Making new

• Making something new requires simpler processes that are easy to automate.

• Making something new may involve the same series of steps done many times, so one can take advantage of economies of scale. You obtain raw materials or parts, and they’re all handled in the same way, many times over.

• You only need to deal with assembly, not disassembly.

Repair

• Devices don’t fail in the same way, so repairs tend to be unique and not amenable to taking advantage of economies of scale.

• Unless you repair a device in the same way multiple times, the scale is necessarily going to be less than manufacturing new, since if you manufacture N devices, you can’t be repairing more than N devices; again, not friendly to economies of scale. Even with things like automobiles that are designed with the intention of being easy to repair, older cars become increasingly less-practical to repair as the pool of cars of a particular year and model shrink over time as some become unrepairable and head to junkyards.

• If repair requires components, those components may need to be manufactured and then warehoused until repair is required, so the storage cost also adds to the cost of repair.

• Repairing is a complex process that likely differs from device to device that is hard to automate.

• Repair involves not just reassembly, but also disassembly.

• Especially if a device was not specifically designed to be simple to repair and especially if scale is not high, repair may involve (expensive) skilled labor from someone who has to be able to craft a specific repair process for this particular device being repaired.

• Repair involves diagnosis of the problem. Diagnosis may be an extraordinarily difficult process — e.g. trying to diagnose a failure inside of a chip without destroying it is something that we may not be able to do today, and the skillset or automated system required to do it may be very complex. Intel spent ages just trying to understand why there were failures in the last two generations of their chips, a situation where they didn’t care at all about destroying chips that they’d use for diagnosis, and they weren’t trying to repair a single chip, but to fix a process that involved huge numbers of chips. Maybe an electrical engineer could diagnose a problem on a device, but his work will repair only a single device. The time of that same electrical engineer could be used to improve a manufacturing process that could produce many more new devices.

Other

• When you repair a device, you get a device which may have other worn components. At some point, something else will fail. If you manufacture a new device, you get all new parts; you “reset the clock” on everything.

• When you repair a device, you get an older device. In many cases, due to the advance of technology, a newer device is preferable. That may not always be true — consider, say, an antique made by a specific artist centuries ago, where the value is in part that that particular person made the thing. But for most functional purposes, something made using present-day technology beats stuff made in the past. And that’s a more-important factor the greater the rate of advance in the field of whatever good it is that you’re trying to repair.

• Repair may compete with recycling.

We live in a world that, due to global trade and large population and probably access to interchange languages, has far more potential for scale than ever before in the past, so economies of scale can be pretty important — design a device once, make a very efficient process for making it, and you can sell to many, many people. Billions of people, even. You couldn’t do that a few hundred years ago, because the world was simply too disconnected. That’s a lot of potential for economy of scale. And economies of scale are, I think, generally more-friendly to manufacturing new.

I can’t think of many factors that would cause repair to become more-efficient versus manufacturing new relative to where they are today. I think that when we get AGI, we may be able to reduce the cost of skilled labor and complex automation, both used in repair, so far that repairing could see a bit of a renaissance. Maybe if we become a multiplanetary species in the future, travel into space and live elsewhere, then we’ll have small populations that are mostly cut off from the rest of the population, and economies of scale will greatly decrease, and repair will become more worthwhile — you fix something on Mars because there aren’t enough people on Mars for building new to make sense, and sending a new item from Earth costs too much. Maybe we’ll crash into fundamental physical limits and the rate of technological advance in many fields will slow way down, so a newer device won’t have many more benefits over an older device. Maybe some sort of new types of goods that are fantastically-expensive and only required in small scale will become increasingly important, and for them repair will be more important than for goods that are produced at large scale. But outside of that, I think that most of the factors will favor manufacturing new, and if anything, probably continue to do so even more than they do today.