Why do local television stations tend to schedule their news shows all at the same time? Do they all get together and conspire to set some standard? There is an good explanation in The Economics of Network Industries by Oz Shy blaming it all on network effects. He uses game theory. The payoff matrix creates game where the stations move into synch because that splits the available of viewers.

As a kid I used to wonder how the TV stations managed to conspire to get their commercials to all play in synch. As an adult I ponder why all my children’s schools have their spring vacations scheduled in different weeks, or why my west coast friends occasionally schedule meetings at my dinner time, but my east coast friends rarely schedule meetings at their breakfast time.

This kind of synchronization, or lack of synchronization is a public good, like a standard.

Here’s another interesting example: the monthly billing cycle. Recently (I’m told by my wife who actually pays the bills) vendors have started playing games with the billing cycle. It used to be that you could pay all your bills once a month. It would seem bill vendors have discovered they can trigger more late payment fines by introducing nonstandard cycles.

These “fines” are probably very nice source of revenue for them. Very similar to how video stores make most of their income on late return fees. The video store that tunes their return policies to maximize fines will be more profitable and thus more likely to survive than one that doesn’t – assuming customers don’t notice.

The billing cycle game is a nice example of cost shifting due to measurement difficulties. Trivial to measure: the increased income for the billing department. Hard to measure: the inconvenience inflicted on customers. Even harder to measure: number of customers that switch vendors do to this marginal decrease in the quality of the company’s service. Management in the billing department gets a bonus in situations like this.

Another example – meeting scheduling software. The gorilla in that domain is Microsoft’s exchange server.

People gloss over how sticky these issues are.

Any theory of standards needs to include standard languages; i.e. French, Japanese, Java, etc.

No widely sucessful human languages were designed by professionals. Languages are extremely sticky (i.e. nobody can learn a second language perfectly after puberty). Most people acquire their primary language by a cognitive short cut. They adopting the local dialect. Preferential attachment indeed. Like any systems with large a installed base rationalization after the fact is extremely difficult. But it does create a vast and entertaining liturature of how we got here.

July 18th is Johann Schleyer’s birthday. Schleyer was a late 19th century German who invented a Volap’k (notice the umlat) a constructed language. It was quickly displaced by Esperanto.

I’m amused that Schleyer was extremely fastidious about retaining control over his IP rights. It must have been an interesting time, the late 19th century, when it first became plausible that individuals might own one of these widely used standards.

This is a list of what drives standardization, i.e. what the goal or purpose of creating a standard is, or what value provides the motive force to cause a standard to come into existence.

I mined many of these out of Industrywide Voluntary Product Standards by Hemenway. Many many moons ago. It has drifted from that original over time. I keep making additions.

frame the competition

A standard make it easier for buyers and sellers to negotiate an agreement. It frames that negotiation. It defines the landscape upon which the competitors will meet. It reduces the complexity of the buyers shopping significantly.

accelerate growth

By clarifying how cooperation takes place, enabling a platform, encouraging complements, simplifying contracting, etc. etc. standards accelerate the growth of the industries around them. For example after the standards for computer interconnection settled in (two examples: the IP protocol stack, or the PC backplane bus) the growth in took off.

cooperation

Enable and frame cooperation between parties to the standard. For example, the standard to drive on the right hand side of the road makes things go smoothly for all parties.

exclude competitors

“A dominant entity (or oligarchy) can use standards to shape the competition and define things to the disadvantage of competitors. This can be done by using the standard to focus on attributes that benefit the standard maker. (See also stability.) The standard maker can gain some first mover advantage, an advantage which (in multiple rounds of the game) can keep competitors “”chasing tail lights.”””

raise the bar

A powerful player may use standards to force suppliers to stretch for a higher level of quality. Governments, for example, do this when they set fuel efficiency or safety standards. Microsoft does this when they define the standard PC.

f(in)->out

A standard can specify the inputs to a good, the process used to make that good, or the outputs. It is often the case that one of these three is far more tractable to measure than the others (it’s easier to measure students per class than it is knowledge instilled). In the construction industries, for example cement construction, the process is critical to a quality output so standards are set to guide inspection steps in the construction.

worse is better

A standard requires both a spec, and a community of users. Attributes that improve a standards abilities to achieve and maintain critical mass often compete with other attributes. Simpler specs are often incomplete; sometimes quite incomplete. The original spec for C, SQL, and Java are all examples of the ‘worse is better’ syndrome. Sometimes the community/spec doesn’t grow – or it fragments – so this is all you get. The early adopters often feel they have been deceived.

brandname displacement

Brandnames are one of the ways used to signal quality to buyers. Buyers maybe forced to use brandnames as a substitute for more accurate measures of quality when buying complex goods. Government standard setting often does this; for example, in the battery industry AA batteries are all built to a standard and don’t vary significantly from brandname to brandname.

platform

Create a platform in service of a bloom of applications above that platform. This is similar to the desire to create compliments.

scale economies

Enable the cost advantages that comes from big: distribution, manufacturing, purchasing, externalities, network effects, etc.

mutual aid

Enables parties to share resources to solve problems. For example, assuring that allied armies share the same size bullets, or that communities share the same size fire hoses. The most generic of these is the sharing of labor, members of the standards community benefit from the problem solving done by other members. For example, if I purchase a Macintosh I get some of that; if I purchase a PC, I get some more. This and a number of other aspects (platform, complements, etc) all create a network effect around standards.

preempt the rush for the bottom

“In the absence of standards vendors may sell efficiency (i.e. lower price) while in fact they deliver is actually lower quality. This is what led to the airline industry writing a standard to define ‘a sandwich’ or the construction industry to write a spec for the “”1 inch board””.”

my way

It is extremely common to find individuals or small groups that have codified their operating rules into what they call standards. Coding conventions are one example of this. Such groups will share these with others; their motives in doing that run the range from enthusiasm to ascription.

aid smaller buyers

Large buyers can afford the overhead of writing their own specifications and have the buying power to enforce them. Historically milspec standards, since they were public, benefited smaller buyers who could free-ride on the public good.

survey current practice

Many standards writing processes start with a capturing a list of current practices – and then some of them never get very far beyond that. The vocabulary they create for describing how things are done in this first stage can be extremely valuable. The standards for medical information systems are like this.

quality

Most goods have numerous heterogeneous attributes. Standards tend to strip down those attributes to a few, and then define quality in those terms. Grade A apples, for example, might mean only that they are blemish free; but say nothing about their taste. This is related to framing the competition.

lawyers or standards

Standards are often used as a device to avoid costly litigation or accidents; in this sense a standard is a cheap form of contracting.

amusement

“Standards often seem designed to amuse. For example, the standard for a 2×4 isn’t 2 inches by 4 inches, but rather something less than that. This kind of thing arises out of the ‘preempting a rush for the bottom’ and the ‘cataloging of current practice’. It also arises out of the need for stability; as with the definition of “”the hundred year storm””; which after it was defined turned out to be misnamed.”

stability

Standards tend to create homogeneous markets; this tends to lower the chance of disruptive innovation.

create complements

Vendors desire that complements be of high quality and low cost; since standards tend to create those aspects it maybe in the interest of a vendor to create standards for those complements. For example when Apple introduced user interface standards for Mac programs they were laboring to make their complementary products, i.e. MacIntosh Applications, higher quality.

signal value

“Vendors often use standards compliance for it’s signal value. “”Version 12, now fully certified XHotStuff support.”” “

IP pooling

IP creates monopolies, which in turn frustrates the creation of markets and complements. Firms may find it necessary to cross-license or create IP pools to work around this barrier to market growth. [See Carl Shapiro’s paper “Navigating the Patent Thicket: Cross Licenses, Patent Pools, and Standard-Setting”]

prevent stranding

In the absence of standards and in the presence of strong network effects early buyers are likely to find they have made happened to choose a vendor that fails to win the leading position. Such buyers are then forced to bear the cost of being stranded and switching to the winning, now standard, vendor(s).

taxonomy

Diversity of practice makes standardization harder. One scheme to move toward something less diverse, more standard, is to catalog the existing practice. This might not be, strictly, a motivation to standardize; but rather a tactic.  This is, of course, self referential and intended to assure the above is put to good purpose – e.g. – epistemic closure.

interoperablity

What to say…

Bracketed by an essay on reading glasses and flaming
Tim Bray pens a nice essay about when to play the standardization card.

Very nice short essay by Tim Bray on the end game in standards for character sets.

Back in the 1970s Brian Reid wrote in one of the earliest versions of the manual for a text formating program called Scribe that the Stanford extensions to the ASCII character set where either innovation or terrorism depending on where you stood. A quote I always find ironic since Brian went on to help found alt.* on usenet.

Some of my favorite stories about standards involve slime-balls and other criminals. For example the international airline industry had to set a standard for a sandwich because some of their members were passing off such low quality things as “sandwiches” the rest of the industry needed to write rules to standardize a minimal “airline sandwich”. Another nice example is that there was a time when people making shoe leather would add salts to the leather until they reached the physical limit on how much salt could be added. That became the standard. This pattern, where the standard is in fact a floor onthe acceptible behavior is a surprisingly common form of standard.

But this is this story is a whole different example of criminals and standards. In this case the criminals would prefer that their victums adopt the standard. So when their victums take the bait the crooks redirect those who were not up-to-date with the latest standards to the the web site of an advocacy group. That in turn slanders the reputation of the advocacy group.

Thanks to the most excellent Dive into the Mark.

When the North won the civil war they forced the South to switch to the northern train gages.

When Nepolean marched across much of Europe he triggered a general switch to driving on the righthand side of the road.

When ever you get a huge disruption it creates an oportunity for alternate standards to displace the previously dominate ones. This war is apparenly no exception.

Same as it ever was? Carpet baggers.

“What few things need to be the same so that everything else can be different?”

              - Michael Tiemann, CTO of Redhat

Boy does that go to the heart of the standardization/hub/network-effect problem in short order. While it is a good question from a public-good point of view, from a private-good/capitalist point of view the more likely question is “What things can be …”

Generally the network effects (demand side scale effects) cause one standard take over and drive out a competing standard. Sometimes multiple standards survive over time because they operate in seperate niches. Which side of the road people drive on is an example of that. Here are some actual numbers; which surprised me.
Almost exactly one third drive on the left while two thirds drive on the right. What’s the chance of that? I wonder if it means something.

In a modern example, it’s hard to see how .NET and J2EE can both survive in the long run. I can’t see how they have different niches.

Getting a winner requires that one bandwagon get out in front of the other. How far in front? Well when the one bandwagon begins to capture enough advantage – enough extra value – usual from it’s network effects to pull people off the other bandwagon – to pay their switching costs.

Easy way to get that network effect advantage is to capture a lot of users fast. Users who are just entering the game for the first time. So they don’t have as high switching costs. New users are, sadly, often quite nieve. They want to pick the right bandwagon. They don’t have much information. They holdback. They look for signs of what the future might hold. They attempt to see thru the fog of standards war advertising. They do what the guy next door did. They build out from what they aready have. They casually pick something, not realising they are picking a bandwagon.

Stefano Mazzocchi posts an interesting note that illuminates another aspect to the bandwagon problem. It’s also an interesting addition to the worse is better thread.

I’m not so optomistic about the good ideas part.

My experiance suggests that the idea must be good, but reasonably conventional. Otherwise the crowd gawks but doesn’t climb on the bandwagon.