microsoft-slammed-misle...