A term often brought up by those who think there are too many LinuxDistributions. They observe the dominance of the Windows and proprietary software markets by a very small number (often only one) of products, and compare this with the over 300 active Linux distributions, and think that this points to something wrong with the latter.

In fact, this is a load of nonsense. In a free, competitive market, the number of competitors is always going to be exactly the number the market can sustain, no more, and no less. Back in the late 1980s, I remember seeing an issue of PC Magazine which did a blockbuster review of about 60 different word-processing packages for MS-DOS. Nowadays you would be lucky to find half a dozen different packages for Windows.

In economics, we talk about the “fixed cost” of setting up to manufacture something (product design, building of factories etc), versus the “unit cost” of building each piece of product for sale (price of raw materials, wages for factory workers etc). Clearly, your profit depends on how much you can sell each unit for, minus the unit cost, plus you also have to amortize (apportion some share of) the fixed cost over the first n units you sell. For instance, you might expect to recover the fixed cost over the first two years of selling the product, just in time to develop a new version with an new amount of fixed cost.

In software, the fixed cost can be quite high for a large and complex piece of software, while the unit cost is really just the cost of duplicating bits, which is as close to zero as makes no difference.

Proprietary software relies on recouping the costs through selling the software. Since the unit cost is essentially zero, the costs being recouped are largely fixed costs. Once these have been paid off, the price of the software can be greatly reduced, and the company can still make a profit.

But conversely, if there are a number of different competitors with proprietary software products, the price could drop to the point where only the number-one vendor is still making a profit, while all the rest are losing money, and eventually go out of business. This seems to happen sooner or later in every market segment with proprietary products, which is why proprietary software is dominated by a small number of very large companies.

So the fixed cost can itself become a competitive weapon: by developing a piece of software with higher fixed costs, you make it harder for your competitors to do the same—you’re raising the barriers to entry into the market. This is why proprietary software tends to suffer from Bloat, where subsequent versions become more complex and need more and more computing resources to run.

With FreeSoftware on the other hand, the fixed cost can be spread out in a number of ways, by reusing code from other projects, and relying on a large pool of contributors to divide up the work between them. This low fixed cost makes it very easy for new players to enter the market: instead of having to develop their products completely from scratch, they can build on work that someone else has done, and just put their own innovative variations on top.

For another example, consider the market for passenger cars: there are dozens of manufacturers and hundreds of models and variations thereon, yet nobody claims that this profusion is “fragmenting the market” or “putting off car buyers”. No single vendor or model dominates the market: their costs are such that most of their models can make a comfortable profit selling just a few hundred thousand units per year worldwide.

And just as with cars, there is no VendorLockIn: if you’re not satisfied with one FreeSoftware support vendor, the cost of moving to another one is nowhere near as great as the cost of moving from one Proprietary software product to another.

Thus, the Free Software world will never create a company as large as Microsoft or Adobe. But it will sustain a lot of smaller ones. Think of this as a good thing, not a bad thing.