So, a few years back a friend of mine used the phrase “Never buy an architect’s second house” as we were discussing the features of a new CNC machine.
I asked him what he meant and he described a mindset and result that I had seen several times, yet did not have a name for.
You see, when an architect gets that first house, they are very careful and conservative. They want to get it right. But, the entire time, they are bubbling over with ideas. Things they could add or design to do more things. But, they are too scared to really try on that first house. The first one is overwhelming just in learning the details of how to build that house.
But, all those ideas are there. Then, comes the second house. That is when the architect tries to add all those ideas. And, then they find the house getting new features and design quirks that clash with each other. The costs go up. The functionality and design begins to clash. The house just does not really go together. No rug will tie the room together.
After that lesson, the architect dials it back a bit and eliminates good ideas that are really only good ideas in some designs and not others. Eventually, the architect reaches the point where the design and functionality mesh well together. This could actually happen on the second or third, or even a little later, but the pattern is almost always the same. There is a steep ramp in ambitions followed by a rather glaring problem, then a ramp back down in what is being attempted. Generally, the more revolutionary the attempt or industry, the further versions are deployed before the crunch. So, a well-known industry like home building will hit early, but a new technology will hit later. Usually in the tech field, it manifests itself in rhetoric promising the world, then a product that sorta does some of what was talked about.
One of the more notable example comes from NASA. After the loss of the Mars Observer mission, NASA moved to a strategy of smaller less costly missions that were complimentary. So, if a single satellite was lost, we would have others that could still function and gather data. The first mission was Mars Global Surveyor. This was mostly built with existing parts left over from Mars Observer. The same sensors and the like mounted on an off-the-shelf chassis. The second mission was the first one to use really new concepts. That was the Mars Pathfinder mission. Its mission firsts were the rovers and airbag delivery system. The mission was a great success and still shows up in pop culture references.
Then, NASA got ambitious. Instead of one mission for the next launch window, they went for three missions. There would be a space probe called Mars Climate Observer, which would orbit the planet and relay back data. The Mars Polar Lander was a second spacecraft. It’s mission was to land in the Mars polar region. The third mission comprised of two probes that were added to the Mars Polar Lander. These probes were to be cut loose from the lander and they were designed to come crashing down into the surface and then try to record sub-surface conditions.
All three were complete failures,
Arguably, the Mars Climate Observer was not lost due to ambition. It was lost due to an error in coding. It might have been caught in a simulation or might not have.
But, the landers were another matter. The penetration probes had very high engineering requirements and a limited budget. There was not a lot of room for testing. They were also on a time crunch to be ready for the next launch window. There was no communication established with the probes, so no specific cause was identified as to their loss, but dropping a probe from orbit into an unknown conditions at high speeds is not a low risk proposition.
As for the Mars Polar Lander, the most commonly accepted view is that the shipboard computers interpreted the vibration sensors data as indicating the spacecraft had landed while it was still in the descent phase. One reason they think this is a likely cause is because they actually saw this behavior during testing. They launched anyway.
The next launch window saw only one mission. It was an orbiter with three primary sensors and a communications relay to be used in future missions from the surface.
The one after that went back to being fairly ambitious, but they dialed back the amount of new things they would try. NASA launched 2 landing missions to Mars in the next launch window. The landers were scaled up pathfinders, but were aimed at areas that were better mapped and where progress could be monitored. The landers had a much larger budget ($850 million for 2 landers vs $110 million for the polar lander). They were also under less of a time crunch.
They matched their budget to the mission. They limited the number of risks to within reason. They scaled from known working technology and they tested it more thoroughly. Those missions were immense successes.
Guess the lessons here might be to scale from known to unknown in increments that constitute an acceptable risk. Know what you can deliver and then deliver that. And something well designed can often be in the field long after it’s expected expiration date. Don’t think because you have done something once that you can do everything at once.