Sunday, June 2, 2019

The Downside of Gutenberg Economics

Pipe organ

18th century pipe organ
at the Monastery of Santa Cruz, Coimbra, Portugal

A friend of mine had a long professional career as an organist and singer in jazz clubs and cocktail lounges. Upon retiring about a dozen years ago, he spent the healthy sum of $18,000 to purchase a top-of-the-line Hammond electronic organ for his home so that he could continue practicing and playing. This organ has now developed problems involving its computer chips and, despite the efforts of various expert mechanics, including the company now owned by Suzuki Musical Instrument Corporation, is considered unrepairable. An instrument that, if it were a human being, would hardly have passed puberty, is now a nice wooden cabinet enclosing a couple of dead circuit boards. And therein lies a tale of the late 20th and early 21st century.

The original organ, familiar from your neighborhood church or a 1920s-vintage movie theater, is an arrangement of pipes connected to valves and an air box. Some of the pipes have notched throats, like a whistle, and some have embedded reeds, like an oboe. All draw breath from the blower feeding the air box or plenum. When the organist presses a key on the console, it activates the valve, letting air into the corresponding pipe and making it sing. This is a design going back 2,300 years to the Greeks, and the oldest extant organs go back about 600 years. If one of these organs breaks, it can be repaired. The job takes an expert who generally has apprenticed himself—because this is rarely a skill undertaken by women—with an older generation of experts and devoted his life to the job. Maintaining a church or theater organ is expensive, probably about as much as building one from scratch, but it can be done.

The original Hammond rotary organ was not based on pipes and blowing air but on something called a tone wheel. Each key on the organ corresponds to an iron disk that has a series of notches cut into its edge. The number and spacing of the notches are mathematically configured so that, with all the wheels rotating at a constant speed on a common shaft, a sensor facing the edge of each wheel picks up a signal in impulses per second that matches a musical tone—for example, 440 impulses per second generates the sound of the A above middle C. To recreate the harmonics that a pipe organ makes by pulling stops to bring in the voices of pipes in various relationships above and below the note being keyed, the Hammond organ uses drawbars to pull in more or less of the electric signals from each tone wheel above and below the wheel being sampled. If one of these original Hammonds breaks, it can be repaired. Again, the job takes specialists who may have to fabricate some of the parts, but the system is mostly mechanical and can, for a price, be restored to working order.

But the organ my friend has, and all the Hammond organs you can buy new today, are electronic. There is no pipe, no tone wheel, nothing mechanical and touchable. The sound of a pipe organ or the cherished buzz of the tone wheel organ has been electronically sampled as a wave that is then encoded in a computer chip as a piece of digital information, the same as if it was a number in a spreadsheet, a word in a document, a line in a graphic, or a pixel in a photograph. When you press the key on such an organ, it calls up the corresponding note from digital memory. When you adjust the drawbars, they pull in the harmonics of other sampled notes. But it’s all just bits and bytes. If your electronic organ has a burned out chip, and that chip is no longer made or available somewhere in stock, your organ is dead.

So, the irony is that with the right people who know how to straighten and fabricate the right parts you can fix a 200-year-old pipe organ or a 50-year-old tone-wheel organ, but nothing can resurrect a 20-year-old electronic organ, piano, keyboard, computer, cell phone, or other digital device if the defunct chips are not available.

And the chips are not available because computing technology—under the force of Moore’s law1—moves forward so rapidly. Designing, plotting, masking, and photoengraving computer chips is a function of what I call “Gutenberg economics,” where the creator makes a single template for a book page, a chip design, or any other intellectual property and then prints off as many as are needed at an ever-diminishing cost per unit.2

The downside with all of this, of course, is that once you have performed all these preparatory steps and finished your print run, you are ready to move on to the next project. If you made computer chips with capacity X and capability Y a dozen years ago, and today’s chips have a thousand times that capacity and a million times that capability but operate slightly differently—say, with different inputs and outputs on the contact points in modern circuit boards—you are not going to go back and do a limited run of antique chip designs just because someone somewhere has a board in an organ, desktop computer, cell phone with a burned-out chip. Like the cost of makeready on a printing press, the costs of setting up for the fabrication of a semiconductor design are the largest part of production. No one pays to start the process over again for just a few copies of the printed book or newspaper—or for a couple of hundred computer chips to repair outmoded products.

So, while the computer chip itself might be virtually immortal, the device in which it’s installed is susceptible to the least defect in just one of the many chips that power it. Burn out a memory component that, on the scale of a large-run chip fabrication, might originally have cost less than a dollar to make, and your $18,000 Hammond organ is electronic waste in a nice cabinet.

In the same way, you can still buy a 1929 Model A Ford and get it serviced and repaired, because there is a small but loyal following for these cars, and gaskets, bearings, carburetors, and other parts can still be sourced, refurbished, or fabricated. You can even restore a 1929 Duesenberg and have yourself an elegant town car—if you have enough time, patience, money, and access to a good machinist. But let a chip burn out in the engine management or fuel injection system of your 2009 Toyota or the charging system of your 2019 Tesla and the chances, over the years, become vanishingly small of finding a replacement. Once the car company notifies you that it will no longer support your model year, you are pretty much on your own, and your treasured vehicle becomes a ton and a half of useless scrap.

In the same way, a document hand-lettered on parchment or printed on acid-free paper can survive for centuries. It will still be readable, even if you require a linguistic expert to translate the words. But recovering the early draft of your novel from a twenty-year-old 5.25-inch floppy disk or a ten-year-old 3.5-inch floppy can now take expert help and, within a couple of years, may not be possible at all. Despite all the standardization the computer industry has made in disk formats and plug sizes and capacities, it’s a sure bet that one day your two terabyte hard drive or 128-gigabyte USB thumb drive will be unreadable, too.

In the same way, many of us music lovers have had to recreate our collections as vinyl gave way to tape, gave way to disk, gave way to MP3. And our movie collections have moved from VHS—or, spare me, Betamax!—to DVD, to Blue Ray, to MP4.

This is not planned obsolescence or some evil scheme by the record companies to make you buy The White Album again. This is simply the advance of technology. Last year’s hip replacement is just not going to be as good as the metal or ceramic joints that orthopedic surgeons will be installing in the year 2028. You might remember fondly that 1970 Ford Mustang you once had and want to own again, but its engine mechanics, gas mileage, and service life will not be as good as today’s model. And when your 2019 Mustang bites the dust because of a dead computer chip, the models that will be available then will be even better yet.

As I’ve said before,3 once the Industrial Revolution got underway—then morphed into the Information Age, and is now fast becoming the Automation Revolution—we all got on a fast escalator to the future. Some of it we can already see. Some only science fiction writers can imagine—and sometimes they get it wrong.4 But it’s all going to be amazing.

1. Moore’s law says—or used to say, starting back in 1970—that the processing power of computer chips doubles every two years. I don’t know if that still holds, because the effort to cram ever more transistors onto a silicon wafer is now approaching physical limits. And new concepts in quantum computing may bring along even greater advances in computing power.

2. See Gutenberg and Automation from February 20, 2011.

3. So many times I won’t bother with a reference.

4. See, for example, Robert A. Heinlein’s early fiction, where we travel in space but still have to program electromechanical computers the size of a room.

No comments:

Post a Comment