Douglas Crockford




2019 Appearances





How JavaScript Works


June 2005

This month's report looks at the Digital Tuner Rule, Current, Digeo, and Software.

Digital Tuner Rule

Screen Size Old Rule New Rule
36" and larger 50% by July 1, 2004 50% by July 1, 2004
100% by July 1, 2005 100% by July 1, 2005
24" to 36" 50% by July 1, 2005 50% by July 1, 2005
100% by July 1, 2006 100% by March 1, 2006
All devices 100% by July 1, 2007 100% December 31, 2006

On June 6, the FCC announced a modification to the digital tuner schedule. It accelerates the deployment of ATSC tuners in integrated television sets. The rule change is intended to benefit broadcasters and to more quickly allow the Analog Switchoff.

As before, the rule only applies to television receivers with integrated analog tuners. It does not apply to monitors.

The ATSC tuners in most sets are not used. The signal source is usually a STB provided by a cable or satellite service.

In February, NCTA announced that it had installed 27,000 CableCards. At that time almost a million DCR TV sets had been sold. That means that most of the people who bought DCR sets are not using the DCR feature.

I believe that it is a bad product design strategy to put expensive, complex features into products that most people do not use.


Last year, Former Vice President Al Gore and Joel Hyatt (the founder of Hyatt Legal Services) bought Newsworld International from Vivendi Universal, possibly with the intention of launching a liberal-biased news service, to compete ideologically and commercially with Fox News. Such a thing appeared to be financially unworkable, so the service was morphed to a video blog channel called Current. (The name is clever because it can mean both the flow of electricity and now.)

The new service launches on August 1. It will target 18 to 34 year olds. They are inviting viewers to submit video material, which if accepted will be adapted for broadcast. Video production tools have gotten really cheap and good. It is possible that Current will become an outlet for homemade video. I think a better outlet would be the Internet. People can make their stuff available to the world now without first getting the approval of anyone. Current's bold new model may be too late.


Paul Allen was one of the founders of Microsoft. He is the Chairman and largest shareholder of Charter Communications, a cable MSO serving about 6M households. In 1999 he founded Digeo to develop interactive technology for cable. The company got off to a bad start, but was able to turn itself around by copying TiVo. The result is Moxi, a dual-tuner multi-room cable-ready DVR that can work in Motorola and Scientific-Atlanta cable systems.

Moxi is now available in some Charter cities, as well as some Adelphia, BendBroadband, and Sunflower cities. Digeo is also trying to make deals with the other MSOs. They are also trying to license their X-Stream chipset.

The Matrix Reified

The Matrix movies were recently released in a 10-disc Ultimate Collectors Edition. There is a disc for each of the three movies, plus a fourth for the Animatrix, a Matrix-themed anime anthology. There are 5 discs containing documentary material about the making of the movies. The tenth disc containing two documentaries: Return to the Source: Philosophy & The Matrix and The Hard Problem: The Science Behind the Fiction.

In the last documentary, there are several interviews with computer scientists and researchers talking about virtual reality and artificial intelligence. Most of these guys completely missed the point of the movies. The Matrix is a technological prison which is used to pacify and enslave humanity. The villain in the first movie is a traitor who decides to betray his friends in exchange for a chance to return to the Matrix. The experts in this documentary are attempting to do the same thing. In fact, they are going significantly farther: They are attempting to build the Matrix.

They cite Moore's Law as evidence that it will be possible to build a convincing simulation of reality: Ultimately, computers will be so powerful that there is nothing they cannot do. This line of reasoning is incorrect because of one obvious fact: Software is not subject to Moore's Law; software is subject to Murphy's Law.

As predicted by Moore's Law, we have seen amazing increases in the capabilities of computers and related technologies. In the past twenty years, we have seen improvements of a factor of 1000 or more in things like processor speed, memory capacity, disk capacity, and network connectivity. However, we have not seen significant improvements in software or any of its characteristics. Software has not become significantly easier to use or cheaper to produce. We have seen little improvement in real software value in twenty years, and most of that improvement is due to hardware factors like increasing memory capacity.

We have made amazing progress in the mechanical design of robots, but not in their programming. We are no closer to building a robot that understands The Three Laws of Robotics than we were when Asimov first stated them in 1942. This is the most important programming component of an autonomous robot. We have made no progress on it because we have such a poor understanding of software.

The Software Crisis

In 1968, the Software Crisis was identified. These problems were routinely observed in software development:

There was a lot of writing about the Software Crisis through the 1970s and 1980s. At some point we stopped talking about it simply because we got tired of it. The problem has not been solved. In fact, it has gotten worse because computers now touch many more people in many more ways. The Software Crisis can now be seen in TV sets that are too complicated to set up and microwave ovens that are too complicated to cook with. The difficulty of software development is the leading cause of bad user interface design.

The root of the Software Crisis is in our poor understanding of software. As software increases in importance and applicability, the pain of the Software Crisis only increases. There is no field of research that can produce greater immediate returns than research in the development of software.

Small Leaps

Improvements in software come in small leaps. In the past 50 years we have seen only four such leaps. Each represents a small factor increase over the previous state of the art. In electronics, we see a doubling in about 2 years. In software, leaps take about 20 years.

Leap Language Description
Symbolic Assembly Language SOAP, BAL Programs expressed as symbols and macros instead of as machine codes.
High Level Language Fortran, Cobol Programs expressed as statements and formulas.
Structured Programming Algol 60, C Structured discipline helped manage statement-level complexity.
Object-Oriented Programming Smalltalk, Java Better modularity and reuse.

There have been many failed leaps as well, such as Automatic Program Generators and 5GL.

A Possible Next Step

It is difficult to predict what the next leap will be, but I think a leap can be taken by combining two previously discarded ideas:

I think that the combining of these two ideas can produce the next leap.