POS 1. an abbreviation for point of sale, that magical moment that all the retailer's efforts have been aimed at achieving, when the customer actually hands over his hard-earned cash in exchange for goods. 2. also a specialized software application which attempts to replace and/or enhance a cash register, with varying degrees of success.

Three years ago, about the time I gave up my coding pen for my writing career, I was knee high in POS. By day I toted that C code (running on a custom DOS 3.x something) that made the Sears cash registers work. At about that same time, Sears was beginning to think about their next POS solution. Would it run on NT or Linux, that was the big question. And Sears wasn't the only giant retailer thinking about trusting the Penguin to handle their sales.

Linux appeared ready to take over the whole world, including the world of retail. Home Depot was working on a Linux solution. The Burlington Coat CIO was giving keynote addresses at Linux World Conference and Expo. Musicland was set to roll out new Linux-based code. But then came the dot com crash. A lot of plans got changed.

Home Depot got a new IT chief and his first decision was to stick with Windows. Best Buy bought Musicland and their Linux plans got quashed. Sears is still running on DOS and trying to make up their mind about the future.

Why the slowdown? Lots of reasons. But before I get into that, let me backtrack a bit on my POS expertise. Prior to the three years I spent working on Sears POS, I spent a year doing the same on Fox Photo's. But to get all the way back to the beginning, I need to jump back in time to 1956.

My first job was tending the cash register at the golf course at Fort Benjamin Harrison, Indiana. Fred Keesling was the pro. Mister Keesling taught me how to open the register and where to put the different denominations of bills and coins.

A couple of years later, at a Lone Star Icehouse at the intersection of Naccodoches and New Braunfels in San Antonio (it's gone now, a bank stands in its place), I learned a lot more about working a register. Buster was the store manager. He taught me the correct way to make change. The first step being to count up from the purchase price while dropping the coins into the customer's hand. Then continue with any bills that might be needed. None of this "Your change is 4.38 cents" followed by placing the bills (uncounted) into the customer's hand and dropping the coins on top of them. Buster would fire anyone who made change that way.

He patiently explained to me why his way was the right way. Coins first, bills on top, allows the customer to use the same hand to put the bills into his wallet or her purse. Reverse the order and the first thing the customer has to do is pour the coins off the bills so they can be put away. Counting the change back as you give it to the customer allows them to be assured they are getting it all. Buster was trying to make the customer's cash register experience at the store as friendly and positive as it could be. Today's retail is a far different world. When is the last time someone made change for you the way Buster wanted it done?

As you can see, with this vast amount of experience at an early age, I was particularly well-suited to advise Sam Skaggs (owner of Skaggs Drugs and Skaggs Supercenters, at the time) on the future of POS in grocery stores. The year was 1978. After viewing a few demos of a checkout station with scanning equipment built-in, I told Skaggs flatly that it would never work in the real world. It had a hard enough time just working in the lab.

Now back to the real issue: why hasn't Linux made more rapid gains in this vital market? Why is it, that three years after it looked like a Linux landslide, it still has not taken the retail space by storm? I think there are several reasons. Let's take a look at some of them.

I mentioned earlier that the Sears POS runs a custom version of DOS. It runs a customer BIOS, as well. The reason for this is to allow it to talk to all the hardware that goes into the register station.

The typical heart of a PC-based register may be nothing but a 386, but it has to be able to handle a lot of strange hardware not normally associated with a PC. Things like bar code readers, credit card readers, Wi-Fi devices, hand-held scanners, touch-screens for input, signature pads, special displays, and printers. I know, everyone has printers. But not like the thermal or impact printers that fit inside cash registers.

Each and every one of those devices require a driver. While Linux is gaining more and more driver support every day in mainstream usage, that's not always the case with these specialty devices. No driver, no device. It's as simple as that.

The application software, however, is not simple. POS systems are doing a lot more than determining how much change the customer gets back these days.

It's tabulating inventory, tracking demographics, handling credit applications, getting approval for credit card purchases, scheduling deliveries, entering special instructions for drivers or assemblers, making refunds and exchanges, taking in-store credits, checking stock, printing questionaires, issuing and accepting gift certificates, doing currency exchange, check processing, driver's license search, and finally, reminding the sales clerk to ask if you would like some fries with that.

In the case of Sears, as elsewhere, the code is not only complex, it is old. Years of change have had its way on the application. Original intent is long forgotten. Each patch, each new requirement, each additional bit of functionality has been laid out atop the old. It's not spaghetti code. It's worse. I'm sure the situation is not unique to Sears. Any legacy code that has weathered as many years in service as the Sears POS code has is going to be in the same shape.

It comes down to this: for many retailers the cost to change is simply too great. New hardware for cash registers is a small percentage of the cost of a new store, so startups are more likely to embrace Linux than existing firms. But when you have 40,000 stores with a dozen or more registers each, and they all have to be replaced to roll out a new POS, well, it's a big deal.

Even in this uncertain economy and with the handicaps mentioned above, Linux has managed to grow its presence. KB Toys, for example. Papa John's Pizza for another. KFC in the Philippines for a third. Add Regal Entertainment's chain of movie houses, and the Sherwin Williams paint stores. It is happening today. Keep in mind that MacDonalds owns Papa John's. If they like the way it works in pizza joints, there could be billions served before long.

In summary I'll say that in spite of delays, Linux is continuing to advance in this critical area of the enterprise. With the support and backing of firms like IBM and Waldorf-Nixdorf, many of the hurdles have already been cleared.

It's only a matter of time before Linux becomes the platform of choice for POS applications. Why? Costs. Free always trumps proprietary, and most firms are operating on pretty slim margins these days. As soon as the cost of not changing exceeds the cost to change, we'll see a tsunami of Linux solutions for cash registers.

Remember those grocery store scanners I predicted would never make it in the real world? It took a long time, but eventually they became the standard. The same thing will happen with Linux and POS.

Related Reading:

Papa John Pizza goes Linux

The Home Depot chooses Linux for mobile POS

Linux based POS grew 80 percent

KFC Philippines runs mission-critical applications on Linux

Guess Who's Taking Your Money? Linux Cash Registers Go Mainstream