[Yawrehto]

It seems like a lot of these government-owned space things last a lot longer than they're made for. There's Landsat, Spirit, Opportunity, Hubble, the Voyagers, et cetera. It seems to be a pretty steep curve - either they fail on launch or landing or very early, or they far outlast expectations. There seems to be little that meets expectations. I can see lots of failures - space stuff is hard - but why so many things exceeding it?

[duskwuff]

> I can see lots of failures - space stuff is hard - but why so many things exceeding it?

The design lifetime is treated as a minimum acceptable value; a vehicle which was designed to last 10 years but has a critical component fail at 9.5 would be considered a failure, for instance. This means that the average lifespan of the vehicle gets pushed out a lot further to ensure it meets its goals.

With that being said, it's not uncommon for space vehicles to reach end-of-life for reasons other than a system failure - one common one being that a satellite or space probe runs out of propellant. Since the underlying mechanism there is predictable, rather than a random failure, there's much less margin needed.

[aziaziazi]

I wonder if the length of all space stuff average out to their acceptable value now ! (fail at launch, rapid unscheduled disassembly and the others going further until exceeding minimum acceptable value)

[wongarsu]

They build and design everything in a way that ensures a 99% chance that after successful launch and deployment it will last for the mission duration in a harsh and still somewhat unfamiliar environment. That happens to translate into a very high chance that it will still work after twice the mission duration, or ten times the mission duration.

Part of this is cultural, part of it is political: nobody wants a failed mission, it's better for the image of the agency and the involved politician to spend a bit more money and underestimate the lifetime. Higher chance of success, and nobody complains if the mission can be extended afterwards.

[bmsan]

For this particular satellite, I think it's actually both. One of the components of the imaging system failed relatively early on[1], but they've worked around the issue for the past 20 years.

1: https://landsat.gsfc.nasa.gov/satellites/landsat-7/

[beerandt]

Because you set a min life, but statistics aside, the design for that minimum life isn't usually something that can be tweaked on a continuous scale, but ends up being binned by design constraints.

Eg, you need an industrial road with a 5-year lifespan over a swamp. To meet this minimum you actually have to build a bridge, which when built to industry standards, might start at lifespans of 20-30 yrs.

Space is a bit different because of budgeting for ongoing operations, so you frontload the cap-x, knowing that asking for addl op-x funds later to extend the program will seem like a no-brainer deal.

Plus sometimes it's as simple as: if you design something to statistically survive space launch, it results in something that is overdesigned to just sit in orbit for years (given that it survives that initial launch).

It's similar to human lifespan statistics- if you get over the historical infant mortality hump, every adult seems 'overdesigned' compared to the historical expected lifespan.

[userbinator]

Space is still a relative unknown, so overbuilding and conservative estimates are far preferred to aggressive cost-cutting and thin margins. You see the same thing with other technologies like cars and appliances -- early versions were mostly very overbuilt.

[noisy_boy]

When smart and dedicated people are allowed time to think, plan and execute instead of being constantly forced to sacrifice everything at the altar of next-quarter results, the quality of the output tends to be better.

[mmooss]

Basic consulting: Underpromise, overdeliver. Nobody knows how long a Mars rover should last; NASA perhaps picks a number so low that they can't fail, and then have another narrative about the amazing little robot that kept going, and which makes the investment look great for taxpayers.

[conception]

The engineering to get it to last a year probably isn’t significantly different from five years, etc.

[bwy]

Lindy effect?

If something is built to last 10 years, it makes it likelier that it can survive another 10.

[callmeal]

>government-owned space things last a lot longer than they're made for

Don't forget that once upon a time, commercial-owned things lasted a lot longer too. Unfortunately the mba's took over and planned obsolescence became a thing.

[Palomides]

all hardware is subject to the bathtub curve