Cord Blomquist did a good post over at TLF about the USTR/open source software issue. It set off a lively debate in the comments that, I think, reflected a common misconception about what free software is and why it’s valuable.
Before digging into that, I should be clear that I don’t generally think it’s a good idea for government software procurement decisions to be dictated from the top down. I’ve had first-hand experience with working at the bottom of large IT hierarchies, and so I know it’s incredibly annoying to have distant bureaucrats constrain the tools you’re allows to use. I wouldn’t endorse a proposal for the federal CTO to mandate that all federal agencies use free software.
But a some people, including a significant number of libertarians, make the stronger claim that government agencies at all levels have an obligation to be neutral between proprietary and free software. This is generally phrased as saying that the government shouldn’t “distort the market” or “pick winners and losers.” On this view, the government has an obligation to make its decision based on the characteristics of the software, without discriminating based on licensing or business models.
To see why this is wrong, consider the following hypothetical. Suppose federal agencies had a long-standing practice of obtaining their care fleets by renting them from companies like Enterprise and Hertz (or, more likely, government contractors that charged ten times as much as Enterprise and Hertz would). Now suppose the GSA did a study and found that the government would save hundreds of millions of dollars by purchasing automobiles rather than renting them. Suppose further that many agencies were finding that the limitations of their rental contracts (mileage limits, reporting requirements, slow repair service, whatever) were making it harder for them to do their jobs. So the GSA issues new guidelines saying that government agencies should henceforth prefer buying to renting.
Now, there are all sorts of good arguments on both sides of the renting-vs-owning decision. But one argument that doesn’t make sense is to say that government would be “distorting the market” if it decided to buy cars rather than leasing them. A purchased car is a different kind of product than a leased car. If car ownership serves the government’s needs better than car rental, the government is entitled to purchase cars without worrying about how this affects companies in the business of renting cars.
The same point applies to software. The difference between Windows Server 2008 and Red Hat Enterprise Linux isn’t just that one was produced by humorless suits in Redmond and the other was produced by dirty hippies in Raleigh. It’s not even that one costs a lot of money and the other one is free. (Support costs will often dwarf licensing fees anyway).
The key difference is that proprietary software comes with a lot of restrictions about how it may be used—restrictions that don’t apply to free software. Probably the most important freedom free software gives provides is the freedom to switch vendors. If a government agency chooses a proprietary software package, the agency is likely to find itself with serious lock-in problems down the road. Switching vendors will mean switching software platforms, which will involve large conversion and training costs. In contrast, if an agency chooses a free software-based solution, the lock-in problems will be less severe. The agency can easily bring in a new firm to provide support for the existing software. Or it may not need to hire outside contractors at all—the internal IT staff may be capable of fully supporting the system internally.
The freeness of free software is not an esoteric detail about how software was produced, nor is it primarily a matter of ideology. Rather, free software provides direct and tangible benefits to their users. If property rights is a bundle of sticks, free software vendors give you all the sticks up front, whereas proprietary vendors give you only some of the sticks so they can charge you later for the others. And some of the missing sticks are things that actually matter to government agencies. So it strikes me as a no-brainer that the government would—all else being equal—prefer the type of software that comes with fewer strings attached.
It’s absurd to say that the government has an obligation to be indifferent between firms that attach strings to their products and firms that don’t do so. Obviously, there are circumstances where a firm makes such a great product that it’s worth putting up with the associated strings. But it should be equally obvious that software freedom is a factor to weigh in software purchase decisions. And I don’t anything wrong with reminding government IT workers to keep this factor in mind when they make software purchasing decisions..