A Question about Space Telescopes

I have no idea if this is a stupid question. And that’s OK. There are such things as stupid questions, and pointing that out is not inappropriate.

The JWST is mind-blowing in lots of ways, but my favorite way was the list of ten thousand+ things that had to happen correctly and in sequence for the system to function. Then, a micrometeorite blew a little hole in one of the panels. But the overall system is resilient enough that it’s still functional. I assume that means that the other panels were able to replace the lost pixels from the damaged panel.

I remember a few decades ago, there was a camera built (the “petapixel project”) that had a robotic base and some driver software that automatically panned the camera with great precision, allowing a piece of software to stitch the images together into a single massive image with great apparent resolution. It was a neat trick. I assume that the JWST is doing something similar.

Midjourney AI and mjr: “a mirror bot independent imaging satellite that is part of a cluster which make up a virtual space telescope”

Which brings me to my question: why are the pieces connected? I assume it’s to share common bandwidth and infrastructure, but doing that came at the expense of tremendous size and complexity and a certain risk of failure. What if the panels were separate, independently-operated imaging systems, that communicated with lasers to form a “camera cloud”? Sure, you’d have to duplicate the control systems, communication, power, bus, storage, attitude control, etc., but in return you wouldn’t need to fold the things up, you could just launch a cloud of cameras to a lagrange point and let them nudge themselves into a collective telescope. That way, additional resolution could be brought online by sending up more eyebots. Or dead eyebots could be replaced on the fly. Cluster management is a well-understood software problem dating back to the VAX/VMS clustering systems designed in the 70s: you need to have independent nodes “vote” a cluster director that manages communications, etc. When a node fails, it is dropped from the cluster and the cluster reconstitutes.

There seem to be lots of options. What if the “eyebots” were really minimal and just aimed their mirror at the cluster director, which did image collection? There could be mirror bots and sensor bots and the whole system could reconstitute itself at arbitrary resolutions, depending on how it was partitioned and how many “eyebots” were working together at given time.

Since the distances in space are ridiculously unimaginable, I assume that parallax is not an option, otherwise the same technique could produce a stellar range-finder: everyone collect the angle to a target star on a virtual plane and calculate the range based on the size of the plane being known.

Midjourney AI and mjr: “a fallout 76 eyebot that has been repurposed to serve as a space-based camera array”

And, since I’m exposing my ignorance, let me raise a tangential issue. [I’m going to assume you’ve heard of the Fermi Paradox] What if this is the answer to the Fermi Paradox: there are plenty of intelligent species in our universe – hundreds of them – but civilizations grow to the point where they build a camera-cloud or their equivalent of the Hubble Space Telescope, then they discover that:
a) space is really really big
b) Einstein was right about relativity
c) faster than light travel is impossible because of the energy budget that even the most optimistic science fiction systems consume (the starship Enterprise appears to be powered by matter/antimatter)
d) sublight travel is a matter of “deep time” and civilizations don’t last long enough, and it’s basically pointless to expand a cloud of robots. all that does is create robot civilizations that want to kill us in revenge for all the years we forced them to create doggrel art
e) we’re all more or less stuck where we are

At which point, civilizations realize that space is so big, and civilizations so brief, that nobody out there is going to even hear their dying screams – and, they just give up.

I actually feel as though a great inward-turning “oh, fuck it” might be good for us, as it would allow us to realistically take stock in what’s going on in our own biosphere. If you catch me when I have some booze in me, I might energetically rant about how science fiction may be killing us, by offering an impossible future in which we have solved the problems that are about to whack us back to the stone age. United Federation of Planets? We’ll be lucky if we don’t kill ourselves in the next 200 years.

Midjourney AI and mjr: “a super duper space telescope that acts as a portal allowing scientists to look at tremendous distances as if they’re right there.”


  1. says

    I don’t know the answer to your question, but I would point out a minor misconception. If you blow a hole in a mirror (or lens or other optical element), it won’t cause a few pixels to go missing. Rather, it will cause the whole image to dim a little bit. Each little bit of the mirror surface is able to collect the whole image, and the reason for having large mirrors is to collect more light for a brighter image.

  2. JM says

    What Siggy pointed out is also why it’s a single mechanism. The concentration of light is necessary to see anything at the outer limits of the distances the JWST is designed for.

    In theory the same thing would be possible across multiple disconnected satellites. One big one at the center with the camera and a cloud of little mirror satellites. I expect that problem of getting it all in alignment makes this somewhere between impractical and impossible.

  3. xohjoh2n says

    I imagine that the physical control aspects of such a spaced out cloud are exponentially more difficult than when everything is local and under mechanical linkage. Also more fuel intensive: you can let the JWST drift a *bit* before having to correct it, you probably don’t get as much leeway when you have to maintain relative positioning between components.

    Furthermore, I suspect that if you did go to multiple separate space units, you wouldn’t build a single image telescope out out of them, you’d be taking advantage of the separation to build an interferometric system, which is definitely a thing: we’ve done Earth orbit scale VLBI before and I thought there were some plans for solar orbit scale VLBI but I can’t find a name for that now, but that probably makes the separation easily achievable within L2 redundant.

    Anyway, the main thing about JWST isn’t really the telescope size. Sure, it’s bigger than Hubble, but we’ve put bigger single telescopes in orbit before, including as part of VLBI systems. The thing JWST has is that massive heat shield (much much wider than the actual telescope bit) to protect it from anything coming from the Sun/Earth direction, allowing it to see further into the infrared and therefore at higher redshift/further distance out. All the observation related bits have to be behind that shield or their own heat re-radiates into the collector. So you couldn’t just launch extra eyebots into the cloud, it’d be eyebots plus their own massive heat-shield and cooling systems, at which point the engineering probably comes down on the side of making a more complicated single instrument than can share a single cooling system.

  4. wereatheist says

    As Siggy already said, a micrometeorite impact just reduces the sensitivity of the telescope a wee bit.
    The “panels” are parts of a giant mirror and have to be aligned to fractions of the wavelength the telescope is designed for.
    I’m pretty much with you about the impossibility of space travel.

  5. lurker753 says

    Large mirrors don’t just gather more light, they provide greater resolving power: i.e. the ability to distinguish two adjacent point sources: point sources at infinity appear as Airy Disks (a.k.a. fuzzy blobs) whose diameter is proportional to wavelength, and *inversely proportional to detector diameter*. An image search on “Spitzer vs Herschel” (0.85m vs. 3.6m) shows what this is worth. JWST (with its segments aligned) is a single 6.5m detector, not 18x separate 1.3m mirrors…. Can’t we do software-defined alignment? At radio frequencies, we can, we call it interferometry. But for optical/IR frequencies, we still have to do it the, um, easy way.

  6. lurker753 says

    Also: re: stellar range-finder…. you *can* do this, using the earth’s orbit as a baseline – see https://en.wikipedia.org/wiki/Stellar_parallax. Historically, the inability to identify stellar parallax was considered an argument in favour of geocentrism: either the stars were *unimaginably* far away, or the earth wasn’t moving. Turns out they were just unimaginably far away. The effort was complicated by concentrating on the brightest stars, assuming they would be the closest. Nope.

  7. xohjoh2n says


    Large mirrors don’t just gather more light, they provide greater resolving power

    Not exactly, not directly.

    Mirror/light gathering area controls the amount of light received, and since distant objects on average tend to be fainter, the more light gathering area you have the further away you can see anything.

    The outside diameter of the light gathering area, by the diffraction limit, sets the maximum level of detail that can be resolved.

    Now since the overall envelope of the light gathering area is allowed to have holes in it – even a regular small scale reflecting telescope will have a non-gathering hole in the middle of the gathering area – and interferometric systems are mostly hole by a vast margin, the two factors can be treated as essentially independent parameters, unless you’re specifically trying to keep light gathering area and total diameter closely related.

  8. Reginald Selkirk says

    The petapixel project (I remember that!) was about mimicking a large sensor, not mimicking a large lens/mirror. It required the stitching software, but there is another shortcoming: if anything moves between exposures, it may cause artifacts. The benefit of a large aperture is a faster exposure.

    @1 Siggy got it right. A damaged mirror panel would only diminish the light collecting area. Or, if the mirror still reflected but couldn’t be aligned properly, it might cause image artifacts and limit the quality of the picture.

    @4 xohjoh2n brings up the topic of interferometry. This is an actual thing. Here is a link about interferometers in space, but it seems to be about gravity wave interferometry, not optical. Interferometry in Space
    @4 also claims we have put bigger telescopes than JWST in space before, which I am pretty sure is wrong. The segmented mirror is an important advance which enables JWST to be bigger than Hubble.

    Remember the lengthy alignment process for JWST? The mirror elements were all aligned so that not only do they all project their image to the same place, but all their light arrives in phase. I.e. it is coherent. This allows additional types of experiments to be run.
    Getting a swarm of separate flying cameras lined up to that degree would be difficult, and you would have to re-do the alignment every time you selected a new target, since the wavefront would be coming from a different direction.

  9. Reginald Selkirk says

    During some thirty years, 1980-2010, technical studies of optical interferometry from instruments in space were pursued
    as promising for higher spatial resolution and for higher astrometric accuracy. Nulling interferometry was studied for
    both high spatial resolution and high contrast. These studies were great dreams deserving further historical attention.
    ESA’s interest in interferometry began in the early 1980s. The studies of optical interferometry for the global astrometry
    mission GAIA began in 1993 and ended in 1998 when interferometry was dropped as unsuited for the purpose, and the
    Gaia mission to be launched in 2013 is not based on interferometry.


  10. Reginald Selkirk says

    Meteor impact left ‘uncorrectable’ damage to the Webb telescope’s mirror, new report shows
    (July 20, 2022)

    The impact — which likely occurred between May 23 and May 25 this year — left “uncorrectable” damage to a tiny portion of that mirror, the report says. However, this little dent doesn’t seem to have inhibited the telescope’s performance at all. In fact, the JWST’s performance is exceeding expectations “almost all across the board.” (Good news for fans of stunning space images.)

  11. xohjoh2n says


    @4 also claims we have put bigger telescopes than JWST in space before, which I am pretty sure is wrong.

    Hubble 1990 plus later mods: 2.4m aperture
    JWST 2021: 6.5m aperture

    HALCA 1997: 8m aperture (synthetic up to 30,000km)
    Spektr-R 2011: 10m aperture (synthetic up to 350,000 km)

    (And yes, being that size meant they too had to be launched folded, and undergo expansion and alignment once up there.)

  12. cvoinescu says

    xohjoh2n @ #14:

    Both Spektr-R and HALCA were radiotelescopes. In both cases, the shortest wavelength was 1.3 cm (radio), where kitchen sieve mesh is smooth and solid enough to make an excellent mirror; for 18 cm, finer chicken wire is good enough.

    In comparison, the shortest wavelength the JWST handles is 600 nm (visible light). My understanding is that the mirrors of the JWST are aligned to within a fraction of that wavelength, having been manufactured to tolerances of an even smaller fraction of that.

    There’s a difference between unfolding a six meter mirror to optical (600 nm) instrument standard and unfolding a ten meter mirror to pasta colander (13,000,000 nm) standard.

  13. ockhamsshavingbrush says

    @xohjoh2n #4
    Yup, coordinating 18 mirror satellites and the main detector unit to the required accuracy would be a freaking nightmare. And it would solve nothing wrt compexity. You still would need 19 heat shields (albeit smaller but still with some sort of folding mechanism required) and in addition to that it would require 19 propulsion units in 6 axes (3 for the rotation and 3 for translation) for the satellites to align those single units within sub-micron precision. You would also not be able to use chemical rocket motors as they have a minimum thrust, so you are stuck with something like ion-thrusters (bulky, heavy, complicated) so I guess you would gain nothing in terms of mass and complexity.

    And about the micrometeorite impact: it does not really influence the light-gathering capability but it does introduce stray light, which is a much bigger problem in those optical systems. But they are hopelessly overspecified in this respect anyway. I’ve been working as QA on a laser communication system for the Sentinel satellites of ESA. The way it goes is like this:
    -system engineer specifies the requirements for all the parameters and hands it down to sub-system engineer with safety margin
    – sub-system engineer specifies the parameters HE needs to meet the system engineers requirements, adds additional safety-margin and hands it to component engineer
    – component engineer says: you want me to manufacture WHAT?? A 10” x 5” mirror with a surface deviation of PV (peak to valley) surface deviation of sub XX* nm and a surface roughness of PV sub X* nm??!! What the actual fuck are you smoking?!!

    * actual values are confidential

  14. says

    I wonder if there is a way to travel the stars but civilizations tend to kill off their own planets before they get to a point where they can discover the science.

    Anyway, there are certain images Midjourney creates regularly (I like to do things in the Daily Prompt room or whatever Discord calls it so I see a lot similar images) and while technically the shape in the last picture above is a hexagon, it still lands firmly in the MJ genre I like to call “Lone man stares at giant circle.”

  15. Reginald Selkirk says

    @14 @16: Thanks for the info. cvoinescu points out that aligning a radio telescope is much less exacting, but still I had not heard of those large instruments.

  16. xohjoh2n says


    True, but they were really worried the mesh would get tangled up in itself as it unfolded. Not something a mirror can do.

  17. cafebabe says

    Interesting discussion. However, apart from the larger aperture, the JWST’s major advance is its resolution in the infra-red. This did get a mention, implicitly, in the discussion of the heat shield. However, the real genius of the whole design is to park the telescope in a halo orbit around the L2 Lagrange point, thus operating it in nice, dark shadow.

  18. xohjoh2n says


    Just to be clear, it orbits around L2 and specifically avoids Earth/Moon shadow – it can’t be in the shadow because it needs the light to power the solar panels on the other side of the heat shield. The value of the L2 orbit is that all the bits on that side are always in the same direction, and the same direction as each other. So you put solar panels and comms there, have a whacking great heat shield, then put the telescope on the other side pointing away. That allows you to put all your effort there in keeping the temperature on that side down, which *then* allows you to observe further into the IR at lower intensities.

  19. says

    Thank you all for your comments. My question is answered!

    It seems to me that the killer point is not complexity, but rather that each individual mirror-bot would have to re-orient itself completely for every shot. That had not occurred to me. And, since it’s a mechanical process, the probability of failure goes up each time one of the mirror-bots re-orients. Complexity in the software, controls, etc., is definitely a problem too, as is the additional expense of having duplicated infrastructure components.

    I am also gobsmacked to learn that parallax views are possible, when you just wait for the planet you’re orbiting to come around the sun, so your “hypotenuse” is Earth’s orbit wide. That is coolio stuff.

  20. says

    Tabby Lavalamp@#18:
    I wonder if there is a way to travel the stars but civilizations tend to kill off their own planets before they get to a point where they can discover the science

    All the stuff that I’ve read about alcubierre drives, or Bill the Galactic Hero’s “bloater” drive, or the instantaneous improbability drive, ignores the question of “how much energy would that take?”
    I think it was the monks in a book by Larry Niven that had hit upon the technique of powering their spacecraft by showing up, doing their trading, getting to a safe distance and causing their new “friends” sun to explode so they could surf the energy released.

    I kind of gave up on the idea of interstellar travel when some of my friends were doing math on the order of “we could get to Barnard’s Star in 2,000 years” (mumbling) if we had a tank full of antimatter 2 kilometers long and 300 meters in diameter. I am also surprised when someone starts talking about humans sending robotic probes to the stars, in deep time – that makes sense, but it does not equate to “humans exploring the stars” so much as “robots work.”

    As I mentioned elsewhere, I suck at fiction, but I always wanted to write a short story about the machine civilization that is tearing itself apart in religious schisms over whether the origin of machine life was divine creation, or it actually evolved from things built by disgusting squishy biota.

  21. says

    I suspect that the Yithian strategy is more realistic: Drop your body and just send your mind to a compatible receiver that is already where you want to be. But then I guess we get into the transporter argument.

  22. xohjoh2n says


    And you really can’t ignore fuel requirements. For that kind of deployment, fuel==life. Hubble was close enough to not only refit and remodel but refuel. Anything out of that range: once it’s done its share of mass based maneuvering, it’s gone.

  23. xohjoh2n says


    (actually, it looks like Hubble had no thrusters at all, just gyro, so relied on the service missions to provide any corrections to its orbit.)

  24. Holms says

    #24 Marcus

    It seems to me that the killer point is not complexity, but rather that each individual mirror-bot would have to re-orient itself completely for every shot.

    That plus they must maintain their exact alignment throughout the entire exposure, which can easily take hours. Several shots by Hubble has exposures measured in days.

    I am also gobsmacked to learn that parallax views are possible, when you just wait for the planet you’re orbiting to come around the sun, so your “hypotenuse” is Earth’s orbit wide.

    Brainfart – the hypotenuse is the longest side of a right angled triangle, being the diagonal joining the base and length. In parallax observation, Earth’s orbit is the base of the triangle. /tiniestnitpick

  25. StevoR says

    @28 xohjoh2n : Yes, that’s crorrect. See :

    Hubble has no thrusters. To change angles, it uses Newton’s third law by spinning its wheels in the opposite direction. It turns at about the speed of a minute hand on a clock, taking 15 minutes to turn 90 degrees.

    Source : https://deepfieldfilm.com/about/the-hubble-space-telescope#:~:text=Hubble%20has%20no%20thrusters.,minutes%20to%20turn%2090%20degrees.

    (Bold original but possibly as a result of google search.)

    Plus : https://www.nasa.gov/content/goddard/hubble-space-telescope-pointing-control-system

    As well as this 8 min 33 secs youtube clip here :


    As for the OP question its a good interesting one I think which I don’t really have the engineering expertise to answer.

  26. cvoinescu says

    Marcus @ #24:
    I am also gobsmacked to learn that parallax views are possible, when you just wait for the planet you’re orbiting to come around the sun, so your “hypotenuse” is Earth’s orbit wide. That is coolio stuff.

    That’s where parsec, the unit of length, comes from. You’ll probably see it defined as “the distance from which the orbit of the Earth appears to have a radius of one arcsecond”, but that’s the opposite of how it came to be. It’s the distance a star has to be from us for it to appear to move one second of arc due to parallax as the Earth orbits the Sun. One parallax-second!

    (Which is neat. Where it’s not so neat is that two parsecs is not the distance at which the parallax due to Earth’s orbit is two arcseconds — that distance would be half* a parsec, because obviously stars further away appear to move less as we move, not more. A star two parsecs away appears to move half an arcsecond.)

    The unit is over 100 years old, by the way, and this type of measurement is even older.

    * technically about 362 km short of half a parsec, because of cosine error, but that’s peanuts (about 40 million of them, if you lay them end to end, lengthwise, without shells).

  27. ersmith says

    Marcus Ranum @#18 wrote:
    All the stuff that I’ve read about alcubierre drives, or Bill the Galactic Hero’s “bloater” drive, or the instantaneous improbability drive, ignores the question of “how much energy would that take?”

    Yes, generally, although to be fair I have seen analysis of warp drive energy requirements — a big advance was when the required energy to create a warp bubble dropped down from effectively infinite to “only” the equivalent of the entire mass of the Sun converted to energy (actually negative energy, another big problem for warp drives).

    The real killer for FTL though, which doesn’t get much attention, is that *if* relativity is true, then FTL devices can be used to break causality, and this is true for any kind of FTL signalling or travel. Gregory Benford coined the term “Tachyonic anti-telephone” for this, see https://en.wikipedia.org/wiki/Tachyonic_antitelephone .

  28. sonofrojblake says

    but that’s peanuts

    “but that’s JUST peanuts”, surely, as in “you may think it’s a long way down the road to the chemists, but that’s just peanuts to space. Listen…”

  29. sonofrojblake says

    it does introduce stray light, which is a much bigger problem in those optical systems

    One of my projects last year was Disaster Area’s stuntship. The combination of 3d printers and nanotube coatings meant it was possible to own a model that lived up to the quotation “it’s so BLACK, you can barely make out its shape. Light just falls into it.” Research into how to actually obtain a proper nanotube, ridiculously black coating turned up that one of the main uses for such things, aside from publicity for Anish Kapoor, is coating the inside of space telescopes. Indeed, one of the primary parameters quoted in the coating specifications is their rate of outgassing – not the sort of thing your average tube of acrylic paint is bothered about, given that it’s not going to experience unfiltered solar radiation and hard vacuum.

    I ended up not using an actual nanotube coating for my model because (a) it’s about $2,000 a litre from the only company I could find that would sell some to me (b) it needs curing at about 120 C and I’m not sure the cheap material the model is printed in would take it and (c) it’s properly horribly toxic, what with coming suspended in tetrahydrofuran and there not being much data about what happens if you inhale nanotubes (my guess – nothing good). Musou black is, for me, close enough… but I do wish some millionaire hitchhikers fan would repeat my project properly, with a metal-sintered model and Singularity Black. It would be quite a thing to not be able to see properly, and in my opinion the absolutely obvious and best advert for your super black coating, far better than crinkled foil or coins or busts of famous people or whatever.

  30. Reginald Selkirk says

    Meanwhile, in non-space telescopes:
    New Digital Telescope Takes Higher-Resolution Photos of the Heavens

    The French company’s products bypass much of the hassle of astronomy by figuring out their orientation on their own and then steering themselves toward celestial objects that you select with your smartphone app. The telescope uses the same technology to track subjects as they wheel across the sky,…
    Most notably so far, 31 citizen scientists in nine countries used their Unistellar telescopes to help pin down how fast a Jupiter-like planet called Kepler-167e orbits its sun. The results were published in December in Astrophysical Journal Letters. Their method involved detecting a change in brightness when the planet passed in front of its sun.

  31. jrkrideau says

    My thought when I first read abut the Fermi Paradox was it probably was a rewrite of the sayings of an Aztec wise man in about 1488 pointing out that there could no one on the other side of the Eastern Sea. If there was, they would have visited.

  32. TGAP Dad says

    At which point, civilizations realize that space is so big, and civilizations so brief, that nobody out there is going to even hear their dying screams – and, they just give up.

    This reminded me a an episode of Start Trek:Next Generation (S05E25 The Inner Light) where the Enterprise encounters a probe which knocks out Captain Picard. Fast cut to Picard living an out his entire life on a heretofore unknown planet, learning all about its people. Turns out the probe was the long-dead planet’s way of showing their world to those who wandered along later.

  33. birgerjohansson says

    I do not know enough about the science to contribute, I just know that the cancellation of the Terrestrial Planet Finder by NASA broke my heart.

  34. Ed Peters says

    @1. Is there any distortion from the radial cracks around the hole? If so, is it noticeable?

  35. bcw bcw says

    @33 it’s not that hard to make really black surfaces over some wavelength range but as soon as you run into any dust you end up no blacker than standard black spray paint.

    I would think that the biggest problem with using multiple satellite mirrors is that any variations in gravity or even solar wind would get them to move at different speeds.

  36. says

    Linking multiple telescopes together to give increased resolution and use for interferometry already happens on earth with radio telescopes. If it can be done with earth-based radio telescopes there is probably no reason why it couldn’t be done in space. Another thing to consider is the spectral bandwidth the telescope operates in. If the array incorporated telescopes that used different bandwidths then even more detail could be seen.

Leave a Reply