In general, I would agree. However, the longer that old technology is deployed, the more expensive support costs. Parts become obsolete, engineers retire or move on, IT has to maintain archaic devices, operating systems, and tools.
> However, the longer that old technology is deployed, the more expensive support costs.
That's true, but military logistics has adapted to mitigate some of these costs.
> Parts become obsolete
There are two ways this is dealt with. The first is an activity called Diminishing Materials Suppliers (DMS). DMS is an ongoing support activity during production where the availability of parts is monitored and alternative parts are selected and evaluated if a manufacturer announces they are stopping or modifying production of a part for which they are the sole supplier.
The second is the lifetime buy. Near the end of a production run, and sometimes earlier, engineering figures out how many of a given part is needed to complete the full production run and provide spares for rework and repair. Purchasing then gets approval under the contract to purchase the entire quantity of materials and the stock goes to a warehouse until it's needed.
> engineers retire or move on
Knowledge transfer is one of the big reasons the defense industry is so big on systems engineering and mountains of documentation. Obviously that only goes so far, but generally there is less tribal knowledge within a DOD program than in most commercial engineering efforts.
> IT has to maintain archaic devices, operating systems, and tools
In general, IT doesn't go anywhere near the software and equipment used to program, test, and troubleshoot hardware and embedded software. Those pieces are owned by specific engineering teams on the program.
> Obviously that only goes so far, but generally there is less tribal knowledge within a DOD program than in most commercial engineering efforts.
Within a well run DoD program. Most of the larger programs are not well run, in my personal experience. And DoD did this silly thing where it let contractors keep the rights to their engineering data until relatively recently, meaning that US taxpayer dollars are funding the purchase of hardware the government by definition can't understand, because the government isn't allowed to know about or look at much of the engineering data. So there's a HUGE amount of tribal knowledge in some of these programs because there's nothing else to reference.
However, practical experience suggests that second-system syndrome is very real, and the cost of replacing 'obsolete' systems is often vastly higher than just maintaining them, even taking into account the need to train people in skills that are difficult to hire for.
But there's an entire industry which exists to soothsay away these concerns and get companies to upgrade, whether or not it really makes any sense. Nobody has an interest in encouraging companies to keep using their old gear and do internal training; many companies have an interest in encouraging upgrades.
