[jiggawatts]

I've been the one selling the "$1,280 mug", not in America, and not to the military, but to state and federal governments all over the place.

It's always the same problem: They write "requirements" that end up being total nonsense, they have an unlimited budget, and they're terrified that they'll get "in trouble" for some slight oversight. This is a recipe for overspending, and is the bane of all such organisations everywhere.

The reason that DOGE had a snowball's chance in hell of fixing government overspend is that this can't possibly be achieved by merely cancelling a few hundred contracts out of millions!

The dynamic has to change, by realigning incentives and changing the rules, but DOGE did not have that power.

Not to mention that nobody knows how to do this at the scale of the US government! Nobody. I don't have the answers, Elon doesn't, neither does anyone else like Peter Thiel.

They keep talking about how the government is bad, but they don't have an alternative that wouldn't be subject to the exact same forces and produce an equally bad (or even identical) outcome.

[Jtsummers]

One thing that would help, but only help, not solve, is to train the people writing requirements. I've seen so much overfitting. "We developed on a Dell 1234ABC, so that's what we need 200 of when we deliver this to the field." That's not how computers work, but that's how they end up writing requirements. That can even make it into the TO for systems so now they have a drawing of the back of a Dell 1234ABC and the front, showing how it's installed at a desk and cabled up.

Once that happens, if the system lasts more than a year, they have to start sourcing Dell 1234ABCs with the same specs. However, that's an item that's no longer sold. So then they switch to maintaining the ones they have, which means a support contractor is hired to staff locations to handle these repairs (because the local IT staff is already responsible for a lot of things, and maintaining obsolete hardware is not their priority). When what's needed is any computer with X GB of RAM, X GB (or TB these days) of storage, and so on. Set the minimum specs, go acquire it from whatever vendor, and move on. It'd cost a fraction of the amount of that multi-million support contract whose entire job is to maintain obsolete computers.

[zbentley]

That would help a tiny amount. The bigger problem, which GP alluded to and which is very, very frustrating to entangle, is the incentives around accountability. Pahlka’s writing puts it better than I could:

https://www.eatingpolicy.com/p/the-water-is-a-mirror

https://www.niskanencenter.org/culture-eats-policy/

[jiggawatts]

Precisely right.

Adherence to internal procedure becomes ever more important as organisations grow larger, eventually becoming by far the most critical requirement for all work, internal or external. Cost, efficacy, customer happiness, etc... become distant secondary requirements, dwarfed by the mountains of procedure, policy, and paperwork.

[lmm]

We get the project management we pay for. You can outsource implementation but you can't outsource accountability; ultimately, the only way to get effective government is to build up project management expertise in-house, and to do that you need to be willing to match the pay and conditions (including but not limited to reliable long-term employment) that skilled project managers could obtain in private industry.

[malcolmgreaves]

Their goal wasn’t to make the government better. It was to destroy it and steal data so that Trump and Musk could get richer.