I was surprised that the company doing the launch is charging $150k for it. Cool idea but a private company charging for promotional content seems a little rude. I'm just a regular software...
I was surprised that the company doing the launch is charging $150k for it. Cool idea but a private company charging for promotional content seems a little rude. I'm just a regular software person, not an aerospace engineer, but this seems like a lot of money to have an engineer review and integrate and test what I'd imagine are a few hundred lines of C code.
AIUI if they did nothing with the empty space it would cost them nothing and they'd earn nothing If they let corporations pay for running something in the space they would make a lot of money,...
AIUI if they did nothing with the empty space it would cost them nothing and they'd earn nothing
If they let corporations pay for running something in the space they would make a lot of money, while incurring an extra $150k in transmission costs, code review, etc
Donating it doesn't earn them any money, and they still get incurred an extra $150k in transmission costs, code review, etc.
So they're not willing to incur extra costs, but they're basically letting him send something at-cost, not making a profit on it. Still seems pretty cool to me
I think the thought is that if they weren't offering it to Matt, some other company would have paid for some of that time on the rover, doing something less promotional but also more lucrative for...
I think the thought is that if they weren't offering it to Matt, some other company would have paid for some of that time on the rover, doing something less promotional but also more lucrative for the company launching the rover. It's not the extra work they're doing, it's the opportunity cost of not selling it to someone else.
There’s also the cost of coordinating pretty powerful send-receive relays from the surface of Earth up to the moon (achievable, but certainly not trivial, and given how few anything’s have ever...
seems like a lot of money to have an engineer review and integrate and test what I'd imagine are a few hundred lines of C code.
There’s also the cost of coordinating pretty powerful send-receive relays from the surface of Earth up to the moon (achievable, but certainly not trivial, and given how few anything’s have ever been on the moon, very unlikely to have simple plug-and-play options available)
It’s also nothing like a regular code review — there’s probably no option for patching once launched, and given the enormous cost to launch at all, you need to be really really really sure that it doesn’t accidentally interfere with any other code, onboard systems, or anything else.
I don't buy that there's a meaningful marginal cost to this code on the data transmission side. In my head the mission was probably spec'ed with some amount of required bandwidth for the mission...
I don't buy that there's a meaningful marginal cost to this code on the data transmission side. In my head the mission was probably spec'ed with some amount of required bandwidth for the mission and then some overhead. They'd be offering to let stand up maths make use of some small part of that overhead or other slack in transmissions. The data needed for sending (number of iterations and current pi value since last upload) is tens of bytes. Even sent hundreds of times it's probably a fraction of one image being sent back.
Also, space missions definitely get patches? The voyager 1 mission launched in 1977 got a code patch recently and it has left the solar system.
The integration testing is indeed the costly part and the bit I have little experience with, so from that respect perhaps the 150k is reasonable. Still feel like a well designed system shouldn't need that much additional review for 'take two random numbers, compute the distance, then store it' but I recognize that is engineering hubris on my part.
The "OS" on the rover likely won't be pre-emptive, all task over runs affect all code, and potentially soft brick the device if they prevent hard-realtime radio code from acquiring sufficient...
The "OS" on the rover likely won't be pre-emptive, all task over runs affect all code, and potentially soft brick the device if they prevent hard-realtime radio code from acquiring sufficient schedule quanta.
I haven't sent anything to space, but I have sent enough iot code into the world that would be cost prohibitive to fix once it's left the factory under certain failure modes to have some idea of how thorough this will need to be.
However, $150k still doesn't account for that. It's not going to take some one a whole year of eng time to review / write / stress test this code. I have no idea about radio relay costs.
Am I missing something here? Is this not taking up a spot that could be used for doing science to do something computationally worthless? Perhaps this is just using a platform to generate the...
Am I missing something here? Is this not taking up a spot that could be used for doing science to do something computationally worthless?
Perhaps this is just using a platform to generate the revenue needed to fund a mission, but I feel like Matt Parker could have done this but also said "50% of the computation time allocated to this will be used to run a science experiment one of you designs", or something
I'm not exactly sure, but it looks like these cycles that will be used to calculate PI will appear at random intervals, so, maybe, they decided to use it just for advertisement, because practical...
I'm not exactly sure, but it looks like these cycles that will be used to calculate PI will appear at random intervals, so, maybe, they decided to use it just for advertisement, because practical applications is limited.
As far as I understood, they’re only putting some extra code on the rover, and they are being given the lowest computational priority (they only run if compute cannot be used for something else)....
As far as I understood, they’re only putting some extra code on the rover, and they are being given the lowest computational priority (they only run if compute cannot be used for something else). I doubt computer memory is a concern for the team. They’re not adding any additional sensors, and the robot won’t add additional behaviors for them.
They’re just using existing sensor data and doing some math to sample from a (hopefully) near-uniform random variable. The extra costs are to pay for testing to make sure the additional software doesn’t cause bugs with the important stuff.
So, this isn’t competing with scientific experiments that would require additional sensors or rover behavior. I suspect any meaningful scientific experiment that is just software could also be added, but it’s likely much cheaper to just run that software on Earth after the data transmission sends the data back here, since you don’t need to test its functionality with the existing rover software. But doing that with the pi calculations would defeat the actual publicity of the computation happening on the moon.
I was surprised that the company doing the launch is charging $150k for it. Cool idea but a private company charging for promotional content seems a little rude. I'm just a regular software person, not an aerospace engineer, but this seems like a lot of money to have an engineer review and integrate and test what I'd imagine are a few hundred lines of C code.
AIUI if they did nothing with the empty space it would cost them nothing and they'd earn nothing
If they let corporations pay for running something in the space they would make a lot of money, while incurring an extra $150k in transmission costs, code review, etc
Donating it doesn't earn them any money, and they still get incurred an extra $150k in transmission costs, code review, etc.
So they're not willing to incur extra costs, but they're basically letting him send something at-cost, not making a profit on it. Still seems pretty cool to me
I think the thought is that if they weren't offering it to Matt, some other company would have paid for some of that time on the rover, doing something less promotional but also more lucrative for the company launching the rover. It's not the extra work they're doing, it's the opportunity cost of not selling it to someone else.
There’s also the cost of coordinating pretty powerful send-receive relays from the surface of Earth up to the moon (achievable, but certainly not trivial, and given how few anything’s have ever been on the moon, very unlikely to have simple plug-and-play options available)
It’s also nothing like a regular code review — there’s probably no option for patching once launched, and given the enormous cost to launch at all, you need to be really really really sure that it doesn’t accidentally interfere with any other code, onboard systems, or anything else.
I don't buy that there's a meaningful marginal cost to this code on the data transmission side. In my head the mission was probably spec'ed with some amount of required bandwidth for the mission and then some overhead. They'd be offering to let stand up maths make use of some small part of that overhead or other slack in transmissions. The data needed for sending (number of iterations and current pi value since last upload) is tens of bytes. Even sent hundreds of times it's probably a fraction of one image being sent back.
Also, space missions definitely get patches? The voyager 1 mission launched in 1977 got a code patch recently and it has left the solar system.
The integration testing is indeed the costly part and the bit I have little experience with, so from that respect perhaps the 150k is reasonable. Still feel like a well designed system shouldn't need that much additional review for 'take two random numbers, compute the distance, then store it' but I recognize that is engineering hubris on my part.
The "OS" on the rover likely won't be pre-emptive, all task over runs affect all code, and potentially soft brick the device if they prevent hard-realtime radio code from acquiring sufficient schedule quanta.
I haven't sent anything to space, but I have sent enough iot code into the world that would be cost prohibitive to fix once it's left the factory under certain failure modes to have some idea of how thorough this will need to be.
However, $150k still doesn't account for that. It's not going to take some one a whole year of eng time to review / write / stress test this code. I have no idea about radio relay costs.
Well, that was quick. 200% funded in a day.
Am I missing something here? Is this not taking up a spot that could be used for doing science to do something computationally worthless?
Perhaps this is just using a platform to generate the revenue needed to fund a mission, but I feel like Matt Parker could have done this but also said "50% of the computation time allocated to this will be used to run a science experiment one of you designs", or something
I'm not exactly sure, but it looks like these cycles that will be used to calculate PI will appear at random intervals, so, maybe, they decided to use it just for advertisement, because practical applications is limited.
As far as I understood, they’re only putting some extra code on the rover, and they are being given the lowest computational priority (they only run if compute cannot be used for something else). I doubt computer memory is a concern for the team. They’re not adding any additional sensors, and the robot won’t add additional behaviors for them.
They’re just using existing sensor data and doing some math to sample from a (hopefully) near-uniform random variable. The extra costs are to pay for testing to make sure the additional software doesn’t cause bugs with the important stuff.
So, this isn’t competing with scientific experiments that would require additional sensors or rover behavior. I suspect any meaningful scientific experiment that is just software could also be added, but it’s likely much cheaper to just run that software on Earth after the data transmission sends the data back here, since you don’t need to test its functionality with the existing rover software. But doing that with the pi calculations would defeat the actual publicity of the computation happening on the moon.
I suspect that the code running is going to be highly vetted, so user submissions might not really work well in this particular arena.
Houston, we have a maths problem