A few defence companies have noticed that failed projects tend to lead to them getting given more money to fix the problem.

They now deliberately make their projects useless (but still technically meeting the contractual requirements) simply so they get the followup contract to make things better.

One project I worked on had 10 times we had deliberately gimped the system and the client had come and paid us more money to ungimp it. (For example, we designed an email client, but "@dhs.gov" was hardcoded in the address field so you couldn't send an email to anywhere else from the system. Then it cost millions of taxpayer money to un-hardcode that, despite it being a 1 line code change)

About software, it is not typical to defense companies and it does not need to be deliberate: most consultancy and subcontracting companies have no incentive to exceed a crappy average level:

Indeed, as you explain, being fast and excellent makes you poorer when you have a day rate and when your next contract depends on the maintenance of the software you create (since you are the only expert).

So in a way, the Peter principle makes money ?

That’s a bit too precise for comfort, if I were you I’d anonymize it a bit more. Civil liability for fraud is a thing.

I changed the hardcoded domain in my example... Either way, it was no secret that or system could only send emails to a specific domain - both we and the customer knew that. And they paid us $XM to give it the ability to send emails to anywhere.

This sounds too blatant to be true, but even so I imagine it's the decision-maker that would be responsible.