Pylimitics

Digital technology, and indeed technology in general, seems like it moves in cycles of usability and utility. I’m not sure whether the cycles apply equally to every user, or if it depends on the features you, as a specific user, find most helpful.

Printers connected to personal computers are one example. Long ago it could be difficult to get a printer (dot-matrix in those days) working well with your home computer. The interface was probably serial, using an RS-232 connector (or for early Macs, an RS-422 for some reason). If you paid extra for the printer supplied by your computer’s vendor, it probably worked. But if you couldn’t afford the brand premium, you might be faced with actually soldering up the connectors by hand. In one case, I even resorted to writing a new printer driver.

Then printing got easy. If you had a Mac, there was “LocalTalk,” a plug-and-play system for a small network that just worked. Of course, once again you were in vendor-premium land for that. For IBM-compatibles, the annoying serial connection could be (usually) replaced by a parallel connection that pretty often just worked, too.

Over the years and decades, printing got more difficult, then easier, then more difficult again. These days I’ve found it pretty easy, except for the fact that when something doesn’t work, it’s harder to diagnose and fix.

Software works that way too. Microsoft Word, which in my opinion is now an unusable snakepit of unwanted and disconnected features and a nightmare of unusable interface design, was once a Very Good and Useful Tool. This was long ago. The same thing happened to a professional writing program called FrameMaker. I think it was around the 200s that FrameMaker, especially when used on a Unix system, was just about perfect. Then the vendor(s) fixed that by making it disastrously worse.

It’s not just computer hardware and software, although that’s also involved in my next example: automobiles. BMW just recently decided to foist unwanted ads on their customers. This on the heels of the same company deciding that you could buy the car, but if you wanted to use the heaters in the seats (very nice feature during New England winters), that was a monthly rental. Mechanically, drive trains in virtually all internal combustion automobiles are becoming less robust, more prone to breakage, and more expensive to repair thanks to complexity and the cost of parts. I don’t have direct experience of this, but I read a lot about cars and watch YouTube videos.

As for design, the tactile controls you used to have for temperature, volume, and other settings in your car are, in new models, often visual controls on touch screens. Instead of using muscle memory and immediate feedback to, for example, make it a bit warmer or cooler, you have to turn your attention to a built-in iPad, tap through a menu structure, and visually adjust the setting with a virtual control.

My family has more than one car, but their average age is 12. The oldest turned 20 last year. They get regular maintenance, and any repairs tend to be age-related. Rubber components eventually wear out. But there’s no reason I can see why they all won’t last for years.

But I was talking about cycles. The automobile cycles are longer than in the computing world, but if you go by decades, cars got very much better around the 1990s. Maintenance schedules lengthened (although don’t believe everything they claim nowadays), many makers vastly improved rust resistance, fuel injection replaced the Cursed Carburetor, and cars got tremendously safer. Reliability improved — cars from the 1970s were, by 1990s standards, appallingly failure-prone.

I don’t have any personal experience with earlier automotive cycles, but the “model-year marketing” model, which I think rose to super-prominence sometime in the 1950s, could not have helped. Of course, cars in those days weren’t expected to last 100,000 miles anyway, although that might have been by design as well.

Home appliances seem to currently be at a low point in their cycle. Appliances are generally less durable now than in the past, and so many computer-based “features” have been added that complexity has gotten comparatively extreme. I could connect my refrigerator, oven, microwave, coffee maker, and even my bed to the internet. Why would I do that? The motivation is supposedly to give me some kind of information, but in exchange the vendors get a lot of information about me and my usage and my household. I remain unconvinced.

I’m well aware that the problem here might be me. I was originally fascinated by personal computing because it offered so much capability locally. There was no need to create a stack of punch cards, deliver them to the computer operator’s window, and then collect the output hours (or sometimes days) later. You could do it all yourself; everything happened right on your desk. Later on, when laptops arrived, you could take the whole system with you.

I was, and still am, enamored of self-contained, independent, distributed systems. I can (up to a point) write my own software when I want or need to. Some of the programs I still use regularly are home-grown and even predate AI assistants. To be clear, all of those are pretty simple programs. I don’t have the programming chops to create my own operating system, networking stack, or programming language. But I wrote the code for the text editor I use — its roots go all the way back to my first home computer, when I wanted a “word processor” but couldn’t afford one. But my computer came with BASIC, so I made my own. I ended up with enough free memory for about a page and a half of text, but it mostly worked. Eventually, when I scraped together the cash for an Okidata, I could even print pages.

I recreated my text editor in various systems and languages until it got to a point where there wasn’t anything more I wanted it to do. That was sometime in the 1990s. By then it wasn’t written in BASIC any more, and it worked on that era’s version of Macs. I’ve had to revisit it a few times since then, but only to keep it working as the underlying platform has kept changing. The program itself has not changed in features or functions since.

I think most tools, whether they’re physical or virtual, can be like that. To take a trivial example, if you find a hammer from 1826, you’ll know what it is and how to use it, and it will still work perfectly well. I recently located a pen that I believe is probably older than I am, and after cleaning out the dried ink and replacing a cracked part, it works. Unless you’re an expert, you can’t tell the pen isn’t from last year. Engineering and design can reach an end point where the tool matches the task so well that no further changes are needed.

So why do our complex tools keep approaching, and even reaching that point, only to be rendered worse? It’s not the technology; it’s the greed, and the decision, which has turned out to have appalling side effects, of corporate personhood. I’m sure there were people in the organizations that created FrameMaker or Microsoft Word who, at the time, knew their products were just about right. But the organization itself — which imposes motivations on its component people — needed the product to change in order to continue amassing cash, hopefully at an accelerated rate.

Regardless of what their spokespeople may say, it is never a corporation’s goal to create the best tool, the finest design, or the perfect solution to a problem. The goal is money. And if quality, refinement, perfection, or even value get in the way of that goal, well, those are aspects, not goals, so they can be sacrificed. To be sure, the people doing the work care. Those designing the automobile, writing the code, forging the steel, do the work of the corporation, and people generally try to learn their craft and do it well. But the corporation does not care about the work. The work is secondary to the goal. There must be growth. Greed must be served.