My first exposure to ebooks was through the Kindle platform. I had money saved from a summer internship, and (among other things) used it to purchase a then-new third-generation Kindle. While I have continued to enjoy some aspects of paper, my Kindle had distinct advantages—most notably that it was actually lighter than many of the books I was reading. With it, I was able to maintain longer reading sessions without having to deal with my arms getting sore.

One of the most remarkable things, looking back on the life of my Kindle, is how long-lived it was. It basically just worked, and continued working, for nearly 12 years. It got firmware updates (including one to enable Amazon’s KF8 format on the device), but those updates never changed how the Kindle looked and felt, or the way it worked. In contrast, I replaced my laptop three times over the same period—of which two were incentivized by major upgrades, and one by the hardware failing out from under me. Apple dragged the laptop world to high-resolution, wide-gamut displays. Internal changes, such as SSDs, swept over the scene, bringing dramatically higher performance. In the land of GPUs, desktops and gaming, we saw a dramatic increase in both compute performance and also the corresponding visual fidelity that could be achieved. In high-performance computing, we saw an increase of 230× in peak performance¹. We even saw the possible start of the end of the dominance of the x86 ISA².

Changes in the software realm were no less dramatic. All three major desktop OSes underwent massive UI overhauls. Many users of Microsoft Windows had their machines force-updated without their permission and woke up to a completely new interface. Mac OS X became OS X, and then macOS, and then stopped using the version number 10.

My Kindle never changed. It performed the same function, with the same interaction model, with the same interface in 2021 as it did in 2010. Until the charging circuit broke, and then it died.

I’m not saying there was nothing to be improved with my Kindle. In fact, I was pleased to see that there were nice improvements that had accrued over the years since I’d purchased my original device. It’s not all sunshine and roses—I’m still on the fence about the touch screen—but it’s unquestionably an upgrade. Still, for all that, my new Kindle has basically the same interaction model as the old one. Aside from moving from buttons to touch, you could nearly pretend it was the same device.

Much of the credit for this stability goes to Amazon, who steered the ship straight for over a decade. That may seem like the obvious (and possibly even easy) thing to do, but it’s really not, especially in a world with as much churn as we’ve seen in the last decade. Consider, for example, the saga of Microsoft, who closed their ebook store and deleted all of the files from customers’ systems. Or the number of perfectly good Google properties that have been shuttered.

But there’s also an extent to which I think this shows what is possible with appliance computing. A device with a well-defined, stable function doesn’t need to change. That doesn’t mean stability is inevitable—as the IOT market has disastrously demonstrated—but it is possible. Or perhaps to be more precise: stability is possible where there is a stable business model that doesn’t depend on selling devices or making them work in a radically different way. Kindle, apparently, is one of those business models.