Why is each space probe is so different? This increase costs. Wouldn't it be beneficial to design one probe that can be sent in many different directions? Or maybe a set of probes, or maybe design a framework what a probe is, like a PC computer (that has many different parts, but all of them has common well agreed interfaces)?
My point is, if we can make a probe, we can benefit from making many of them (serial production) to decrease cost and gain more science at the end.
There already is some economy of scale in the space industry which I describe a bit more below. When it comes to interplanetary missions however, there is a significant limitation on the destination of the probe, and the mass of the vehicle.
Several (all?) of the major geostationary spacecraft manufacturers have a standard platform which is customized per client. For example, Boeing has the 702, Maxar (ex- SS/Loral) has the SSL 1300, and Airbus produces the EuroStar series.
It is typical for constellation operators to partner with a manufacturer. For example, OneWeb partnered with Airbus for their fleet and O3b with Thales.
For interplanetary missions, the greatest limitation is the amount of fuel onboard the spacecraft (usually referred to as the "delta-v budget" of the spacecraft). The rocket must place the spacecraft on the trajectory computed by the mission designers. For interplanetary and science missions, the team usually wants to add as much science payload as possible, at the expense of fuel. Therefore, the rocket usually releases the spacecraft at its maximum reach.
Let's switch to thinking in terms of economy of scale for science missions. A balance needs to be reached: if one wants economy of scale, dozens of spacecraft must be manufactured.
Therefore, accounting for the previous fuel limitation, one must decide whether to limit the onboard instruments to launch dozens of probes to same destination on just a few launches, or instead launch a single probe which can conduct much more science at that same destination.
So far, space program managers have opted to build science spacecraft on a one-off basis with a slew of scientific payloads instead of building dozens of interplanetary spacecraft with limited scientific capability. However, not all is lost between missions as most will reuse components or even working strategies. For example, the Mars 2020 rover uses a similar bus and landing mechanism to the Curiosity rover.
The benefits of tooling up for mass production only matter if you plan to make a lot of something. A cheap consumer computer case that sells for $63 requires millions and millions of dollars of tooling and equipment to produce them for that cost, and it requires that a large volume of them are sold to pay for that initial capital outlay.
If you only needed one computer case it would probably cost you many thousands of dollars (instead of $63) to make a single one as a bespoke item. The cost for one is tens or hundreds of times more than it could be, but if you only need one or two then the millions and millions you'd spend making a mass-produceable design make that one or two even more expensive yet. It's just not worth it.
A computer case is basically a box - its fundamental components are cheap and manufacturing costs dominate its pricetag. Modules for satellites are not cheap - they are filled with expensive and complex parts. Even removing the production costs, these are expensive components, so targeting the manufacturing costs, or especially the assembly costs, isn't necessarily going to provide the same gains, especially if the overall demand volume is weak.
Satellites are like this. Sure, we could make an assembly line to crank them out, but there's no point if you only even need two or ten or twenty of them. Exceptions exist, of course. SpaceX Starlink is a good example - there you have a heavy filled with dozens and dozens and dozens of mass produced, identical satellites. Naturally the production process for thousands of Starlink satellites is going to be much different from that of a single atmospheric probe destined for a highly specialized mission to Venus.
Space missions are usually highly specialized - everything about the satellite is optimized for its specific mission profile. How many batteries do you need? Are solar panels going to even be useful? etc? If you're going to the inner system it's much different from heading out to Jupiter - the way you power the satellite is different, the heat handling requirements are different, the amount and types of propellant will be different, the radiation shielding requirements will be different, the science you're trying to do will be different, etc.
All of these special challenges mean that it is usually worthwhile to customize the entire satellite for the specific mission, and no two missions have the same requirements. This greaty eats into the potential benefits you might find trying to recycle sub-components when so many parts need to be modified for this mission or that. Add to that the fact that every kg you can shave off of your design by removing generic things you don't need saves you almost $3k in launch costs alone (to say nothing of hauling that mass around for the entire mission), even today when launches are cheaper than ever.
There's very little that is generic about most space missions.
Two points that the answers so far have missed. First, missions are not all that frequent, and technology improves between each mission. For instance, consider the difference between the Voyager cameras https://voyager.jpl.nasa.gov/mission/spacecraft/instruments/iss/ and https://solarsystem.nasa.gov/missions/cassini/mission/spacecraft/cassini-orbiter/imaging-science-subsystem/
Second, missions are going to different places, and so have to deal with considerably different conditions. New Horizons had to deal with the low light conditions of the outer solar system, the Messenger Mercury orbiter had to deal with intense solar radiation, the Dawn misson needed lots of delta-V to visit Vesta and Ceres. A lunar lander has to use rocket power to land, while a Mars lander can use aerobraking and parachutes for most of its descent. And so on...
Sometimes we do build identical probes, though. The two Viking landers, the Spirit & Opportunity rovers, the two Voyagers...
PS: And a third point, which is that you need a team of scientists (or at least students or interested members of the public) to analyze the data returned by probes. Just the Mars Reconaissance Orbiter has returned something over 300 terabytes of data. (Some of that may be relayed from other missions.)
There are a lot of costs associated with operating an inter-planetary probe. Customizing the hardware is not a major driver.
For example, this article claims that the rocket to launch Curiosity was 20% of the 2.5 billion total price tag. The sky crane wasn't included in that figure, so the "rocket" that slowed Curiosity down on the other end increases the price even more. So roughly a quarter of the total budget is spent just getting the rover to it's target location!
If you think about all the other things that go into that total cost - especially the things that incur ongoing costs over years of operations, like paying to maintain communications systems, hardware to operate the rover and store data, paying the salaries for mission operators, paying the salaries for scientists to clean and process the data - you see that actually constructing the probe is not the main cost driver.
Further, using a set design template won't drastically impact the cost of construction. Keeping all the parts free of contamination during assembly costs money. Economics of scale don't really impact "clean room" costs.
Obviously, there would be cost savings by using a standard design, but the savings are low enough that it makes better sense to spend a little more to maximize the science value of each launch.
We actually do try to make more generic more standardized probes. CubeSat and especially the delfi program is based on the idea of "simple satellites that can easy be replaced".
Hardware in those satellites is actually really "standard", with (I know for delfi c3/n3xt) the goal being that nothing is more complex or harder to get than in your standard smartphone. - The antennas for the n3xt were actually nothing more than simple tape measure, which they knew would deploy due to the internal spring force.
The starlink constellation is a follow up on that, though that is once again purpose built hardware. Just a lot of similar stuff.
This is exactly Rocketlab's new business strategy! They are developing a satellite platform that you can mount instruments to. I believe this is only showing up now because of the reduced launch costs from the new players in the orbital launch business. Until recently launch costs were way too high to justify cheap payloads, thus the motives in the other answer apply.
Good question. It's because NASA prefers customization of hardware to standardization. Never the less; a space probe going out very very far has to be very light weight so it has as much fuel as possible and inertia travelling out.
