4 ms·

Why is tech so unreliable? Why does testing and building take so long time that tech is outdated?

ZoFreX 14y ago
Just to play devil's advocate here, is it really /that/ "outdated"? Yes, the camera is modest by 2004 standards, but by 1994 standards it's out of this world! To me, eight years to build and test equipment to fly to Mars and wander around on another planet's surface is a pretty reasonable turnaround time.
thaumaturgy 14y ago
I used to wonder that too. Well, I wondered why nobody could seem to make simple products that worked reliably without any duds and without a lot of cost. Then I got a job with a company that manufactures sensors used by a huge number of companies and governmental departments, including NASA and Boeing. And my job was Output Smoothness Technician -- which meant that I was the one guy who was responsible for verifying the electrical specifications of every single part that left the plant. (The next two steps were QA -- which tested things like watertightness -- and then shipping.) I learned a lot from that job. I learned that there are a lot of decisions behind every single little thing. For example, let's take the tiny little linear potentiometers that Disney uses in its animatronics. Someone at Disney decided they needed them; one team of engineers came up with a spec; another team of engineers figured out how to turn the spec into something that could be made; another team figured out how to make it. Then someone decided what kind of metal to use. Someone else decided how to tune the potentiometer to provide the desired output. Someone else decided what kind of grease to use. And then during manufacturing, someone decided whether or not the housing was good enough, someone decided whether or not everything fit together right, someone decided whether or not the weld was good enough, and so on. By the time that little thing got to me, there were thousands of little decisions stored in it. Some good, some not. Then I had to test it and decide whether or not it would do what the customer wanted. Should I throw it out, wasting the company a lot of money? Or should I ship it, and let the customer decide, and hope the QA on their end is better than me? I pretty quickly developed a reputation as one of the toughest OS techs they'd ever had. I threw back a lot of parts. The sales manager (who basically ran that location) hated my guts. But we still had defective parts come back every month! Now imagine that you're trying to build something that you can't service, and you can't fix. You get only one shot to get it right, and a lot of money and a lot of people are riding on you. Best of all, you're building it to survive in an environment that you just can't really create here on Earth, so you don't get to test it the way that you'd test a lot of things. That's why it takes 8 years and a lot of money: because it's very, very hard. I think that one day sending things to Mars will become something that we're used to, and at that point, it will get a lot faster and a lot cheaper and a lot more reliable. But, right now, we're still trying to do things that we don't really know how to do. That's hard, and it takes time.
- lovskogen 14y ago
  So tech is unreliable because so many people work on different parts, that everything is a fragmented mess? How can one fix this problem?
  - astrodust 14y ago
    Having a large number of people in the chain isn't the problem, it's that no manufacturing process is perfect and defects in design or production are often not obvious until you've tested the item in the field. Further, since it's not practical to test things on Mars, so you can imagine how difficult this can be for the engineers here on Earth. The process described here is typical of many companies because it's not possible for one person to do all of these steps. The level of expertise required is too high. You may be a great mechanical engineer, but do you know about the performance characteristics of the hundreds of different kinds of grease? About the way the sleeve should be machined? There's a list of specializations thousands of items long, and at best you'll be able to truly master only a small percentage of these even through a whole career. The ultimate reason technology is unreliable is because the world is a complicated, crazy place that often exposes you to situations you're not prepared for.
  - HeyLaughingBoy 14y ago
    Let's not say it's unreliable. That doesn't really have much meaning without context: the reliability I expect from my car is much lower than the reliability I expect from a space probe. Instead, let's consider that everything ever built will fail under a certain set of conditions. Understanding a particular component's reliability means understanding all the things that are likely to go wrong over its lifetime and fixing some, and characterizing others. The many people working on different parts is an outcome of different concerns seeing different problems and applying solutions. These solutions are sometimes contradictory: a weak potentiometer shaft may be improved by using a particular alloy of steel instead of, say, aluminum. But now that alloy may make the part too heavy, or too expensive, or it may create problems for the bearing it has to sit in, etc. What you see in the long development time isn't that something is necessarily "unreliable" it's that it takes that long to understand how it can fail and make it reliable enough to sustain the mission. This is already why engineers will tend to use tried and true parts: they already understand them completely.

Just to play devil's advocate here, is it really /that/ "outdated"? Yes, the camera is modest by 2004 standards, but by 1994 standards it's out of this world! To me, eight years to build and test equipment to fly to Mars and wander around on another planet's surface is a pretty reasonable turnaround time.

thaumaturgy 14y ago

I used to wonder that too. Well, I wondered why nobody could seem to make simple products that worked reliably without any duds and without a lot of cost. Then I got a job with a company that manufactures sensors used by a huge number of companies and governmental departments, including NASA and Boeing. And my job was Output Smoothness Technician -- which meant that I was the one guy who was responsible for verifying the electrical specifications of every single part that left the plant. (The next two steps were QA -- which tested things like watertightness -- and then shipping.) I learned a lot from that job. I learned that there are a lot of decisions behind every single little thing. For example, let's take the tiny little linear potentiometers that Disney uses in its animatronics. Someone at Disney decided they needed them; one team of engineers came up with a spec; another team of engineers figured out how to turn the spec into something that could be made; another team figured out how to make it. Then someone decided what kind of metal to use. Someone else decided how to tune the potentiometer to provide the desired output. Someone else decided what kind of grease to use. And then during manufacturing, someone decided whether or not the housing was good enough, someone decided whether or not everything fit together right, someone decided whether or not the weld was good enough, and so on. By the time that little thing got to me, there were thousands of little decisions stored in it. Some good, some not. Then I had to test it and decide whether or not it would do what the customer wanted. Should I throw it out, wasting the company a lot of money? Or should I ship it, and let the customer decide, and hope the QA on their end is better than me? I pretty quickly developed a reputation as one of the toughest OS techs they'd ever had. I threw back a lot of parts. The sales manager (who basically ran that location) hated my guts. But we still had defective parts come back every month! Now imagine that you're trying to build something that you can't service, and you can't fix. You get only one shot to get it right, and a lot of money and a lot of people are riding on you. Best of all, you're building it to survive in an environment that you just can't really create here on Earth, so you don't get to test it the way that you'd test a lot of things. That's why it takes 8 years and a lot of money: because it's very, very hard. I think that one day sending things to Mars will become something that we're used to, and at that point, it will get a lot faster and a lot cheaper and a lot more reliable. But, right now, we're still trying to do things that we don't really know how to do. That's hard, and it takes time.

lovskogen 14y ago
So tech is unreliable because so many people work on different parts, that everything is a fragmented mess? How can one fix this problem?
- astrodust 14y ago
  Having a large number of people in the chain isn't the problem, it's that no manufacturing process is perfect and defects in design or production are often not obvious until you've tested the item in the field. Further, since it's not practical to test things on Mars, so you can imagine how difficult this can be for the engineers here on Earth. The process described here is typical of many companies because it's not possible for one person to do all of these steps. The level of expertise required is too high. You may be a great mechanical engineer, but do you know about the performance characteristics of the hundreds of different kinds of grease? About the way the sleeve should be machined? There's a list of specializations thousands of items long, and at best you'll be able to truly master only a small percentage of these even through a whole career. The ultimate reason technology is unreliable is because the world is a complicated, crazy place that often exposes you to situations you're not prepared for.
- HeyLaughingBoy 14y ago
  Let's not say it's unreliable. That doesn't really have much meaning without context: the reliability I expect from my car is much lower than the reliability I expect from a space probe. Instead, let's consider that everything ever built will fail under a certain set of conditions. Understanding a particular component's reliability means understanding all the things that are likely to go wrong over its lifetime and fixing some, and characterizing others. The many people working on different parts is an outcome of different concerns seeing different problems and applying solutions. These solutions are sometimes contradictory: a weak potentiometer shaft may be improved by using a particular alloy of steel instead of, say, aluminum. But now that alloy may make the part too heavy, or too expensive, or it may create problems for the bearing it has to sit in, etc. What you see in the long development time isn't that something is necessarily "unreliable" it's that it takes that long to understand how it can fail and make it reliable enough to sustain the mission. This is already why engineers will tend to use tried and true parts: they already understand them completely.