When I was a lad in high school I worked for Walmart. I learned quite a bit about retail at my early age but one of the fascinating things I used in the late 1990s was a wireless inventory unit, colloquially known as a Telxon. I was amazed by the ability to get inventory numbers on a device without a cable. Since this was prior to the adoption of IEEE 802.11 it was a proprietary device that only worked with that system.

Flash forward to the 2020s. I went to Walmart the other day to look for an item and I couldn’t find it. I asked one of the associates if it was in stock. They said they could check and pulled out their phone. To my surprise they were able to launch an app and see that it was in stock in the back. As I waited for them to return with the item I thought about how 25 years of progress had changed that hardware solution into something software focused.

Hardware Genesis

All problems start as hardware problems. If there’s a solution to an issue you’re going to build something first. Need to get somewhere fast? Trains or cars. Need to get there faster? Planes. Communcations? Phones or the Internet. All problems start out by inventing a thing that does something.

Technology is all about overcoming challenges with novel solutions. Sometimes those leaps are major, like radio. Other times the tool is built on that technology, like wireless. However they are built they almost always take physical form first. The reasoning is simple in my view. You have to have the capability to build something before it can be optimized.

If you’re sitting there saying to yourself that a lot of our solutions today are software-focused you would be correct. However, you’re also making some assumptions about hardware ubiquity already. Sure, the smartphone is a marvel of software simplicity that provides many technological solutions. It’s also a platform that relies on wireless radio communications, Internet, and cloud computing. If you told someone in 2005 that cellular phones would be primarily used as compute devices they would have laughed at you. Because the hardware at the time was focused on audio communications and only starting to be used for other things, like texting or PDA functions.

Hardware exists first to solve the issue at hand. Once we’ve built something that can accomplish the goal we can then optimize it and make it more common. Servers seem like a mainstay of our tech world today but client/server architecture wasn’t always the king of the hill. Cloud computing seems like an obvious solution today but AWS and GCP haven’t always existed. Servers needed to become commonplace before the idea to cluster them together and offer their capacity as a rentable service emerged.

Software to the Rescue

Software doesn’t like unpredictability. Remember the Telxon example above? Those devices ran proprietary software that interfaced with a single server. There was no variation. If you wanted to use the inventory software on a different device you were out of luck. There was zero flexibility. It was a tool that was designed for a purpose. So many of the things in our lives are built the same way. Just look at your home phone, for example. That is, if you still have one. It’s a simple device that doesn’t even run software as we know it. Just a collection of electrical switches and transistors that accomplish a goal.

However, we have abstracted the interface of a telephone into a device that provides flexibility. Any smart phone you see is a computer running software with a familiar interface to make a phone call. There’s no specialized hardware involved. Just an interface to a system that performs the same functions that a traditional phone would. There are no wires. No switches like an old telecom engineer would recognize. Just a software platform that sends voice over the Internet to a receiving station.

Software becomes an option when we’ve built a hardware environment that is sufficiently predictable as to allow the functions to be abstracted. The Walmart Telxon can function as an app on a smartphone now because we can write an app to perform those same functions. We’ve also created interfaces into inventory systems that can be called by software apps and everyone that works for Walmart either carries a phone or has access to a device that can run the software.

Tom’s Take

Programs and platforms provide us with the flexibility to do things any time we want. But they only have that capability because of the infrastructure we’ve built out. We have to build the hardware before we can abstract the functions into software. The most complicated unsolved problem you can think of today will eventually be solved by a piece of kit that will eventually become a commodity and replicated as a series of functions that run on everything years later. That’s the way that things have always been and how they will always be.