Hacker News Reader

Comment by cromwellian | Hacker News Reader

I really fear for this project. Because if it blows up on launch, misses an orbit, or has a failure to deploy, or some flaw, basically decades have been lost.

As another commentor said, the marginal cost of these things must be must cheaper. Why aren't they just launching new iterations like every year or so. Rather than starting from scratch after the HST, there should have been an HST 2.0, HST 3.0, a whole bunch of marginal iterations. This would create an efficient pipeline around producing many of these one-off specialized parts.

Building the mirrors would become a pipeline of incremental improvements instead of 2-decade redesigns.

I feel like if Elon Musk ran this project, he'd have built 5 different versions already, 3 of them would have failed, we would have learned something useful, and two of them would be in space delivering 2x the science.

View on HN

There _are_ incremental improvements, in the form of 5 gamma ray, 6 x-ray, 5 UV, and 4 visible satellites. Actually, there are more than that since Hubble went up.

Unfortunately there are real physical limits to visible light telescopes, and there are not many ways to get around them without drastically increasing the size of the lens/mirrors, and drastic increases in size and weight are not something that plays well with our space capabilities.

I know it seems ridiculous from the outside, but do you really think thousands and thousands of astronomers, physicists, etc., wouldn't be exploring every single possible way to improve visible light telescopes? That the idea of "incremental improvement" never occurred to anyone? They are using the current approach because it is the only approach, at least for the time being. Perhaps in 10, 20, 30 years we'll be able to launch hundreds of smaller telescopes that can achiever a higher effective resolution, but, again, it's pretty damned hard.

All true. On the other hand, the current JWST is way less capable, and way, way more expensive, than what was initially planned. I think there's a very legitimate argument to be made that maybe, with hindsight, it would have been better not to put all that money into the very large bucket that it's ended up being.

It seems the way this goes is "we don't have money for more than one mission, so we have to make it count." But now you only have one mission, so you have to make really sure it doesn't fail, and the cost starts running away.

Also, after JWST started, NASA has pretty much completely changed their strategy away from grand missions into many smaller ones with the "faster, better, cheaper" motto.

> It seems the way this goes is "we don't have money for more than one mission, so we have to make it count." But now you only have one mission, so you have to make really sure it doesn't fail, and the cost starts running away.

This is recognized by NASA - in particular, I've recently seen one report[0] with a diagram explicitly mentioning this problem. It was called "The Tyranny of Space Transportation Costs", and depicts a spiral with nodes like "High Launch and Space Transport Costs" -> "Infrequent Launch" -> "Inability to Service" -> "Over Engineering" -> "Cost Growth" -> "Need for Extreme Reliability" -> "Higher Launch Costs" -> ...

That's one of the reasons NASA is interested in in-orbit mining & manufacturing - to enable rapid iteration on future missions. That seems to also be the reason why they've outsourced LEO transportation to private companies.

[0] - "Stepping Stones: Economic Analysis of Space Transportation Supplied From NEO Resources". Found it on HN some time ago. https://www.nasa.gov/sites/default/files/atoms/files/eso_fin... (warning: 140-ish MB); the diagram in question is on page 9.

"Faster, better, cheaper" has not been NASA strategy since the 90's (i.e., pre-JWST). The resulting failures were judged by the public as intolerable.

https://en.m.wikipedia.org/wiki/Daniel_Goldin

There are indeed very successful missions at smaller scales than Hubble/Chandra/JWST, but they are by necessity less powerful and more specialized. There are regular calls for a "balanced program" of large and small programs, but those run into the astropolitical reality that large missions have more "constituents" in science, government, and industry.

I want to believe there are very smart people studying these "strategic" issues, with mathematical models and stuff. That is : beyond the technical aspects, the implementation strategies, science return/cost ratios and whatnot.

NASA used to be good at implementing incremental approaches to projects, weren't they? I mean isn't the Apollo program a perfect example of it? If they seem less incremental in their approach to space observatories, I assume they have good reasons.

Am I being naive?

Am I being naive?

I think so. The Space Shuttle was not incremental at all, it was supposed to be a great leap forward.

I think calling Apollo incremental is also a stretch. They build a rocket that's still the largest the world has ever seen and flew it twice before putting people on it a year later.

I'd say Apollo was pretty incremental. Hell, it was very much "move fast and break things", given how many things they designed and tested in the 12 years between Sputnik and the first Moon landing.

Yes, it was very incremental. They even sent the orbiter around the moon on a mission before even attempting to land. And there were a number of other tests (manned and unmanned), and even ones without an official Apollo designation.

More info: https://nssdc.gsfc.nasa.gov/planetary/lunar/apollo.html

Sorry, but Apollo was a clash of german waterfall with american move fast and breaking things.

https://en.wikipedia.org/wiki/Wernher_von_Braun#NASA_career

I assume to have those two directions as two camps in your company, fighting each other generates the most optimal result. If you fail, the cautious camp wins and carefull analysis and slow methodic itteration become stronger.

If you win, the move fast team takes over and pushes towards new exploration data and scaling up towards new problem areas.

Pre-sputnik and pre-Gagarin, the US approach to space exploration was conservative, and my understanding is that Braun was pushing for a less cautious approach [1]. One issue was apparently a reluctance to lean too heavily on 'Nazi' technology. This caution contrasted with the US' approach to airplane development, which had a lot of moving fast and breaking things (and people.)

Ironically, if von Braun had got what he wanted then, there may have been no Moon race and no Apollo program.

[1] Angle of Attack: Harrison Storms and the Race to the Moon, Mike Gray, ISBN 978-0393325133

Before I get the chance to read the book, can you please explain in short what von Braun wanted that wouldn’t be a Moon race?

Von Braun wanted to send people to the Moon (and he really wanted to get them to Mars), but I am guessing that if the US had adopted his plan to orbit a satellite with a modified Redstone rocket, and done so before Sputnik, followed by the first human orbital flight, there might not have been the impetus for a Moon race, no Apollo, and no Moon landing in 1969 or any time soon after. This speculation, of course, depends partly on how the USSR would have responded to these developments.

Von Brauns push for additional testing prevented the first human in space from beeing american.

"After the flight of Mercury-Redstone 2 in January 1961 experienced a string of problems, von Braun insisted on one more test before the Redstone could be deemed man-rated. His overly cautious nature brought about clashes with other people involved in the program, who argued that MR-2's technical issues were simple and had been resolved shortly after the flight. He overruled them, so a test mission involving a Redstone on a boilerplate capsule was flown successfully in March. Von Braun's stubbornness was blamed for the inability of the U.S. to launch a manned space mission before the Soviet Union, which ended up putting the first man in space the following month.[citation needed]"

I guess the story is more complicated than generalizations will allow. It is rarely obvious when decisions made in good faith will come back to haunt you, as with the decision to give the Apollo command module a sturdy hatch that could not be quickly jettisoned. For everyone saying the Shuttle booster 'O' rings were in danger of failing, there was someone saying that they had been fine so far.

In the case of MR-2, was the schedule to have an orbital flight before what turned out to be Gagarin's day of destiny? The response to Glenn's flight leaves little doubt that, in the popular view, it is orbital flight that counts as being 'in space', and the start of the Moon race might have hung on that perception.

> “Sorry, but..”

So are you refuting that the missions were incremental? Because if you look at the mission outcomes, they certainly were..

The Apollo program cost, according to https://en.wikipedia.org/wiki/Apollo_program , $25 billion in 1973 dollars, over 11 years. According to https://www.bls.gov/data/inflation_calculator.htm that's about $146 billion in 2018 dollars.

That's an annual cost averaging more than the entire program cost of the JWST. Of course Apollo spending wasn't very even over those 11 years; for example in 196 according to https://en.wikipedia.org/wiki/Apollo_program#Costs it cost about $23 billion in 2018 dollars. So triple the entire JWST program cost in that one year.

If we were willing to throw this much money at the problem, sure NASA could build multiple iterations of things quickly, destructively test them to get lots of data, etc, etc.

Could the lens be made in situ from distilled, frozzen again moon-ice and a mirror be crafted from foil? Im sure, this would not be optically optimal, but given good enough camera software, learning about the physical flaws of the system, these flaws could be filtered out?

There's a much simpler plan for a moon-based telescope: ship a big vat of mercury to the moon and spin it. The centrifugal force would curve the mercury perfectly and elemental mercury is a really good mirror on its own.

I found this- but no mention of mercury. https://www.space.com/5628-nasa-envisions-huge-lunar-telesco...

Diffrent solution entirely- dont build a mirror- build a fragment, and rotate it with a counterweight along a axis?

There was one in Canada: https://en.m.wikipedia.org/wiki/Large_Zenith_Telescope

Now, if you had a perfect simulation of fluid dynamics of mercury- you could use piezo-pumps and image correcting software to angle it partially and compose a image at a angle.

Actually- if you do know the fluid dynamics of mecury in openspace, and how to push it where you want it- you can do the same thing up there.

Screw my rotor idea, all you need is a way to influence quecksilber in space, push it and take a series of images - that is recomposed into one giant hiRes picture.

So, one needs a strong magnetic field to push the mercury into shape, a closed tube, able to withstand that magnetic force and a shielded camera. Or you spin the tube, while applying the magnetic force, thus getting the mercury to collect at the side. Also the mirror would be selfrepairing..

Wow, this would be how much weigth? Anyone with a engineering degree and a calculator nearby?

Well, filtering flaws works much better when you've got a strong prior on what you expect to see. Unfortunately, these priors infect what you think you're seeing, which is an even bigger problem when you're hoping to see new, novel things never seen before.

Here's a nice paper on the problem of overfiltering in electron microscopy. They famously found Einstein by aligning images of white noise: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3831464/

These crazy expensive one-off projects are the natural result of trying to spend as little as possible each year while completely ignoring spending in the long term.

I'm unsure why this is being downvoted. I think it's a completely fair assessment given what happened with Hubble. At that time we had the Shuttle to head up and actually fix things that were broken. I have doubts that SLS is going to be as quick as they want it to be, and crewed Dragon isn't operational yet either. By the time this thing launches we might see trials of both, but certainly there won't be something readily available for service operations on the JWT at the time it launches.

JWT is suppose to be at Earth-Solar L2 Lagrange point. Anything that's not a failure in the initial earth orbit would require a repair mission that involves sending astronauts further than we've ever sent them (Earth-Solar L2 is significantly beyond lunar orbit).

Indeed! At minimum we would need the Mars Dragon or Orion with a long range trunk attached to carry supplies and repair pieces. It would be really interesting to see an orbital pairing (a-la the Lunar Module/Command Module) for an L2 repair mission.

L2 is about 4 times further away than the Moon, to be precise.

Also, it's not being made to be serviceable.

That was a feature after HST.

"What Would Elon Musk Do" is a tiresome theme. Mostly because the guy is a total anomaly.

He and his company have accomplished a lot but that does not mean NASA is wrong for running space programs the way they run them. This stuff is complicated, NASA doesn't always get it right but that doesn't mean they need to reinvent themselves in the image of a technocelebrity.

With regards to SLS, I do trust the NASA professionals; they originally argued against the system, and then were forced by direct Congressional mandate (pushed forward by representatives of places building the hardware) to develop the thing anyway.

Appeal to authority is a dumb argument, whether Musk or NASA fans are using it.

>By the time this thing launches we might see trials of both, but certainly there won't be something readily available for service operations on the JWT at the time it launches.

Servicing the JWST with astronauts like we did with Hubble will be impossible, as it will be parked in one of earth's L2 points, not in LEO. Basically, everything has to work absolutely flawlessly on the first try, or that's it. Billions of dollars and decades of work gone.

>I really fear for this project. Because if it blows up on launch, misses an orbit, or has a failure to deploy, or some flaw, basically decades have been lost.

The thought of this literally keeps me up at night. Watching that thing launch is going to be incredibly stressful.

That's possible with a private company as the owner/CEO takes the blame and that goes as a part of profit/loss of some major stakeholder.

Its a lot more harder to do it when tax payers are footing the bill. If you have, say, more than a few failures people would be complaining and asking for less iterations and more of one-time-done-well projects.

Then we should apply that same logic to the military because they don't seem to care about just blowing money up.

NASA has an HST 2.0. Two of them. They're sitting in storage, and they cost NASA absolutely nothing out of its budget to develop.

Than why not launch them? I mean, isn't HST oversubscribed and wouldn't having 3x observing time a net boon for the science community?

I doubt there's any way to easily estimate the cost, but the two HST-superior telescopes NASA was given by the NRO, need to be made science-ready. There are no instruments on them.

Say for example that the cost would be $500m to $1 billion over several years to get them ready, then another $1 or $2 billion over ten years to operate them both. If you're NASA and you have $23 billion to spend per year average over that time, do you invest into the two telescopes and divert resources, or stick with HST until its end and focus on Webb. Obviously if the US Government were behaving more rational, NASA would be given the resources to prep these telescopes and launch them. They're exceptional gear going to waste, surely we could task something interesting for them.

I was unaware we had more, so this is pure speculation, but I would imagine that having continuous coverage is considered critical, so if we launch another one and it experiences a failure soon after the current Hubble fails, we’ve lost that margin of error.

Fundamentally it’s much safer sitting in storage.

How would having the backup in orbit reduce the amount of coverage that an HST-type device could provide?

It seems like having another telescope in orbit could basically be a "hot-spare" for the highest priority tasks while continuing to do lower priory work in the mean time.

Because it's safer on the ground. If you put everything you have up in the sky, and they all fail for various reasons (solar storm, collision, wear & tear) then you're screwed.

Regardless, per another comment it sounds more like a budget problem, so my speculation was crap anyway.

Yeah, I guess I understand that, it just seems like most things that aren't a design failure are unlikely to affect more than one orbiting telescope.

In other words, to return to your example, it seems unlikely to me that a collision would happen to more than on telescope ever, let alone in a time window too tight for a third (or fourth, etc) telescope to be built and launched.

(and, anyway, it looks like you are right: this is all budgetary anyway, so whatever)

Storing a satellite in space is significantly cheaper and not that much riskier than storing it on the ground

You can't repair satellites beyond LEO very easily though, so if your spares fail in spaces, that's it.

If your spare gets damaged in storage on earth, you send the electrician over with a car.

Satellites don't break significantly faster in orbit than in storage. Storage can easily cost $5 million a year, but on top of cost savings there's other benefits to sending your payload up early (assuming you've budgeted for a bit more battery wear, fuel consumption, etc.)

The most dangerous time for a satellite is by far the first couple months. This is when most design and QC failures become apparent. If you launch before you need your satellite, you can do your on-orbit checkouts and not be in a rush to deal with any anomalies. When your preexisting operational satellite dies, you can have one ready to go immediately instead of needing to beg borrow or bribe your way to a better position in a launch provider's manifest.

This is not quite true. These telescopes were built by the military to look at things a few hundred miles away, not millions of miles away, so the optical dimensions are different. They are still useful, but they are not clones of Hubble.

Furthermore, they lack the expensive part: the one-of-a-kind scientific instruments. These will still have to be developed from scratch.

NASA is developing one of them into a telescope called WFIRST, and by some estimates, little to no money will be saved by using the donated military telescopes.

> NASA is developing one of them into a telescope called WFIRST

Yes, the Trump Admin just tried to kill that.

Correct me if I'm wrong, the NRO telescopes are just as powerful and capable as the HST as telescopes, the difference is the field?

Basically, yes. The mirror is the same size as Hubble, and the field is much wider. However, there are some technical differences that make it less useful for imaging extrasolar planets and the most distant galaxies. Details are harder to come by, as this is ex-military hardware, so certain things are still classified.

The NRO instruments are just optics, optics make up a tiny fraction of the cost of a space telescope. NASA was not gifted fully functional space telescopes they could just launch and then start observing. They still need all of the components to actually turn them into functional instruments, spacecraft systems, cameras, attitude control, etc. The estimated cost for WFIRST alone is $3 billion, only slightly less than if the instrument had been built from scratch, if at all.

Reply on news.ycombinator.com