14 ms·

I know there is a very good reason for this, but it seems like we're sending a lot of satellites to these moons and planets to perform VERY similar missions. Or

by jdhawk 11y ago

I know there is a very good reason for this, but it seems like we're sending a lot of satellites to these moons and planets to perform VERY similar missions. Orbit around, take photos and readings. Why are we creating a new spacecraft every time this happens? Why are they not "mass producing" the same generic exploration sat, with a generally useful set of features, then slightly modifying it for any specifics? Launch in 2020's? snooze. Just launch the same sat from the last mission and start getting data now.

devindotcom 11y ago
Boy, where to start. First of all the conditions around various planets and bodies are very different. Jupiter for example is oozing with ugly radiation so we can't just put a "stock" probe into orbit around Europa, it would fry. Second, these missions aren't actually that frequent. It takes months or years for a spacecraft to arrive at its destination, and its design had to be finalized and tested rigorously years before its launch. That means by the time we even get a close up picture of Ceres, for example, the science of spacecraft building will have advanced by several years and we now can include new sensors and improve fuel efficiency and so on. Third, what do you think science is? Taking photos and readings. That is a huge, huge part of basic science, especially when it comes to astrophysics, astrogeology, astrochemistry, and so on. Landing on a planet provides further opportunities to take photos and readings, but it's phenomenally difficult to do. Finally, if you are bored by cutting-edge space travel and the study of distant worlds that may hold the possibility of life... I don't even know. I'm mystified by your perspective on the universe we live in.
- Achshar 11y ago
  But what's stopping us from making exact copies of this probe/sat (say... 10) and launching them all at once? Or at an interval of say 6 months a piece? That way we will have low cost since the r&d will be basically 0 for all the extra models and we get 10x the data and larger time span of coverage. Plus we won't be putting all our eggs in the same basket.
  - devindotcom 11y ago
    The day (or month, or year) you want to send a probe to Pluto and the day you want to send one to Mercury, not to mention the direction, are going to be totally different, and aren't easy to schedule — you might have a 3-day window to be sure something can get to Saturn, since you're relying on a gravity assist from Mars and a configuration of the planets that only happens every couple decades. Also, even if such a thing could be organized, there's simply no way the budget will support 10 missions launching at once. They have to be tracked, developed, and funded separately for a lot of really good reasons. Stuff that's shared (launch vehicles, software) can be co-developed but each mission has wildly different timelines and requirements.
  - Pelerin 11y ago
    Are launch costs still high?
    - IndianAstronaut 11y ago
      It will be a long while before launch costs are brought down. They aren't all that different than what they were a few decades ago.
  - billyhoffman 11y ago
    Because of budget. The closest thing to what you are describing that NASA has done recently was sending 2 rovers to Mars at the same time (Spirit and Opportunity). It cost nearly $1B dollars in 2003 money to do it [1]. Nasa's entire budget for 2015 is $17.5B In contrast, NASA only spent a further ~$130M keeping the rovers going for the last decade. R&D, as well as getting stuff into space, is incredibly expensive. [1] - https://en.wikipedia.org/wiki/Mars_Exploration_Rover https://en.wikipedia.org/wiki/Mars_Exploration_Rover
    - anigbrowl 11y ago
      Having two rovers turned out to be insanely popular. Yes, it was expensive and NASA's budget is too small. But let's separate out the marginal cost here, because I'm pretty certain that if we had (for example) sent 4 rather than 2, it wouldn't necessarily have cost twice as much.
- mturmon 11y ago
  To amplify your comment, here's a web page with documents from the science definition team -- the panel that decides what exactly the mission will study: https://solarsystem.nasa.gov/europa/sdt2013.cfm https://solarsystem.nasa.gov/europa/sdt2013.cfm A key document is the very granular "Science Traceability Matrix" which is very focused on "can instruments be made to satisfy these science goals" (zoom in to notice a lot of requirements flow-down from science goals): https://solarsystem.nasa.gov/europa/docs/Clipper_Trace_Matrix_V2.2_140609.pdf https://solarsystem.nasa.gov/europa/docs/Clipper_Trace_Matri... And regarding Europa, one particular item of concern has been, can markers of life be detected in a plume as sampled by an orbiter. Here's a recent workshop on the subject: https://astrobiology.nasa.gov/calendar/europa-plume-workshop/ https://astrobiology.nasa.gov/calendar/europa-plume-workshop... It's a colossal understatement to say that a standard design plus tailoring isn't really going to lift the burden of figuring out if detecting life in the plume is possible. But that said, there are re-usable components. Some of the spectral radiometers ("take photos") are largely re-used, also communications devices, etc.
  - golergka 11y ago
    Apart from being extremely interesting from space perspective, this matrix seems like a great instrument to help manage a project of this complexity and uncertainty. I wonder what other things software industry can learn from NASA about this stuff.
    - fsloth 11y ago
      Well, NASA did run a software engineering laboratory for 25 years https://www.cs.umd.edu/~basili/publications/proceedings/P94.pdf https://www.cs.umd.edu/~basili/publications/proceedings/P94....
    - chii 11y ago
      > I wonder what other things software industry can learn from NASA they take their software seriously. This is just one shallow article describing their process: https://rhodecode.com/blog/code-review-learn-nasa-codes/ https://rhodecode.com/blog/code-review-learn-nasa-codes/ But i recall reading that they do simulation testing of their rocket firing software - by exhaustively running through all possible input ranges from all their sensors, and add in invalid ranges to boot, and their software _musts_ pass it. Takes days to run i heard.
kryptiskt 11y ago
Science instruments aren't generic, you have to choose a scientific payload that gives you a chance to attain your objectives while operating both under monetary and power constraints. The instruments have also improved tremendously over the last couple of decades. Here is an overview of what we know about the mission now: http://futureplanets.blogspot.se/2015/06/nasa-goes-first-class-for-europa.html http://futureplanets.blogspot.se/2015/06/nasa-goes-first-cla... "Galileo’s camera and spectrometers revealed that the icy crust is fractured and frequently covered with material that appears to have originated in the ocean below. The Clipper’s radar instrument (Radar for Europa Assessment and Sounding: Ocean to Near-surface (REASON) – principal investigator Dr. Donald Blankenship of the University of Texas, Austin) will see below the surface to investigate the structure of the shell, potentially all the way to the interface with the ocean below." "The MAss SPectrometer for Planetary EXploration/Europa (MASPEX) (principal investigator Dr. Jack (Hunter) Waite, of the Southwest Research Institute (SwRI)) will measure the composition of gasses, ices, and organic molecules." "The wide angle camera in the Europa Imaging System (EIS) (principal investigator Dr. Elizabeth Turtle of APL) will map the surface of Europa at 50 meter resolution in color to document the surface structure." "During each of the planned 45 flybys, the spacecraft will travel close to the surface of Europa. At each encounter, the wide and narrow angle EIS cameras will record the surface geology in high resolution recording details as small as one meter."
avian 11y ago
Different missions require vastly different hardware. > Orbit around, Orbiting Europa requires a different amount of propellant than orbiting Mars. Solar panels that power a probe at Mars won't power one farther away from the sun. Radio that can talk with Earth from the Mars orbit won't be able to do that from Jupiter. > take photos and Different cameras required due to different amount of light, expected distances for imaging, filters optimized for materials being imaged, etc. > readings. What readings? Basically no two probes carry the same set of scientific instruments. Often, the instruments carried are one-of-a-kind, specifically designed for the specific thing the mission wants to investigate.
- anigbrowl 11y ago
  Basically no two probes carry the same set of scientific instruments. That's the problem. Great as those are, by continually making one- or two-of-a-kind devices we are endlessly prototyping. What if we picked one or a few different designs, selecting for greatest generality, and then worked to get the costs very low by manufacturing a lot of them, accepting that they will be suboptimal for almost every target? Of course crappy probes would give crappy results, and many of them would fail altogether. But what if we deployed hundreds of cheap crappy probes on a regular basis - ie build a shotgun instead of a series of sniper rifles? Obviously finding a good general-purpose design is easier said than done. We have I think 3 lunar probes in orbit at the moment, and a few more on the way, as well as various probes that have gone to the moon, sent back a bit of data, and then crashed into it by design. That's not very many at all. The moon's not that far away, why not try putting 50 low-cost probes around it with the same instrumentation and see what we learn from that? If we can get better astronomical observations from arrays of relatively low-power telescopes, surely we will get better planetary observational data from arrays of low-quality probes? Simply getting an array of probes up there and running and learning hwo to handle the networking, data flow, and and the inevitable variety of unexpected failure modes will provide us with a vast amount of experience, not to mention a vast amount of additional data about the moon that can be benchmarked against a whole lot of excellent data we have already for accuracy. I don't mean this as a dig at you or the other commenters critiquing this proposal, but you remind me of Thomas J Watson suggesting that "I think there is a world market for maybe five computers" (not withstanding the apocryphal nature of this quote, similar sentiments were expressed by other experts around that period, eg https://en.wikipedia.org/wiki/Thomas_J._Watson#Famous_misquote https://en.wikipedia.org/wiki/Thomas_J._Watson#Famous_misquo...).
uxcn 11y ago
The destinations the probes go to are drastically different, and generally the things you want to study aren't identical. What's more is that the mission plans for probes need to be extremely detailed, which impacts the design of the actual probes. The extreme level of detail necessary for a mission is also one of the reasons probes aren't launched on short timeframes. However, they are absolutely used for more than one purpose, and this is part of the reason the mission plans are so detailed. We generally try to get as much usable science out of what we send. For example, the mission for Cassini Huygens[1] included Venus, Jupiter, and obviously Saturn and its moons. Some of the other easy examples... Voyager[2] I and II, Spirit and Opportunity[3]. [1] https://en.wikipedia.org/wiki/Cassini%E2%80%93Huygens https://en.wikipedia.org/wiki/Cassini%E2%80%93Huygens [2] https://en.wikipedia.org/wiki/Voyager_program https://en.wikipedia.org/wiki/Voyager_program [3] https://en.wikipedia.org/wiki/Mars_Exploration_Rover https://en.wikipedia.org/wiki/Mars_Exploration_Rover
jccooper 11y ago
The spacecraft itself isn't a particularly big deal on these missions. They will usually use an existing bus or put together existing subsystems. So basically they already do as you suggest. Cassini, currently around Saturn, is directly descended from the Mariner/Voyager/Viking line, as were Galileo and Magellan. MAVEN, MRO, and Mars Odyssey are all basically the same type. Mars Observer was a comsat, and most Earth observers use a commercial sat bus. Historically the Explorer, Mariner/Voyager/Viking, Pioneer, Ranger, Venera, Zond, and other families shared quite a lot. Most of the effort goes into novel and usually hand-made sensors to deal with specific scientific questions. There's not particularly a point in sending the same instruments to the same places again, and the accessible parts of the solar system are fairly well studied from a generic standpoint. If you're going through the (fairly expensive) effort at all, you want better or different instruments, and that's what takes so long. There's a reason all these programs are led by a "principal investigator" and not a "head engineer". I do think a common design with fairly generic instruments would be useful for a pair of ice-giants orbiters. Uranus and Neptune are basically unstudied up close, compared to all the other planets. But two does not mass production make.
srdev 11y ago
> I know there is a very good reason for this, but it seems like we're sending a lot of satellites to these moons and planets to perform VERY similar missions. Orbit around, take photos and readings. I think you're overly abstracting the missions. It may seem similar at such a high level, but the actual readings being taken are likely very different and require very different hardware.
fsloth 11y ago
The same reason every modern automobile is not a model T ford and every computer is not just an Apple 2? Also, space missions do not have the luxury of activities on earth where if things go wrong one can quickly try again if funds are available. Thus, much more preparation is needed.
jasonkester 11y ago
Because we know what happens when you try to make an expensive thing that takes the place of several somewhat related other expensive things and can perform all of their capabilities: https://en.wikipedia.org/wiki/Joint_Strike_Fighter_program#Cost_overruns https://en.wikipedia.org/wiki/Joint_Strike_Fighter_program#C... https://en.wikipedia.org/wiki/Joint_Tactical_Radio_System#Problems_and_restructuring https://en.wikipedia.org/wiki/Joint_Tactical_Radio_System#Pr... You end up with a super duper expensive thing that takes 15 years longer than you expected to build and ends up not being able to do the job of any one of the things it was intended to replace. There's a lively discussion on the front page about one of the examples above.
lmm 11y ago
A lot of the time there are good reasons, but the funding structures of this kind of large-scale science also generate some perverse incentives. No-one's employed full-time to "be a scientist" at the kind of scale that a space mission takes - at best some of the people will be tenured researchers, but their loyalty tends to (reasonably) belong to their specific institution. Usually at least some of the people involved will be commercial contractors who will then move elsewhere once the money goes away - and it will, as soon as the probe is launched. Finally, science tends to be funded on a grant basis, and these grants are always for specific things. So imagine someone working on the Foo Mars Orbiter. They're probably employed by the University of Bartown, and their work is being funded by an NSF grant for the Foo Mars Orbiter project. Neither of those bodies is interested in making something reusable for another satellite ten years down the line: the university may well not be involved in the next one, and they're unlikely to get much academic prestige from having their designs reused, they'd rather their professor spend time making their paper nicer and getting it in better journals. Meanwhile the NSF wants its grant to be used for the grant's specific purpose, not as a general slush fund, and will be hostile to any expenditure that's not directly related to getting the Foo Orbiter into orbit. Even if the team did put together something reusable, the team's going to be dissolved as soon as it launches (with a skeleton crew remaining to check when it arrives) or at very best once the primary mission completes, so no-one's going to be in a position to handover the assets to the team making the next satellite. If you've ever tried to reuse some code that was written two years ago by someone who thought it was a one-off and has since left the company, you get the idea.

by jdhawk 11y ago

devindotcom 11y ago

Boy, where to start. First of all the conditions around various planets and bodies are very different. Jupiter for example is oozing with ugly radiation so we can't just put a "stock" probe into orbit around Europa, it would fry. Second, these missions aren't actually that frequent. It takes months or years for a spacecraft to arrive at its destination, and its design had to be finalized and tested rigorously years before its launch. That means by the time we even get a close up picture of Ceres, for example, the science of spacecraft building will have advanced by several years and we now can include new sensors and improve fuel efficiency and so on. Third, what do you think science is? Taking photos and readings. That is a huge, huge part of basic science, especially when it comes to astrophysics, astrogeology, astrochemistry, and so on. Landing on a planet provides further opportunities to take photos and readings, but it's phenomenally difficult to do. Finally, if you are bored by cutting-edge space travel and the study of distant worlds that may hold the possibility of life... I don't even know. I'm mystified by your perspective on the universe we live in.

Achshar 11y ago
But what's stopping us from making exact copies of this probe/sat (say... 10) and launching them all at once? Or at an interval of say 6 months a piece? That way we will have low cost since the r&d will be basically 0 for all the extra models and we get 10x the data and larger time span of coverage. Plus we won't be putting all our eggs in the same basket.
- devindotcom 11y ago
  The day (or month, or year) you want to send a probe to Pluto and the day you want to send one to Mercury, not to mention the direction, are going to be totally different, and aren't easy to schedule — you might have a 3-day window to be sure something can get to Saturn, since you're relying on a gravity assist from Mars and a configuration of the planets that only happens every couple decades. Also, even if such a thing could be organized, there's simply no way the budget will support 10 missions launching at once. They have to be tracked, developed, and funded separately for a lot of really good reasons. Stuff that's shared (launch vehicles, software) can be co-developed but each mission has wildly different timelines and requirements.
- Pelerin 11y ago
  Are launch costs still high?
  - IndianAstronaut 11y ago
    It will be a long while before launch costs are brought down. They aren't all that different than what they were a few decades ago.
- billyhoffman 11y ago
  Because of budget. The closest thing to what you are describing that NASA has done recently was sending 2 rovers to Mars at the same time (Spirit and Opportunity). It cost nearly $1B dollars in 2003 money to do it [1]. Nasa's entire budget for 2015 is $17.5B In contrast, NASA only spent a further ~$130M keeping the rovers going for the last decade. R&D, as well as getting stuff into space, is incredibly expensive. [1] - https://en.wikipedia.org/wiki/Mars_Exploration_Rover https://en.wikipedia.org/wiki/Mars_Exploration_Rover
  - anigbrowl 11y ago
    Having two rovers turned out to be insanely popular. Yes, it was expensive and NASA's budget is too small. But let's separate out the marginal cost here, because I'm pretty certain that if we had (for example) sent 4 rather than 2, it wouldn't necessarily have cost twice as much.
mturmon 11y ago
To amplify your comment, here's a web page with documents from the science definition team -- the panel that decides what exactly the mission will study: https://solarsystem.nasa.gov/europa/sdt2013.cfm https://solarsystem.nasa.gov/europa/sdt2013.cfm A key document is the very granular "Science Traceability Matrix" which is very focused on "can instruments be made to satisfy these science goals" (zoom in to notice a lot of requirements flow-down from science goals): https://solarsystem.nasa.gov/europa/docs/Clipper_Trace_Matrix_V2.2_140609.pdf https://solarsystem.nasa.gov/europa/docs/Clipper_Trace_Matri... And regarding Europa, one particular item of concern has been, can markers of life be detected in a plume as sampled by an orbiter. Here's a recent workshop on the subject: https://astrobiology.nasa.gov/calendar/europa-plume-workshop/ https://astrobiology.nasa.gov/calendar/europa-plume-workshop... It's a colossal understatement to say that a standard design plus tailoring isn't really going to lift the burden of figuring out if detecting life in the plume is possible. But that said, there are re-usable components. Some of the spectral radiometers ("take photos") are largely re-used, also communications devices, etc.
- golergka 11y ago
  Apart from being extremely interesting from space perspective, this matrix seems like a great instrument to help manage a project of this complexity and uncertainty. I wonder what other things software industry can learn from NASA about this stuff.
  - fsloth 11y ago
    Well, NASA did run a software engineering laboratory for 25 years https://www.cs.umd.edu/~basili/publications/proceedings/P94.pdf https://www.cs.umd.edu/~basili/publications/proceedings/P94....
  - chii 11y ago
    > I wonder what other things software industry can learn from NASA they take their software seriously. This is just one shallow article describing their process: https://rhodecode.com/blog/code-review-learn-nasa-codes/ https://rhodecode.com/blog/code-review-learn-nasa-codes/ But i recall reading that they do simulation testing of their rocket firing software - by exhaustively running through all possible input ranges from all their sensors, and add in invalid ranges to boot, and their software _musts_ pass it. Takes days to run i heard.

kryptiskt 11y ago

Science instruments aren't generic, you have to choose a scientific payload that gives you a chance to attain your objectives while operating both under monetary and power constraints. The instruments have also improved tremendously over the last couple of decades. Here is an overview of what we know about the mission now: http://futureplanets.blogspot.se/2015/06/nasa-goes-first-class-for-europa.html http://futureplanets.blogspot.se/2015/06/nasa-goes-first-cla... "Galileo’s camera and spectrometers revealed that the icy crust is fractured and frequently covered with material that appears to have originated in the ocean below. The Clipper’s radar instrument (Radar for Europa Assessment and Sounding: Ocean to Near-surface (REASON) – principal investigator Dr. Donald Blankenship of the University of Texas, Austin) will see below the surface to investigate the structure of the shell, potentially all the way to the interface with the ocean below." "The MAss SPectrometer for Planetary EXploration/Europa (MASPEX) (principal investigator Dr. Jack (Hunter) Waite, of the Southwest Research Institute (SwRI)) will measure the composition of gasses, ices, and organic molecules." "The wide angle camera in the Europa Imaging System (EIS) (principal investigator Dr. Elizabeth Turtle of APL) will map the surface of Europa at 50 meter resolution in color to document the surface structure." "During each of the planned 45 flybys, the spacecraft will travel close to the surface of Europa. At each encounter, the wide and narrow angle EIS cameras will record the surface geology in high resolution recording details as small as one meter."

avian 11y ago

Different missions require vastly different hardware. > Orbit around, Orbiting Europa requires a different amount of propellant than orbiting Mars. Solar panels that power a probe at Mars won't power one farther away from the sun. Radio that can talk with Earth from the Mars orbit won't be able to do that from Jupiter. > take photos and Different cameras required due to different amount of light, expected distances for imaging, filters optimized for materials being imaged, etc. > readings. What readings? Basically no two probes carry the same set of scientific instruments. Often, the instruments carried are one-of-a-kind, specifically designed for the specific thing the mission wants to investigate.

anigbrowl 11y ago
Basically no two probes carry the same set of scientific instruments. That's the problem. Great as those are, by continually making one- or two-of-a-kind devices we are endlessly prototyping. What if we picked one or a few different designs, selecting for greatest generality, and then worked to get the costs very low by manufacturing a lot of them, accepting that they will be suboptimal for almost every target? Of course crappy probes would give crappy results, and many of them would fail altogether. But what if we deployed hundreds of cheap crappy probes on a regular basis - ie build a shotgun instead of a series of sniper rifles? Obviously finding a good general-purpose design is easier said than done. We have I think 3 lunar probes in orbit at the moment, and a few more on the way, as well as various probes that have gone to the moon, sent back a bit of data, and then crashed into it by design. That's not very many at all. The moon's not that far away, why not try putting 50 low-cost probes around it with the same instrumentation and see what we learn from that? If we can get better astronomical observations from arrays of relatively low-power telescopes, surely we will get better planetary observational data from arrays of low-quality probes? Simply getting an array of probes up there and running and learning hwo to handle the networking, data flow, and and the inevitable variety of unexpected failure modes will provide us with a vast amount of experience, not to mention a vast amount of additional data about the moon that can be benchmarked against a whole lot of excellent data we have already for accuracy. I don't mean this as a dig at you or the other commenters critiquing this proposal, but you remind me of Thomas J Watson suggesting that "I think there is a world market for maybe five computers" (not withstanding the apocryphal nature of this quote, similar sentiments were expressed by other experts around that period, eg https://en.wikipedia.org/wiki/Thomas_J._Watson#Famous_misquote https://en.wikipedia.org/wiki/Thomas_J._Watson#Famous_misquo...).

uxcn 11y ago

The destinations the probes go to are drastically different, and generally the things you want to study aren't identical. What's more is that the mission plans for probes need to be extremely detailed, which impacts the design of the actual probes. The extreme level of detail necessary for a mission is also one of the reasons probes aren't launched on short timeframes. However, they are absolutely used for more than one purpose, and this is part of the reason the mission plans are so detailed. We generally try to get as much usable science out of what we send. For example, the mission for Cassini Huygens[1] included Venus, Jupiter, and obviously Saturn and its moons. Some of the other easy examples... Voyager[2] I and II, Spirit and Opportunity[3]. [1] https://en.wikipedia.org/wiki/Cassini%E2%80%93Huygens https://en.wikipedia.org/wiki/Cassini%E2%80%93Huygens [2] https://en.wikipedia.org/wiki/Voyager_program https://en.wikipedia.org/wiki/Voyager_program [3] https://en.wikipedia.org/wiki/Mars_Exploration_Rover https://en.wikipedia.org/wiki/Mars_Exploration_Rover

jccooper 11y ago

The spacecraft itself isn't a particularly big deal on these missions. They will usually use an existing bus or put together existing subsystems. So basically they already do as you suggest. Cassini, currently around Saturn, is directly descended from the Mariner/Voyager/Viking line, as were Galileo and Magellan. MAVEN, MRO, and Mars Odyssey are all basically the same type. Mars Observer was a comsat, and most Earth observers use a commercial sat bus. Historically the Explorer, Mariner/Voyager/Viking, Pioneer, Ranger, Venera, Zond, and other families shared quite a lot. Most of the effort goes into novel and usually hand-made sensors to deal with specific scientific questions. There's not particularly a point in sending the same instruments to the same places again, and the accessible parts of the solar system are fairly well studied from a generic standpoint. If you're going through the (fairly expensive) effort at all, you want better or different instruments, and that's what takes so long. There's a reason all these programs are led by a "principal investigator" and not a "head engineer". I do think a common design with fairly generic instruments would be useful for a pair of ice-giants orbiters. Uranus and Neptune are basically unstudied up close, compared to all the other planets. But two does not mass production make.

srdev 11y ago

> I know there is a very good reason for this, but it seems like we're sending a lot of satellites to these moons and planets to perform VERY similar missions. Orbit around, take photos and readings. I think you're overly abstracting the missions. It may seem similar at such a high level, but the actual readings being taken are likely very different and require very different hardware.

fsloth 11y ago

The same reason every modern automobile is not a model T ford and every computer is not just an Apple 2? Also, space missions do not have the luxury of activities on earth where if things go wrong one can quickly try again if funds are available. Thus, much more preparation is needed.

jasonkester 11y ago

Because we know what happens when you try to make an expensive thing that takes the place of several somewhat related other expensive things and can perform all of their capabilities: https://en.wikipedia.org/wiki/Joint_Strike_Fighter_program#Cost_overruns https://en.wikipedia.org/wiki/Joint_Strike_Fighter_program#C... https://en.wikipedia.org/wiki/Joint_Tactical_Radio_System#Problems_and_restructuring https://en.wikipedia.org/wiki/Joint_Tactical_Radio_System#Pr... You end up with a super duper expensive thing that takes 15 years longer than you expected to build and ends up not being able to do the job of any one of the things it was intended to replace. There's a lively discussion on the front page about one of the examples above.

lmm 11y ago

A lot of the time there are good reasons, but the funding structures of this kind of large-scale science also generate some perverse incentives. No-one's employed full-time to "be a scientist" at the kind of scale that a space mission takes - at best some of the people will be tenured researchers, but their loyalty tends to (reasonably) belong to their specific institution. Usually at least some of the people involved will be commercial contractors who will then move elsewhere once the money goes away - and it will, as soon as the probe is launched. Finally, science tends to be funded on a grant basis, and these grants are always for specific things. So imagine someone working on the Foo Mars Orbiter. They're probably employed by the University of Bartown, and their work is being funded by an NSF grant for the Foo Mars Orbiter project. Neither of those bodies is interested in making something reusable for another satellite ten years down the line: the university may well not be involved in the next one, and they're unlikely to get much academic prestige from having their designs reused, they'd rather their professor spend time making their paper nicer and getting it in better journals. Meanwhile the NSF wants its grant to be used for the grant's specific purpose, not as a general slush fund, and will be hostile to any expenditure that's not directly related to getting the Foo Orbiter into orbit. Even if the team did put together something reusable, the team's going to be dissolved as soon as it launches (with a skeleton crew remaining to check when it arrives) or at very best once the primary mission completes, so no-one's going to be in a position to handover the assets to the team making the next satellite. If you've ever tried to reuse some code that was written two years ago by someone who thought it was a one-off and has since left the company, you get the idea.