So a couple of things are happening here: most important is that the budget for multiple launches and physical spacecraft to do the same mission just isn't there, hasn't been there since the 1970's. Viking and Voyager each got two bites at the apple, but since then NASA has decided that their reliability is good enough to get away with one actual launch. And so the easiest form of reuse is off the table. And then since every mission is different, you end up needing different things for components.
https://en.wikipedia.org/wiki/Mars_Observer is what happens when you try and reuse components without fully taking into account the differences in mission. This was an early 1990's NASA Mars mission which failed because they reused an engine from Earth orbiting satellites which would always fire the engine shortly after getting into space. Probably its fuel tank failed because it was expected to lie dormant for months of travel and then fire. So the environment that HST, JWST, the Parker Solar Probe, and Juno are expected to operate in are enormously different from each other and re-using a component and having it fail because it was designed for a different environment would be really bad.
Oh, and building a new JWST now would be pretty expensive, because a lot of the cost is in testing and certifying that it will do what it's supposed to do, and all of the people who built and tested the JWST have moved on and changed jobs, so all of that workforce would have to be re-created, and that's where the money goes, to pay people. If you wanted to build 8 JWST's at once, the cost wouldn't be that much more than building one JWST, but building a single new one now would be expensive.
This is a fact that is often overlooked in these types of forums. Quality control is an enormous expense for high reliability systems. A bolt doesn’t necessarily cost $1k because it took a lot of R&D to make it. It costs a lot because they have to maintain a chain of custody, certifications, inspections. They have to keep raw material coupons for testing. Sometimes all of them need to be shipped and/or warehoused in tightly controlled environments, with security and so on. Those quality administrative costs are what add up to $1k per bolt.
You can work harder or smarter. I’m not convinced that all the quality control isn’t harder instead of smarter. The proof point is how spacex managed to drastically reduce the cost of a space shuttle launch vs what NASA was able to do. NASA does good work and what they accomplish in their environment is impressive but culturally they seem to have stagnated a bit - not entirely their fault and having a mercurial political situation adds to the distinction but it does indicate that the costs that NASA projects take may be unnecessarily expensive and there may be cheaper ways of accomplishing this stuff. Repeatability in particular is huge in driving down costs and it’s why spacex costs are so low. Unfortunately there’s no money to be made from scientific telescopes and even national security decisions are driven by the US government which often makes incorrect cost/benefit decisions because of politics.
I’m really happy to see them exploring things like the Mars copter to do a super cheap “unreliable” robot that turns out to actually work reliably. I think people seriously overestimate how much has to “go right” to get amazing results and decoupling things from the “must succeed” parts of the mission are ways to bring down that cost by having more non-critical pieces of the project (in addition to reuse). It’s easy to point tot reuse failures one offs but the data is irrefutable that repetition drives down cost and improves reliability. So we should simply learn from the repetition failures and just keep repeating anyway.
I think this ignores some of history. The most glaring example is the strut failure that lost a launch. It is common practice in aerospace to test material coupons for structural specs for critical parts. SpaceX didn't and it cost them a rocket. Luckily it was uncrewed, but it forced them to layer on those quality requirements afterwards. They had to re-learn what much of industry was already doing but I bet if you asked them beforehand, they'd claim those weren't value-added requirements. With low probability events, it can be easy to conflate being lucky with being good.
I will say I think NASA is probably too risk adverse. This gets more into a philosophical debate, but I think that stems more from NASA being as much about national prestige as science. There are many people at NASA who think the engineerinng and quality requirements are too much, but very few people want to stick their neck out and waive them. Maybe a hot take, but I think that's a big part of CCP: it allows private companies to skirt many of those requirements so civil servants don't have to make those decisions. When civil servants have raised the flag on some CCP issues, they've been told it's not their place because NASA is just buying a ride and it's up to the contractor to drive the design. It essentially allows them to get rid of those requirements and outsource responsibility for the increased risk. (This is also a somewhat similar dynamic with JPL that made the Ingenuity copter, although it tends to be less pronouced. JPL is quasi-NASA because it's mainly managed via CalTech)
You say that, but then I counter with Boeing. Their testing and certifying are definitely not very expensive since they self-certify.
There's a large part of any government awarded contract's budget that goes into institutional largess or bureaucratic accounting practices that divert monies from intended purposes. With Boeing and the evidence showing their diversion from standard engineering into more MBA style management, I would not be surprised to learn that a small fraction of government awarded money went to actual engineering with the largest portion going to management.
Are all government contractors as bad as Boeing? I hope not, but it wouldn't be shocking if they came close. But I'm guessing that $1k bolt wasn't strictly from design/testing costs as much as managerial costs being a large factor.
>Their testing and certifying are definitely not very expensive
If you're referring to the 737-Max, that's a different animal because: 1) it's a different division than their aerospace and 2) it was taking advantage of a previously certified airframe. It doesn't mean they got rid of all the upstream quality requirements. Also, you might need to qualify what you mean by "not very expensive" because I don't think they would agree.
If you're talking about Starliner, there is no "certification" process like what the FAA requires. And NASA was heavily involved in the testing, especially after the mishaps with the Demo flight. They (NASA) felt the need to levy on many more requirements after the fact because of the (real or perceived) increase in risk.
>But I'm guessing that $1k bolt wasn't strictly from design/testing costs as much as managerial costs being a large factor.
That's my point. A large part of the cost isn't R&D but administrative. The difference I think we're making is you're considering the "managerial" costs to go to some C-suite, but I'm saying there are lots of quality requirements that would also go into the bucket of "adminstrative" or "managerial" costs. Look at what it costs to ship a flight article: you have to have a quality engineer inspect it and a photographer document it. There's all kinds of tracking documentation. You have to have special packing and labeling. Of course, the shipping itself is more expensive; it may have to have be shipped in a certified/refrigerated/climate-controlled truck and have accelerometers attached to the package to measure g-forces in transit. Then once it's received, you have to have a quality engineer open it, a photographer document it, more tracking paperwork, and a climate controlled, secured warehouse. All that paperwork has to be managed for the duration of the mission (and beyond). And this isn't just for the big items like a assembled satellite; this is often the case for the small parts too. You can get rid of all those requirements and just drop it to the UPS guy, but then you might risk being back in the same place after a mishap, like SpaceX.
https://en.wikipedia.org/wiki/Mars_Observer is what happens when you try and reuse components without fully taking into account the differences in mission. This was an early 1990's NASA Mars mission which failed because they reused an engine from Earth orbiting satellites which would always fire the engine shortly after getting into space. Probably its fuel tank failed because it was expected to lie dormant for months of travel and then fire. So the environment that HST, JWST, the Parker Solar Probe, and Juno are expected to operate in are enormously different from each other and re-using a component and having it fail because it was designed for a different environment would be really bad.
Oh, and building a new JWST now would be pretty expensive, because a lot of the cost is in testing and certifying that it will do what it's supposed to do, and all of the people who built and tested the JWST have moved on and changed jobs, so all of that workforce would have to be re-created, and that's where the money goes, to pay people. If you wanted to build 8 JWST's at once, the cost wouldn't be that much more than building one JWST, but building a single new one now would be expensive.