[alistairSH]

I never understood why there aren't performance requirements in government contracts. Same thing with infrastructure construction here in DC... new offices go up in a year, but changes to the intersection in front of said office take 5 years and cost multiple of the office itself. [hyperbole, but only just]

[spamizbad]

There used to be a lot more defense contractors during the cold war. Also, more accountability. So there was competitive pressure to be seen as an organization that could reliably deliver projects. Over the years, many of the major companies merged, then shifted strategies to geographically dispersing their jobs to bolster congressional support.

Take the Joint Strike Fighter / F-35 for example: 4 bids. Lockheed won the contract but subcontracts major parts to another JSF bidder (Northrop) and Boeing/MD ending up merging. The whole thing is so incestuous.

[perl4ever]

What I find humorous is a few years ago I observed apparently the federal government wanted to outsource some legal services and document management, and they went with one of these airplane makers (I forget which one). Who do you think of when you think of paperwork? The aerospace industry!

[chii]

"Most of our paper work is already for dealing with these contracting firms! Might as well outsource the paper work completely to them!"

[jariel]

"There used to be a lot more defense contractors during the cold war. Also, more accountability. "

Gear today is far more complicated. Schedule and cost increase geometrically with complexity.

Many of the 'innovations' on early era fighters were quick iterations made by tiny teams.

But it's no longer the 100's of fighters that will win, it's the one squadron with the best gear, radar, comms backed up by the rest of it (AWACS etc.) that will dominate. Of course, it has to actually work (!) but there's something to be said for that.

The basis of the 'over-budget makes money' is still reality however, there's no doubting that.

So it's going to be a matter of how we apply operational integrity in this new era of sophistication.

[spamizbad]

We have powerful computer software today that lets a single CAD engineer replace a room full of draftsman. We can spin up 20 different virtual designs then test them in virtual wind tunnels before a single physical mockup is made. Material science advances also can make your job easier in areas like friction and keeping weight down.

So while I agree things are orders of magnitude more complex, our tools are also much more powerful.

JSF studies started in 1993, when the fastest computer was a 235.8 GFLOPS behemoth -- roughly the speed of a midrange gaming GPU today.

[ska]

It's worth drawing a distinction between increasing costs and schedule complexity, and increasing overruns in both.

Regardless of how complex or expensive a project it, if the people doing it aren't strongly incentivized (positively or negatively) to keep these things under control, the will not. That doesn't matter what era you are talking about.

[ceejayoz]

The argument is that something the size and complexity of the JWST has never been attempted, which leaves a lot of Rumsfeld's famous "unknown unknowns" it's unfair to saddle the contractor with. There really needs to be a middle-ground on cost-plus contracts; something that allows for unanticipated issues to be dealt with, but punishes shoddy work or insanely silly lowball bids.

https://www.seattletimes.com/business/boeing-aerospace/shodd...

> In a vibration test of the telescope earlier this year in California by prime contractor Northrop Grumman, dozens of loose fasteners — some 70 pieces in all — came off. A few pieces are still missing and could well be inside the observatory. The locknuts were not tightened properly before the test, according to a report by the board.

This sort of thing (IIRC, this individual issue cost $150M) should come directly out of the contractor's profits.

[alistairSH]

I see your point for true research programs. But, we see similar patterns all the way down to building subways stations - massive overruns and delays and the taxpayer stuck with the bill. Building an above-grade rail should be a mostly solved problem. Just now, in DC metro, we have the Silver line expansion to Dulles - crap concrete is causing massive delays, but instead of forcing the contractor to replace it, the solution is just "eh, we'll just inspect it more often and hope it doesn't fall on anybody's head." And that doesn't solve the delay - it's still years overdue.

[coliveira]

The answer is the sunken cost fallacy. A government institution will view the sunken cost as too big, and solve that problem by throwing more money at it. The same thing that happens in Wall Street when they get into big problems.

[perl4ever]

Without actual knowledge, I'm certain your analysis is wrong, because it's too anthropomorphic. An institution isn't a person, and isn't subject to common fallacies, for better or for worse. Even if it appears to do things wrong that a person would do, the reasons aren't the same. You may say that there is a boss who is in control and to whom the explanation applies, but institutions don't work like that - they would never last if the leaders completely controlled them.

[scarejunba]

Organizations are even more susceptible to making poor decisions because they don't have information about the other participants. Take, for instance, a choice you and I have to make and where the best choice is either the one that gets you all the benefit or one that gets me all the benefit. e.g. a project that could be done by either of our teams. Instead of doing it that way, we will both choose the third (and worst) option: both our teams do the work and split the work and coordinate in meetings. This way we both get credit. This way the organization gets the worst possible outcome.

Now take the same thing but reward any attackable situation. What you get if you cancel the big losing project: You just threw out $100 million of work. The first mistake you make, I will capitalize on that to say why I should be director instead.

[fgonzag]

I'm starting to believe it'd be cheaper to just start out the same project with 2 or 3 companies, and then just keep the be st performing one and dump the other two as soon as its clear that one will be finishing the project in time and within budget. That would only double the cost (if you dump early enough) of a govt project, and since most projects seem to have a budget overrun much larger than that, it'd actually end up cheaper.

Companies with no competition and a guaranteed contract will never deliver the goods.

[pavon]

That is exactly what NASA did for Commercial Crew. The started out awarding 5 contracts, then at each stage down-selected to 4,3, and finally 2 suppliers, Boeing and SpaceX. Also commercial crew contracts are fixed-price, pay-for-performance, not the standard cost-plus contracts. So Boeing won't get a dollar to repeat this test; they have to pay for it out of their own pocket.

[ceejayoz]

> Also commercial crew contracts are fixed-price, pay-for-performance, not the standard cost-plus contracts. So Boeing won't get a dollar to repeat this test; they have to pay for it out of their own pocket.

LOL. Boeing knows how to handle minor obstacles like that.

https://spacenews.com/nasa-inspector-general-criticizes-addi...

"NASA paid Boeing nearly $300 million more than originally planned in its commercial crew contract in part because of agency concerns that the company might drop out of the program, a new report claims."

[njarboe]

SpaceX also has a contract with NASA to build a rocket and spaceship for astronauts to get to the space station. It has passed its last big test recently. If there are no further delays, NASA will be using it in a few months.

[wmf]

I've seen that done with DARPA contracts but I wonder if it just incentivizes contractors to back-load all the overages.

[wmf]

So if the contractor runs out of money and goes bankrupt when the project is 90% complete, then what? Spend far more money with a different contractor to try to pick up the pieces?

[ceejayoz]

For starters, don't give contractors projects so large they're an existential threat to the company. Northrop Grumman makes billions in profit every year, and the JWST has been in the works since 2003. They (and their insurance) could afford to pay for the fuckups.

When a bank collapses, the FDIC steps in, takes over operations, and ensures it keeps running while they find a different bank to take it over.

It's a rather fascinating process: https://www.npr.org/templates/story/story.php?storyId=102384...