Spreadsheets seem backward to software people because we are confident with manipulating data with more flexible tools. We see them as backward primarily because they are batch-oriented and cannot easily handle distributed writes, elegant types, reflective programming, serialization, automated updates, etc.

However, consider a small to medium scale manufacturer. They have millions of dollars invested in equipment. They have dedicated employees who know that equipment, its settings, its maintenance, and its operation. To those employees, software is a sideshow.

Chances are, for any given manufacturing order there is going to be the following: pricing the order, sending out a quote, taking a downpayment, allotting or acquiring the raw materials, preprocessing, temporary storage of parts, scheduling the machine and operator time against other orders for all unit operations necessary, post-processing, final assembly or QA, consolidating and packaging, actual shipping, settlement.

Can you see how batch-oriented processing actually makes reasonable sense for this sort of scenario? Going real time is quaint but if it doesn't buy you anything and you need to hire "$oftwar€-d€v$" to create and maintain interfaces that still require human input and have less familiarity to your very-busy-operators than a simple spreadsheet - why not use a spreadsheet? It's cheaper, faster, more familiar (read: less errors, ~no training required) and therefore from a business standpoint more effective.