Tangled Webs
Moore's Law Repealed
Jul 1, 1997
Issue 2.9

Upholding the Law

Perhaps the most repeated half-truth in the computing industry is that computers are constantly becoming faster and cheeper. Over the past 15 years, I have owned over a dozen personal computers. Each one has been more expensive than the last, and has taken longer to boot up.

This misunderstanding stems from a 1964 paper published by Gordon Moore. The paper received little attention at the time, but laid the foundations for what would become known as "Moore's Law" -- that computer processing power doubles every eighteen months. Unfortunately, over the last decade, each advance in hardware performance has been matched by a corresponding drop in software efficiency. For the most part, users are spending a lot of money just to tread water.

Gordon Moore went on to co-found Intel, and Moore's Law went on to become the bedrock of today's software industry. By and large, software developers consider it axiomatic that the next generation of hardware will be able to run the bloatware they write today.

Furthermore, almost every advance in software development, from object oriented programming to RAD tools, has decreased the time and effort required to develop software at the expense of more bloated, slower products. For example, despite advertising implications to the contrary, the Windows95 versions of Access, Word and FoxPro are considerably slower than their Windows 3.1 ancestors.

With notable exceptions in image processing and high-end database servers, users have derived little benefit from recent increases in hardware power. It takes me just as long to write and edit these articles in Word97 as it did in Word 1.0. For history buffs, the original Microsoft Word was a Macintosh-only product, and back in 1985 the entire MacOS, MS Word and dozens of document files fit neatly onto a single 400K floppy disk.

Bloatware is sound business strategy in today's market where Moore's Law holds, new computer purchases rather than replacement purchases drive the market, and most software achieved its requisite functionality years ago. Under these conditions, it is simply bad business to write tight, stable, efficient software. Bloatware is cheeper to produce and easier to market.

Civil Disobedience
As the number of replacement purchases catch up with the new computer purchases, however, we are seeing the beginning of a bloatware backlash. Corporate users in particular have begun to question the wisdom of spending millions of dollars to buy new software that runs slower than what they have now. Not a few companies are even sticking with Windows 3.1 rather than upgrading to Windows95. Microsoft, however, is trying to ensure that this will not be a viable long-term option.

Gordon Moore does not share the faith the software industry places in his law, and there is really no rational reason to believe that processing power will double every 18 months until the end of time. In fact, there are signs that the rate of growth may be slowing already.

Furthermore, every generation of chips costs twice as much to develop as the last, this is often known as Moore's Second Law. Will the semi-conductor industry -- or any industry for that matter -- be able to spend $10 billion on plants that will be obsolete in a few years? If trends continue, a decade from now the cost of developing the next generation product may well be beyond than the R&D budget of the entire industry.

The Revolution will not be Shockwaved
Without a doubt, someday the acceleration of raw computing power will slow. When that happens, we will see a fundamental shift in the way software is developed and marketed. But as we say in the computer business, "shift happens."

The post-Moore market will hit the software industry like an oncoming train, but it could well be a light at the end of the tunnel for the average user. Software companies will be unable to simply add more bloat and release an upgrade every year. Perhaps they will actually find it beneficial to turn inwards and reengineer software to consume fewer resources, and to run faster and more stably.

Production programmers who get paid by the number of lines of code would become a thing of the past. The ability to coax the most performance out of the fewest resources would once again be the hallmark of a valuable programmer.

The end of feature bloat may also force magazines to move away from reviews that are just feature comparisons; where the package with the most dots in its column is declared the best choice. Perhaps they would focus on interface, efficiency and usability. And perhaps, just perhaps, the end of Moore's law will result in something virtually unthinkable in today's software market; software being sold based on the quality of the programming.

[ Home Page] [ Back to Index ] [ Previous Issue ] [ Next Issue ]

© Copyright 1997, Tim Romero, t3@t3y.com
Tangled Webs may be distributed freely provided this copyright notice is included.
The Tangled Webs Archive is located at http://www.dotco.com/t3/tangledwebs/index.shtml