"Interfaces and integration are certainly the best places to spend dollars on improving MANMAN," Floyd said. But the cloud is not free, just a lot less costly. Estimating a 30-user implementation -- plenty of the remaining MANMAN sites are small -- he still came in at $6,000 a month. That's $72,000 against the $17,000 plus the expense of purchasing the 3000 and its storage devices.

"You're spending a heck of lot less than that just for the electricity," Floyd said of the cloud solution.

Of course, most companies running MANMAN -- or nearly any other application -- have long ago paid off capital costs for hardware. The costs that remain fixed are the OS and application support ($12,000 in Floyd's estimate) plus the third party software support at $10,000.

Let's see, $22,000 plus that $72,000 is $94,000 yearly. You're up in the cloud in this picture, running a virtualized 3000 server. The license for that virtualization software and its support fee varies, but nobody is reporting much under $10,000. It's a big advantage when you consider the emulated 3000 will operate many times faster than a Series 939.

So someone who stays by rising to the cloud will be up in the $100,000 annual range for five years, annually, using the solution with the longest lifespan (virtualized OS, virtualized hardware) with an application that's just about the most senior in your community. Factor in the costs of purchasing MANMAN over 10 years and you'd add $25,000 yearly. (This is, of course, another cost that most MANMAN sites have paid off long ago.)

But even if you're able to do computing from the cloud, "IT costs do not go away," Floyd said. "Even if you're in the cloud, for any manufacturing company, better utilization requires an IT function." That IT function is a programmer for ongoing development of modifications, at the least. FORTRAN programmers might be hard to find in the middle of nowhere, Floyd added. Lots of US-based manufacturers using MANMAN operate in such small towns, to keep labor costs contained.

The counterpoint of all that expense of working to stay on MANMAN? "The biggest cost of leaving MANMAN is data migration and implementation of the new system," Floyd said. You would retain the cloud costs, the OS and vendor support costs in this scenario -- while the MANMAN site must pay for SAP, or Oracle, or some other ERP solution.

When calculated with this much detail, "It's not a crazy idea to think of staying on MANMAN another five or 10 years," Floyd said. Mobile connectivity will demand bandwidth that might not exist. "MANMAN is cheaper to operate than either an on-premises replacement or a cloud-based replacement. Inertia is the driver, especially if you're retiring in the near future."

The companies using MANMAN aren't retiring, of course. They face a cost to select, acquire and implement and migrate data to a replacement ERP system from "hundreds of thousands to millions of dollars," Floyd said. "Why did all those companies leave MANMAN? Ten years or so ago, they might have had management with some high ambitions."

Or did they really leave because their users lacked a basic understanding of MANMAN, so relied on tribal knowledge in the organization -- "and then they forgot the way their were going things the way they were. And then a couple of really key users left. And you wonder how this stuff works, and why it works like it does. If you change the heck out of MANMAN and didn't leave a good trail, there's no way you could keep track of why you did that modification."

But after more than three decades in the field, there just aren't many bugs left in MANMAN. The 3000 sites that tracked their mods, can keep knowledge of their implementation documented, can keep a FORTRAN developer available somehow -- these are the sites that have added up the costs to stay on an app that was first released in the 1970s, even it hasn't been changed in more than 15 years.