You're ignoring the lifting costs. The cost of lifting a payload into space and putting it into orbit is non-trivial compared to the cost of developing the probe. Can you imagine the cost of trying to put 50 probes into Julian orbit? Or even Lunar orbit? It would be cost-prohibitive. It gets worse if you use a bunch of mass-produced designs because you'd have a bunch of instruments of marginal value to the mission increasing lifting costs.
I also think you're assuming its possible to reduce to a general set of equipment that can answer the scientific questions that we're trying to answer. If you consider, for example, the difference between the things that Philae, Curiosity, New Horizons, Dawn, and this Europa probe are testing and the conditions that they are testing them in, then its hard to arrive at a common design that can handle all these conditions.
Edit: Another thing to consider is launch windows. You typically don't want to just regularly launch stuff to put near Mars or Jupiter whenever. You time your launches such that you can get them there within a certain amount of dV budget, otherwise you're dramatically increasing your deployment costs. This means that you only have a short window in which you can send out your shotgun probes -- you can't send them out constantly even if the lifting costs were feasible.
Let's focus on the lunar orbit idea as a relatively straightforward objective, since it is practically in our backyard and we have so much experience there.
I am not suggesting we perform 50 different launches into space. That would be hopelessly wasteful. But suppose we did one launch, sent one ship towards the moon, and then have it release 50 probes as it got close, each with a small amount of propellant sufficient get itself into orbit.
I am not suggesting a common design that is adequate to handle all the different conditions of the different missions you mention. I said specifically that we should focus some effort on developing cheap probes, accepting that they will be suboptimal in almost every case.
Let's consider one of the most basic things we like to do, which is to simply take pictures of things. Pictures help sell science to the public because most people are interested in how things look, and they are scientifically useful. When we aggregate multiple pictures of the same subject we often get even more useful scientific data. Downsides, the data transmission requirements are large and visual spectrum is just a small slice of the information we'd like to collect. Upsides, you can buy a COTS camera that takes 4k video or ~50mp stills for a few thousand $. Likewise you can buy a fast lens of high optical quality very cheaply, and record onto very cheap solid-state media. Let's accept that it will fail in some situations and that we don't expect it to keep working for ever, but we would like it to work for a while. so we need some power (onboard or renewable or some combination of the two), an antenna of some sort to transmit the data back and listen to requests from our end, some shielding to protect it against the slings and arrows of outrageous fortune, some sort of propulsion to get it into position and point it roughly where we want it to look, and a little control system to run it all.
Technologically this is no longer a tall order. We can stick a consumer video camera & phone in a lunchbox, attach it to a balloon, send it up to the stratosphere, and retrieve it afterwards for only hundreds of dollars, it's a middle-school project by now. I think that we could make a pretty decent camera probe that would take relatively high resolution pictures at a relatively low frame rate and last for at least a year for a marginal cost of $100,000, maybe quite a bit less. 50 of those would be $5 million, which is the sort of sum you can raise on Kickstarter. Now, the fixed costs of launch, building a deployment module and numerous other things would be a lot higher, let's say they started at $50 million. Well, that's quite a lot of money but you could still raise it pretty easily. Donald Trump plans to waste twice that amount on promoting himself as a public figure while pretending he wants to be President, a summer blockbuster movie has launch costs of about $200m including marketing. There are lots of people in Silicon Valley who could write checks for that whole amount if they really wanted to. I pick $50 million as a benchmark because India managed to get a probe going around Mars for ~$75 million, so I don't think it's a totally outrageous idea to think we could deploy a bunch of lunar microsatellites for 2/3 of that.
OK, let's say we even went overbudget by a factor 2 but we managed to do it. We have 50 probes in lunar orbit sending back, i dunno, 43 4k photographs of the lunar surface at the rate of 1 frame/minute (7 of them failed to deploy properly). None of them works properly for longer than 18 months. Within a few years they have all fallen out of orbit and are space junk on the lunar surface. Well, I think that we'd get a ton of useful knowledge from doing that.
> Technologically this is no longer a tall order. We can stick a consumer video camera & phone in a lunchbox, attach it to a balloon, send it up to the stratosphere, and retrieve it afterwards for only hundreds of dollars, it's a middle-school project by now. I think that we could make a pretty decent camera probe that would take relatively high resolution pictures at a relatively low frame rate and last for at least a year for a marginal cost of $100,000, maybe quite a bit less.
Where are you getting the $100k from? I would guess that a control system, communication system, reaction control system, propellant, batteries, solar panels, and a camera would cost much more than that if you're designing it to withstand space. The Indian probe you mentioned was 15 kg and cost $24 million. If you normalize that down you're talking $1.6 million per kg for a mission notable for its low cost. The LRO cost $504 million total, and Atlas V costs around 230 million If we take $250 million to be conservative, then it cost $2.5 million per kg of scientific instruments.
Note that the Indian orbiter had a dry mass of 500 kg and a launch mass of 1,337 kg to support that payload. The LRO has a dry mass of 1,018 kg and a launch mass of 1,916 kg. So the LRO had 10 kg of supporting dry mass for every 1 kg of science, and the indian orbiter had about 30 kg of supporting dry mass for 1 kg of science. Lets add Kaguya as another data point that had 1,984 kg dry mass, 2,914 kg launch mass, and the mission payload seemed to be around 300 kg, for a ratio of 6 kg weight for 1 kg of science. In other words, as your satellite gets smaller, the weight of all the other stuff required to keep your satellite ticking and pointing in the right direction becomes dominant. That means that one satellite with 100 kg payload will likely have less total weight than 50 satellites with a 2 kg payload. That increased total weight will come from installing redundant systems on all your satellites. So we're talking higher launch costs and higher part costs. Its certainly going to be difficult to improve upon costs by a literal order of magnitude.
So then the question becomes, what science can we accomplish with 50 satellites that we can't with 1? Is there a benefit to all this added cost and complexity?
I don't have the experience to address any of your points, but I do enjoy learning about this kind of stuff and there are some very smart people with long track records of past success working on these challenges:
"Through the use of multiple ARKYD 300 spacecraft per mission, Planetary Resources will distribute mission risk across several units, and allow for broad based functionality within the cluster of spacecraft.
The ARKYD 300 series spacecraft also demonstrate low-cost interplanetary capability, which is of interest to potential customers such as NASA, scientific agencies or other private exploratory organizations."
"Very often satellites are more expensive than the rocket. So in order for us to really revolutionize space, we have to address both satellites and rockets. We’re going to start off building our own constellation of satellites, but that same satellite technology that we develop can also be for science — Earth science and space science — as well as other potential applications that others may have. We’re definitely going to build our own, but also it’s something we would be able to offer to others."
Again: what would be the benefit of collecting hundreds of mediocre (and probably predictable after the first few samples) data points, without being able to do the kind of science you really want to do?
If we had a few things we knew we wanted to monitor for a long time, then this sort of proposal might make sense. But we don't, and there are always new questions to be answered, necessitating the use of different instruments on different missions. Those are rarely as simple as a camera simply taking pictures.
Moreover, I think you are overestimating the reliability of COTS hardware and underestimating the environment in space -- or underestimating the cost of radiation hardened hardware while overestimating its capabilities. You don't just grab a SD card or a CCD and send it into space. Likewise with lenses/filters (we are more interested in some wavelengths than others), etc.
COTS hardware may work for cute balloon projects that stay within the Earth's atmosphere, or (maybe!) satellites in low earth orbit. Space is a very different environment.
Even with "economies of scale" pushing down the cost per probe, you still have substantial fixed operational and science costs on top of that. Since running a large fleet of probes would be, overall, more complex than a single probe, I'm not sure if it would even be cheaper to operate than a few specialized probes. So you get worse science for, at best, the same cost, with much increased operational complexity. Again: what is the benefit?
I've already explained what the benefits are: practice, because we are going to want to run networks in space sooner or later anyway; the observational benefits from aggregating an array of relatively low-quality observations, which is something we already do for astronomy; and knowledge of failure modes and fault tolerances.
I don't expect COTS stuff to work that well or that long. But I would like to know how well or poorly it does perform. Some kinds of hardware are so cheap that we can afford to waste it on such experiments. Your reference to 'science costs' suggests to me that you've missed the point; I don't want to do any innovative science, I am perfectly happy to try something as simple as taking boring pictures to begin with as proof of concept, so we can concentrate on operational issues. Learning how to do things fast and cheaply even if the results are not especially good is a perfectly worthwhile goal in its own right.
> benefits are: practice, because we are going to want to run networks in space sooner or later anyway
As I've said earlier, learning how to manage a fleet of probes is a non-issue until we actually have a legitimate need for a large fleet of probes operating in concert, which won't be for a (very) long time. Note that we already have experience managing satellite constellations in the 50+ range.
> I don't expect COTS stuff to work that well or that long. But I would like to know how well or poorly it does perform. Some kinds of hardware are so cheap that we can afford to waste it on such experiments
TBH, neither do I, but 1) Someone does already, hence why we don't hear about Nikons on interplanetary missions, or even in orbit. 2) We don't need to send dozens of probes up to space to find out. In fact, we don't even need to leave the Earth. 3) Many instruments are not simply cameras, let alone COTS.
> Your reference to 'science costs' suggests to me that you've missed the point
You are moving the goal posts, your original comment was about sending dozens of cheap probes up to do different missions. Quoting:
> 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?
Even taking your statement that this isn't meant to be good science, beyond the perceived benefits of increased operational experience after the first few batches of probes (using the moon as your example), continuing with the "small, cheap, lousy" form factor is not going to outweigh the loss of spending the money instead on fewer solid science missions -- because, right now, what other purpose do we have for sending probes into space? We don't have the resources or knowledge to do anything else "useful" yet -- the science needs to come first. srdev has made better arguments on why it's still infeasible from a cost and technical perspective.
I am not moving the goalposts at all. I made a general point about what I'd like to happen I wrote an entire post addressing a single example of an initial project with the specific goal of doing nothing more exotic than taking pictures of the moon, our nearest neighbor to see what would be achievable at a low cost. I made it very clear from the outset that I wanted ot explore theidea of leveraging quantity and low cost at the expense of quality and speficity. To claim otherwise is not an honest way to carry on an argument.
Feel free to keep right on arguing about why this is stupid and a waste of time until someone gets around to doing it, which I predict will happen between 2025 and 2030.
"Feel free to keep right on arguing about why this is stupid and a waste of time until someone gets around to doing it"
This is something HN does a lot. Always mystifies me, for a site based around a supposedly disruptive startup industry. Lots of little mental boxes in many of the conversations.
EDIT- To be fair, the reason your proposal hasn't happened yet is that it's still very early days for space exploration. Satellites in Earth orbit are often like you describe, so contrary to some of the arguments against it, obviously it can be done. But even though we have countless photos of Jupiter, it's a lot more mysterious that it seems. So much still totally unknown, so they have to optimize for learning it all. Your approach is actually pretty good for refining general knowledge once the basics are locked in. As such, you're very likely correct that the Moon will be a target of such efforts soon.
I know of at least one case where the relevant science community would love to have more identical spacecrafts that are launched about one year apart: Follow ups to the very successfull stereo mission. In 2006 we launched two spacecraft that (compared to other science missions) were basically identical, with the goal of having solar observations not just from earth, but two other vantage points. During the mission it we learned that it is quite helpful to combine Stereo data with data from the Soho satellite that is stationed near earth at the L1 Lagrange point. And while having two or three observations from different directions is a start, having a couple more stereo-like satellite would make things like 3d reconstruction of CME fronts possible.
...except Pocket Spacecraft is even smaller. One of their probes is a CD-sized mylar disk with a solar panel and an Arduino-compatible microcontroller printed on it. The antenna is a wire ring around the outside. Sensors are minimal (I believe one of their models has a single pixel camera) and of course there's no propulsion. They were planning on launching hundreds at a time via cubesats.
They did a Kickstarter last year --- £99 for a vehicle in Earth orbit, £199 for one in lunar orbit --- which failed, but apparently they got funding elsewhere. Their website's short on updates but their twitter feed is active.
I also think you're assuming its possible to reduce to a general set of equipment that can answer the scientific questions that we're trying to answer. If you consider, for example, the difference between the things that Philae, Curiosity, New Horizons, Dawn, and this Europa probe are testing and the conditions that they are testing them in, then its hard to arrive at a common design that can handle all these conditions.
Edit: Another thing to consider is launch windows. You typically don't want to just regularly launch stuff to put near Mars or Jupiter whenever. You time your launches such that you can get them there within a certain amount of dV budget, otherwise you're dramatically increasing your deployment costs. This means that you only have a short window in which you can send out your shotgun probes -- you can't send them out constantly even if the lifting costs were feasible.