7 ms·

I feel the same way, and I'm reminded of a quote from Elon Musk recently when discussing Starship's Raptor engine: "It's 10-100x more difficult to design the p

by holler 5y ago

I feel the same way, and I'm reminded of a quote from Elon Musk recently when discussing Starship's Raptor engine: "It's 10-100x more difficult to design the production system than the engine itself" So what we really need is to invest and innovate in the production of complex telescopes/probes/etc such that it doesn't cost $10 Billion and 20 years to build, and the loss of one wouldn't hinder the overall production. https://youtu.be/t705r8ICkRw?t=238 https://youtu.be/t705r8ICkRw?t=238

stopping 5y ago
The cost doesn't come from a lack of production efficiency. A significant fraction of the cost is for testing/validation of the spacecraft, which must be done for every unit you produce. Simulating operating conditions on Earth is extremely expensive and time consuming.
- aero-glide2 5y ago
  When you build 20 scopes you can use the fail fast approach to better iterate on the next versions. It's okay if the first few fail.
  - petschge 5y ago
    Except after the second failure you have a REALLY hard time to get funding out of politicians...
    - loonster 5y ago
      Its about setting proper expectations. Its not a failure if its expected. Get the politicians to commit to all 20 launches in advanced.
      - rizky05 5y ago
        How do you know it only needs 20 launches ?
- holler 5y ago
  It's the same reason SpaceX is eating Boeing's lunch with their space program. One uses an iterative build process, and the other uses the more historical waterfall method. That's not to say waterfall didn't or doesn't have merit here, just that it seems like technology has advanced to a place where hopefully in the future we can build these things much faster and cheaper, such that the stakes aren't so high for a single unit.
  - stopping 5y ago
    I don't disagree with your point re: Boeing v SpaceX, but precision scientific instrumentation is a very, very different domain where "move fast and break things" cannot be naively applied. The sensitivity to failure is so ridiculously high that you need 6 to 7 sigmas of reliability for each of ten thousand critical components in order to even have just a 90% chance of mission success. Something as simple as an instrument bring slightly out of tolerance, faulty rad-hardening on a CPU, the sun shield having a minor tear, are all enough to completely jeopardize the entire mission. As a fully integrated system, JWST is at least an order of magnitude more complex than Starship, whose main design goals (launch, orbit, land, don't explode) tend to favor a more iterative approach.
    - gfodor 5y ago
      Pretty sure this argument could have been levied against rocketry. For scientific telescopes most likely a frame shift would be needed to break out of this mindset that would be analogous but distinct from the one applied to rocketry.
      - stopping 5y ago
        There are two key differences between rockets and space telescopes (any space-faring probe, really) which prohibits "iterate to failure" as a development technique: 1. Forensics difficulty. You need extensive data to debug issues for complex systems. You can collect so much more data from terrestrial testing, which allows you to do the extensive forensic analysis required to achieve the required component reliability. Once you put a telescope in space, you can't inspect it anymore. A lot of the sensitive components which might fail on JWST have to be inspected to microscopic precision in order to perform adequate failure analysis. 2. Design requirements are far, far more precise. Many failures in deep space are effectively impossible to correct in later iterations due to point 1. Telemetry and sensor data is enough to debug rockets, but for JWST you would need to ship so much extra data/sensor infrastructure alongside the telescope that the whole project becomes recursively intractable. The system complexity is so high that you absolutely must have the ability to make arbitrary system corrections because the chance of "building everything correctly the first time" is effectively zero, even with perfect hindsight! Essentially, any time you build the thing from scratch, you will always find mission-ending statistical deviations. The objective of on-ground testing is to identify and correct those specific deviations, until the whole system is within design tolerances. If you were to totally rebuild it, the next iteration will have a totally different set of statistical deviations which will need to be corrected. This process of development involves "hardening" the entire system through extensive testing, because it impossible to build a fully-hardened system to start with, even after "learning" from previous attempts.
- gpm 5y ago
  If you can produce them cheaply but not test them cheaply, the solution is really simple, just don't test them. Send them to space, see what happens. It's not like space telescopes need any of the valuable orbit space, and even if they did it's easy to put thrusters on board that can de-orbit them or move them to a graveyard orbit (depending on their operational orbit) in the case of failure.
  - stopping 5y ago
    I addressed this in another comment: https://news.ycombinator.com/item?id=28452001 https://news.ycombinator.com/item?id=28452001 To summarize, you have to have the ability to correct statistical aberrations after manufacture via extensive validation and testing, as opposed to design aberrations which can largely be eliminated through destructive iteration. Without the former process, you'll launch a million telescopes and not a single one will work.
marcodiego 5y ago
This is not a device that will be produced in an assembly line. Efforts to improve its production would be nullified by the fact that every unit is itself a prototype.
- BobbyJo 5y ago
  I wonder about that. Hubble, after all, could have been looked at the same way from the outside, but it was essentially a proof of concept for a constellation of top secret satellites still in use today.
  - anonymousiam 5y ago
    https://www.americaspace.com/2012/06/06/top-secret-kh-11-spysat-design-revealed-by-nros-twin-telescope-gift-to-nasa/ https://www.americaspace.com/2012/06/06/top-secret-kh-11-spy...
  - asmithmd1 5y ago
    The opposite is true. Hubble is a stripped down version of a Keyhole spy satellite that was already in production. One of the major repairs (replace solar panels) was required because of long duration exposures used in astronomy that are not used in spying. Remember that Hubble was essentially useless until heroically repaired the first time by Story Musgrave. Many subsequent repairs were required to get the best pictures from it - like the deep field, long duration picture.
- devnulll 5y ago
  There's another way to think of this. Let's use Starlink as an example, which is ~1300 today growing to 40k satellites in LEO. If each of these had a high-grade commercial camera lens pointed towards earth, you could have real-time viewing of the entire planet with just a little bit of math. With a bit more fun (and for each click of optics improvements), you could "Bullet Time" view a great number of things... If each of these had a high-grade commercial lens pointing outwards, what could that be used to capture? At scale, it seems like doing diff's against all the images would reveal a huge number of previously unidentified asteroids and other near-earth objects. What if you put "very good" optics on them? Or a radio antenna pointed out? Could you make earth sized optical or radio telescopes? Seems like you could. It's increasingly looking like the assumption each one needs to drive the state of the art, and each one needs to be "space grade at 100x the cost" is a fallacy simply because launching was historically so expensive.
AdamJacobMuller 5y ago
I largely agree, even though you're being downvoted. What we really need is to standardize as much of the telescope as possible. Communications? Don't need to be bespoke per satellite. Guidance/Navigation? Doesn't need to be bespoke. Ultimately only the actual optical payload really needs to be bespoke. Commonality of components and design can drive down prices and reduce time-to-launch.
- Rebelgecko 5y ago
  The problem is that if you can't really just standardize every subsystem unless you're willing to give compromise your mission in the name of a one-sizw-fits-all approach. Presumably the people who built JWST passed on of-the-shelf busses for a reason. Imagine if you to tried to force JWST's bus onto every satellite: people who don't care about pointing accuracy end up spending zillions of dollars so they can have fraction of an arc-second precision satellites that are in Earth orbit have different thermal needs than one in L2 satellites that are less weight sensitive waste money on carbon fiber an instrument that needs to be cooled down to 50K doesn't need a JWST-tier cryocooler, etc. You could try having a range of things, e.g. small, medium, and large bus with a handful of interchangeable subsystems. But you're still asking designers to sacrifice their precious size, weight & power requirements

by holler 5y ago

stopping 5y ago

The cost doesn't come from a lack of production efficiency. A significant fraction of the cost is for testing/validation of the spacecraft, which must be done for every unit you produce. Simulating operating conditions on Earth is extremely expensive and time consuming.

aero-glide2 5y ago
When you build 20 scopes you can use the fail fast approach to better iterate on the next versions. It's okay if the first few fail.
- petschge 5y ago
  Except after the second failure you have a REALLY hard time to get funding out of politicians...
  - loonster 5y ago
    Its about setting proper expectations. Its not a failure if its expected. Get the politicians to commit to all 20 launches in advanced.
    - rizky05 5y ago
      How do you know it only needs 20 launches ?
holler 5y ago
It's the same reason SpaceX is eating Boeing's lunch with their space program. One uses an iterative build process, and the other uses the more historical waterfall method. That's not to say waterfall didn't or doesn't have merit here, just that it seems like technology has advanced to a place where hopefully in the future we can build these things much faster and cheaper, such that the stakes aren't so high for a single unit.
- stopping 5y ago
  I don't disagree with your point re: Boeing v SpaceX, but precision scientific instrumentation is a very, very different domain where "move fast and break things" cannot be naively applied. The sensitivity to failure is so ridiculously high that you need 6 to 7 sigmas of reliability for each of ten thousand critical components in order to even have just a 90% chance of mission success. Something as simple as an instrument bring slightly out of tolerance, faulty rad-hardening on a CPU, the sun shield having a minor tear, are all enough to completely jeopardize the entire mission. As a fully integrated system, JWST is at least an order of magnitude more complex than Starship, whose main design goals (launch, orbit, land, don't explode) tend to favor a more iterative approach.
  - gfodor 5y ago
    Pretty sure this argument could have been levied against rocketry. For scientific telescopes most likely a frame shift would be needed to break out of this mindset that would be analogous but distinct from the one applied to rocketry.
    - stopping 5y ago
      There are two key differences between rockets and space telescopes (any space-faring probe, really) which prohibits "iterate to failure" as a development technique: 1. Forensics difficulty. You need extensive data to debug issues for complex systems. You can collect so much more data from terrestrial testing, which allows you to do the extensive forensic analysis required to achieve the required component reliability. Once you put a telescope in space, you can't inspect it anymore. A lot of the sensitive components which might fail on JWST have to be inspected to microscopic precision in order to perform adequate failure analysis. 2. Design requirements are far, far more precise. Many failures in deep space are effectively impossible to correct in later iterations due to point 1. Telemetry and sensor data is enough to debug rockets, but for JWST you would need to ship so much extra data/sensor infrastructure alongside the telescope that the whole project becomes recursively intractable. The system complexity is so high that you absolutely must have the ability to make arbitrary system corrections because the chance of "building everything correctly the first time" is effectively zero, even with perfect hindsight! Essentially, any time you build the thing from scratch, you will always find mission-ending statistical deviations. The objective of on-ground testing is to identify and correct those specific deviations, until the whole system is within design tolerances. If you were to totally rebuild it, the next iteration will have a totally different set of statistical deviations which will need to be corrected. This process of development involves "hardening" the entire system through extensive testing, because it impossible to build a fully-hardened system to start with, even after "learning" from previous attempts.
gpm 5y ago
If you can produce them cheaply but not test them cheaply, the solution is really simple, just don't test them. Send them to space, see what happens. It's not like space telescopes need any of the valuable orbit space, and even if they did it's easy to put thrusters on board that can de-orbit them or move them to a graveyard orbit (depending on their operational orbit) in the case of failure.
- stopping 5y ago
  I addressed this in another comment: https://news.ycombinator.com/item?id=28452001 https://news.ycombinator.com/item?id=28452001 To summarize, you have to have the ability to correct statistical aberrations after manufacture via extensive validation and testing, as opposed to design aberrations which can largely be eliminated through destructive iteration. Without the former process, you'll launch a million telescopes and not a single one will work.

marcodiego 5y ago

This is not a device that will be produced in an assembly line. Efforts to improve its production would be nullified by the fact that every unit is itself a prototype.

BobbyJo 5y ago
I wonder about that. Hubble, after all, could have been looked at the same way from the outside, but it was essentially a proof of concept for a constellation of top secret satellites still in use today.
- anonymousiam 5y ago
  https://www.americaspace.com/2012/06/06/top-secret-kh-11-spysat-design-revealed-by-nros-twin-telescope-gift-to-nasa/ https://www.americaspace.com/2012/06/06/top-secret-kh-11-spy...
- asmithmd1 5y ago
  The opposite is true. Hubble is a stripped down version of a Keyhole spy satellite that was already in production. One of the major repairs (replace solar panels) was required because of long duration exposures used in astronomy that are not used in spying. Remember that Hubble was essentially useless until heroically repaired the first time by Story Musgrave. Many subsequent repairs were required to get the best pictures from it - like the deep field, long duration picture.
devnulll 5y ago
There's another way to think of this. Let's use Starlink as an example, which is ~1300 today growing to 40k satellites in LEO. If each of these had a high-grade commercial camera lens pointed towards earth, you could have real-time viewing of the entire planet with just a little bit of math. With a bit more fun (and for each click of optics improvements), you could "Bullet Time" view a great number of things... If each of these had a high-grade commercial lens pointing outwards, what could that be used to capture? At scale, it seems like doing diff's against all the images would reveal a huge number of previously unidentified asteroids and other near-earth objects. What if you put "very good" optics on them? Or a radio antenna pointed out? Could you make earth sized optical or radio telescopes? Seems like you could. It's increasingly looking like the assumption each one needs to drive the state of the art, and each one needs to be "space grade at 100x the cost" is a fallacy simply because launching was historically so expensive.

AdamJacobMuller 5y ago

I largely agree, even though you're being downvoted. What we really need is to standardize as much of the telescope as possible. Communications? Don't need to be bespoke per satellite. Guidance/Navigation? Doesn't need to be bespoke. Ultimately only the actual optical payload really needs to be bespoke. Commonality of components and design can drive down prices and reduce time-to-launch.

Rebelgecko 5y ago
The problem is that if you can't really just standardize every subsystem unless you're willing to give compromise your mission in the name of a one-sizw-fits-all approach. Presumably the people who built JWST passed on of-the-shelf busses for a reason. Imagine if you to tried to force JWST's bus onto every satellite: people who don't care about pointing accuracy end up spending zillions of dollars so they can have fraction of an arc-second precision satellites that are in Earth orbit have different thermal needs than one in L2 satellites that are less weight sensitive waste money on carbon fiber an instrument that needs to be cooled down to 50K doesn't need a JWST-tier cryocooler, etc. You could try having a range of things, e.g. small, medium, and large bus with a handful of interchangeable subsystems. But you're still asking designers to sacrifice their precious size, weight & power requirements