why software is a natural monopoly and some repurcussions

an unfinished essay by Bill de la Vega

 

What is a natural monopoly?

Typical examples of so-called "natural monopolies" are the telephone & railway systems. All the phones need to be connected and work together to attain the highest benefit. All trains must "fit" on the same track for similar reasons. The monopoly is called natural because many competitors in these markets tend to die out leaving just one or a few providers for any given geographic area. The subject of monopolies and near-monopolies is a large area of economics, so I'm glossing over what could be a several books worth of theories and case studies (because you wouldn't read them and I can't write them). But, the essence of the concept of a natural monopoly is there, that these are markets that tend to have one or few providers.

The theory goes that the larger company has lower costs and better services. For example, if there were 2 (incompatible) phone companies with a 60/40 split in the number of customers, the 60% company will grow to become the 100% company, other things being equal. This is because consumers joining the system (getting a new phone) would prefer the larger company so they can reach more people. For similar reasons, those in the 40% camp have an incentive to switch in order to reach more people. If consumers are aware of this difference in service the switching will happen very quickly as the incentive to switch grows with the split.

Note that the better service is not a technical superiority like faster trains or less noise on the phone lines as I stated "other things being equal". The improved service is just an attribute of being bigger. For the railway example, the larger company might travel to more locations or provide more frequent runs. In the phone example, the larger institution provides access to more people.

Note also that it is difficult to compare two quantifiable differences in the quality of service. Consider our phone system example, imagine the provider with 60% of the users had a somewhat noisy sound quality, while the 40% phone system had superior sound quality. Even though engineers can quantify the noise on the line, it is very hard to tell whether the market will prefer the technilgical difference or the size difference. If for instance there is a sizable number of people who prefer the sound quality, they may switch fast enough to make the 40% system dominant at which point it will again snowball into the one phone system. On the other hand if very few people care about the noise again we have the 60% system becoming the single provider.

So what?

What difference does it make that something is a natural monopoly, you might ask? Well, in the US it means the governement treats it differently than other industries. Traditionally, the government has converted natural monopolies into quasi-public institutions or regulated their prices.

The rational has been that monopolies hold prices higher than a competitive system would, so the  government is protecting the consumer. The consumer gets the benefit of a one uniform system, while paying about the same as in a competitive system.

In practice the government action is questionable on several grounds.

1. It has often been competitors that have conviced the government to step in. This raises the question of who is benefiting from the intervention.
2. How will the governement decide what is an acceptable price? In a competitive system the consumer benefits from inovation from any company.
One could easily imagine a monopoly, like the one phone company charging much more for a service than they might under a multi-provider system.
 

How is software a natural monopoly.

1. Software is much more efficient with uniformity.
Uniformity benefits the users because they only have to learn one set of arbitrary quirks.
Consider phones and cars. These two ubiquitous tools have essentially one interface each. On all phones the 12 main buttons are all in the same positions and labeled the same. All cars have a wheel, blinker switch, gas pedel & brake in the same place. This uniformity is key to the interchangability of skills with these tools.

Software needs this uniformity even more so than cars and phones because software is so much more complicated to "drive". Speaking as someone who has spent more than a few weeks getting comfortable in each of VMS, UNIX shells, Motif, Open Look, MacOS, OS/2, MS-DOS, Windows 3.x and Windows 4.x, they all have there standard ways of doing things. A person can become fluent in any one of them. However, any time spent learning one of them is time spent forgetting the others. What a pain.

This uniformity benefits users in several other ways. Consider looking for an systems person. If all computers had the same interface, it would be easier to find and judge candidates for the job. The pool of applicants would be bigger.

2. Monopoly provides cheaper &/or better products due to the vast economies of scale in software
Software creation has a giant fixed cost (mainly development) and a tiny variable cost (mainly distribution). This leads to what are called "economies of scale". Basically economies of scale mean that a large producer can either do it better or cheaper than a small producer, other things being equal.
 
3. Software is revised frequently, and it very brittle. This causes buyers to be (reasonably) reluctant to buy from smaller vendors. This is unfortunately a self fullfilling belief. Since software is brittle everyone knows they may be hung out to drive when a smaller company folds or discontinues a product. Oddly enough this leads to users buying from the largest vendors. The funny part is that these vendors set the standards for correctness. One of the first company to stop writing software for 16-bit Windows was Microsoft...

 

Rebuttals of ways that software is not a natural monopoly.

It is cheap to enter the market.

This is certainly true compared to manufacturing intensive markets like cars and railroads. Any 2 or 3 smart people can start a software company. Many do. However, none of them compete well on the huge, ubiquitous markets. On a dollar basis, most software is produced by the big companies.

It costs a certain amount to innovate a new idea. Many software startups are one idea shops. They typically start with a few acedemics (who often come up with these new idea while doing research) and exploit the hell out of it. However, the big companies are waiting and watching. If any new idea flops they ignore it, if it succeeds they either reproduce it at a fraction of the cost (reimplementing a proven concept is both much cheaper and often superior to the initial implementation) or they decide the mindshare is worth it, they buy the inovator.
 

Standards allow many players in a market.

Two things about standards falsify this claim.
First standards typically fall into 2 areas. The first are successsful, open, multiply implemented standards. Email is one of these. The RFCs that define basic email have been implemented in many products and are making money for many companies.
The second are those have not really been multiply implemented. Before MS entered the browser arena it was an open standard with one grossly dominant implementor (mosaic, netscape). For the past 2 years the 2 companies have been racing to implement their own extensions, so as to cut out the other. Oh yeah both companies will tell you they are implementing the extensions to provide features the users were screaming for. I can't imagine any user caring that much about the protocol's features. COM/OLE is another one of these standards. Probably 80% of all com components are made by MS. Don't get me wrong COM is a great idea.
 

it is hard to switch (training, hardware, software overlap)
 
 

In this essay I attempt to show that software is essentially similar to these systems

One phone is useless, two phones are nearly useless, and having a phone when everyone you know has a phone is almost indispensible.

In this discussion I'm considering generic software.

What about MS?

What I believe they provide is good degree of ease of use and the backing of a large company that is commited to providing that degree of ease of use.
 

What it isn't about.
Cost of software: consider LINUX, GNU, ...
Technical supoeriority: consider B, Next, LINUX, ...