If the factories are operating with giant iron cog wheels older than your grandmother, then maybe yes, we should.

The world of software is moving a lot faster than the physical world, so things can become outdated pretty quickly. Compare a 15 year old browser to a 15 year old factory.

I think that wasting money on solved problems is, essentially, bonkers.

Sure, it's nice to have the latest and greatest. But I don't think it gives you magical powers. You still need to do all the other stuff.

That, and the well-known problems with rewriting (the track record is notso hotso) mean that really, the onus for making a case to rewrite is on the rewriters.

The difference between a browser and a factory that makes say, steel wheels for cars, is that while the web has changed dramatically in 15 years, the basic requirements for a steel wheel have not. As long as your machinery can hold the required tolerances and make product profitably and while complying with the laws that apply (like environmental), it would be foolish to scrap the old factory.

If the market starts shifting to alloy wheels, and the reduced volume of sales for your steel wheels starts to affect the profitability of your factory, then maybe it makes sense to start looking at investing in updating your process to make steel wheels more cheaply at lower volume or to dump steel wheels entirely and start manufacturing alloys.

The Air Force Heavy Press Program made the rounds on the internet last year when Alcoa refurbished its 50,000 ton forging press in Cleveland [1]. It's been around since 1955, and it's almost unique in its capabilities within the USA. The idea that it should be replaced simply because it's old is, frankly, insane.

That's obviously a pretty extreme example, but the fundamental requirements for a metal lathe or end mill haven't changed in years. Yes, there are multi-axis CNC machines that can do vastly more operations before somebody has to change the setup on them than on a basic lathe or mill. If you aren't making something that requires a lot of steps, the basic lathe or mill might still viable.

Just because the rate of change in requirements over time is huge in software doesn't mean that it's true everywhere. If an old factory can continue to meet the requirements, it's probably hard to justify changing it just to be up to date.

EDIT: Forgot the link

[1] http://www.theatlantic.com/magazine/archive/2012/03/iron-gia... [1]

This attitude strikes me as wasteful. By all means, if a system has shortcomings, and you can afford to build a new system with fewer shortcomings, and migrate to it, then do that; but by itself, "it's old" is not a reason to rewrite something.