[wslack]

I suggest the book "Failure is not an option" by Gene Kranz, an Apollo flight director (played by Ed Harris in Apollo 13). He describes how the primary work of flight controllers in all missions is risk management. You are constantly balancing mission needs, fuel needs, mass needs, temperature needs, and etc etc.

I don't think that a raw metric of the number of SPOF is the right way to measure the risk of this spacecraft. It's a fun term for PR purposes (and emphasizing the risk here) but the actual risk posture is more complex.

I imagine that in the course of developing this, they worked out a possible strategy without all of those SPOF - but doing so doesn't eliminate the risk, and the impact to mission is likely massive.

[onphonenow]

What I don't understand is for $11 billion - can't we do 3 $3B space telescopes that work reasonably well? Spread risk? More science?

Let's say a Falcon 9 launch is $90M. Falcon heavy let's say $200M.

So you take your 3x $3B. Put $200M/instrument into launch, have $2.8B per telescope leftover.

There just seems to be something wrong that it costs THIS much to build a telescope.

That said, the Thirty Meter Telescope is also a sort of "forever" job, the delays have stretched on and on.

I wonder if you did something like bid out and paid just on performance instead of this forever cost reimbursement thing. Right now if you can get onto one of these mega projects, and can stretch it out with delays, it basically can cover your career (ie, 20 year projects).

[sjburt]

There are many people in NASA and around that have made similar arguments.

I think the real reason that they never have a lot of traction, sadly, is that if you propose 3, Congress will give you 2. And then when 2 are over budget, it will get trimmed to one. Better to propose one big mission and get it to the point where it can’t be cut easily.

[dwohnitmok]

Yeah when you have a fickle funding source often the only way to get any long-term project done is to make it too big to fail, but then you see all the other terrible waste, corruption, etc. that accompanies that.

[azalemeth]

I'd bet that the vast majority of that cost isn't going to be materials -- it'll be staff time to design and optimise the telescope and make something that can work, including where necessary how to make new materials or processes to make that telescope. I'm not saying the actual hardware is cheap by any stretch of the imagination, but making three different telescopes isn't a linear function of that budget. They're literally pushing the envelope of what's possible here. If something terrible happens to the rocket, lots of forms will be written and people will be sad but fundamentally I think they'll build something a bit better on a few years and nail it. A bit like New Horizons (awesome Mars rover) vs Beagle 2 (awesome Mars rover that died on arrival to Mars).

[captn3m0]

The seemingly costly part of JWST - The Gold Mirrors are cheaper than one might guess, for eg. it’s a 0.1 micron coating that only used 45g of gold.

[picture]

The material cost for a lot of "high technology" things can be very small, but the process is expensive. For example fancy chips are only some x grams of silicon, and expensive inverted mesa quartz crystals might cost a good bit while only being x microns thick piece of quartz..

[lamontcg]

JWST is driven by the size of the mirror, the need for an IR instrument and the temperature you need to keep the instrument at to do the observations. Spamming a bunch of hubble-sized instruments up into orbit won't accomplish the same thing. And I don't think you can do space VLBI in the optical/IR or it would have been done already (but I don't know why?).

It is kind of like asking why 5 Ford Rangers can't replace one Lamborghini or something.

[petschge]

The reason why we are barely getting started into VLBI in the IR on the ground (and nothing that I have heard of yet in space) is that the different apertures need to be stable relative to each other with a precision better than a small fraction of a wavelength. (One tenth, one twelves and one twenties are often used, depending basically on which performance drop relative to the theoretical optimum you are willing to live with.)

For radio astronomy, where we do VLBI everyday, we have to handle waves of wavelength 1 cm and position antennas to a precision better than a millimeter. Not easy when the antennas are scattered across the country, but something we can pull off.

For IR astronomy we are talking wavelength in the range of 1000 nanometers to 30 microns. So at the easiest end of the spectrum you would have to position satellites to a precision better than 3 microns relative to each other, while flying on orbit and being pulled and pushed by tidal forces, gradients in the graviational fields and solar wind pressure (which contains turbulent fluctuations). For it to actually work in near IR you would have to get the positioning right to within 100nm.

For comparison: The mirrors of JWST itself are flat to within about 25nm. And in some sense we ARE doing IR VLBI with JWST since we have separate mirror segments that we all position correctly relatively to each other. But doing so we separate freeflying satellites is something we just aren't capable of yet.

PS: Yes, LISA Pathfinder has demonstrated measurements of spacecraft separation down to a few picometer, so we are slowly getting there.

[lamontcg]

That's kinda what I guessed. So on the ground you can measure the distance between two observatories that are a few hundred meters apart with incredibly high precision and track that over time as the ground under them slowly deforms and moves.

[petschge]

Geodesy is indeed one of the side products of VLBI observations on Earth.

[CamperBob2]

Spamming a bunch of hubble-sized instruments up into orbit won't accomplish the same thing.

It will, and then some, once we get optical interferometry nailed down. JWST is great, and needed to be done, and I'm glad it's finally getting deployed. But if I had an argument to make against it, it would be, "Let's wait until we know how to do this properly. We're not there yet."

That's a weak argument and should almost never be heeded, but it's also not wrong.

[lamontcg]

> That's a weak argument and should almost never be heeded

Well that's an incredibly strong opinion with almost nothing to back it up.

Maybe it all goes boom tomorrow on launch or the deployment is ultimately completely FUBAR'd and you can feel satisfied with yourself, but I don't buy that your principle is what we should always follow.

Hopefully we start getting data from JWST in a few months and then I invite the scientific community to figure out how to spamcraft optical and IR instruments into LEO and achieve VLBI with them in the future, but we won't have to keep holding our breath for the technical breakthroughs there.

And even as a jobs program and Keynesian stimulus the JWST beats digging holes and filling them up again, even if it all goes boom. Better than building yet another weapons program as well. And the Senate just tacked on an extra $23B to the 2022 Defense Appropriations Act for one year of spending. The $10B that JWST cost over 25 years of the program doesn't really matter at all.

[CamperBob2]

True, I don't disagree with any of that, but:

Maybe it all goes boom tomorrow on launch or the deployment is ultimately completely FUBAR'd and you can feel satisfied with yourself

A more charitable interpretation of my post would read something like this: "If it does go boom, we can either start building a new one now and try again in 20 years, or we can wait 15 years and build multiple better, cheaper ones in 5 more. Meanwhile, maybe a few hundred other researchers can get funded with the same money."

[aargh_aargh]

The reason we can't achieve the same resolution as JWST with interferometry has been mentioned in a recent startalk podcast. The distance between the telescopes would have to be coordinated to a precision somewhere on the order of nanometers.

[CamperBob2]

The distance between the telescopes would have to be coordinated to a precision somewhere on the order of nanometers.

That'll happen. It's fundamentally a timing problem, and we can already build clocks that will tell you what floor of the building they're on.

Less hand-wavingly, it's fundamentally a data-acquisition and correlation problem of the sort that was solved long ago for microwave VLBI. Back in the day, the individual stations had their own maser clocks, but now I imagine they're all GNSS-based. It is by no means trivial to go from RF interferometry to optical, or to move the antenna elements from earth to space, but that's the basic approach that will ultimately be used.

[foobarian]

I think this is because of how the various constraints work against each other. We need the telescope to be as large as possible to work as well as possible. But the rockets can only be so big (i.e. we're constrained by the largest available booster). The only thing here under our control is how hard we work on fitting the biggest possible instrument in the available payload envelope, and that is exactly what happened.

With a machine this complex I think it's also not easy to crank out multiple copies since I'm imagining most of it is made by hand without the benefit of a production line.

[rtsil]

JWST was (re)launched in 2005, the very idea of cheap spaceflight didn't exist back then, so you'd have ended up with three expensive flights. Also consider that the expected cost back then was $4.5 billion.

[pkaye]

To make an infrared telescope significantly better they need it in an orbit not facing the sun always. But other types of lower cost projects might be feasible.

[mongol]

Perhaps that is what will happen from now on. But when JWST was initiated, the options looked different.