Epistemic Status: Observation, Intuition, Not sure if there is something there
In "Living with Complexity", Donald Norman notes that complexity is inherent in a task and cannot be arbitrarily removed. In the following, I give anecdotes to show that complexity in the real-world is often shifted from a task into the tool used to achieve that task. Thus removing it from the usage/process. Borrowing from probability theory one could maybe describe it as the "Complexity Mass" being moved, where the entire complexity stays the same, but the distribution is changed.
There are tools for many tasks that are either bare bones and need a skilled user, or are simple to use, but complex. In the one case, the task complexity resides in the tool's usage and its user (muscle memory, experience, heuristics, ...), in the other case the tool itself is complex. What I mean by that is, that it contains a lot of interacting parts, and is thus difficult to repair, maintain, and reinvent.
Examples & Contrast
For example, consider traditional woodworking tools, such as chisel, saw, hammer, drill, or nail. Each one of these devices is either very simple or there exists a simple version (manual drill). A chisel needs to be sharpened to be usable, but you can sharpen metal on rocks, and you can prepare rock surfaces, just by using other rocks . A saw is arguably more complex, but still only consists of one or two pieces. If using a saw-string, it is extraordinarily simple.
What is needed to render these tools usable is someone who knows how to utilize them, and who has the experience to know how to saw a particular kind of plank, where to attach the tool, what force to use, etc. Thus: tool simple, usage complex.
Contrast this to modern power tools: router, power saw, nailgun are all much easier to use. A router is much easier to use than a chisel, a power saw makes straight cuts much simpler and a nail gun is much faster than hammer and nail. And the nails are all straight. But this comes at the price of the tools themselves being very complex. Who could reinvent a nailgun? Who could even repair one? Tool complex, usage simple.
Or cooking: The traditional, simple, tools are knives, pots, cutting boards. They require you to know how to cut without endangering yourself, or how to cook rice. But they are simple to make, maintain, and clean. Complex modern tools: Food processor, rice cooker, egg slicer. All of them arguably much easier to use. But more complex in themselves, requiring power, microelectronics, injection molding.
Or in computing: simple command-line tools that can be chained together, but need some knowledge level of the user, on the one hand. On the other hand are complex single-use-case GUI applications, that are easier to use.
So the trade-off we are making is ease of use vs simplicity of the tool and ease of use vs maintainability of the tool. It's also ease of use vs versatility, since simple tools can be used for more, different things. But not only that, but it is also ease of use vs space usage, since fewer, more versatile tools use less space in your kitchen or workshop. Add to that ease of use vs repairability. A knife can be resharpened with a flower pot, a hammers' stick replaced by about anything. How about a nailgun?
I'm sure there are more tradeoffs, but the most important one might just be ease of use vs thinking. Since simple tools have so many more ways of being used, you decide what you can do with them to a larger extent, than a complex tool allows you to. In a rice cooker, you can only cook rice (and maybe a handful of other recipes). A normal pot you can heat to very high temperatures to melt aluminum cans, you can weld a new handle on it, you can wear it as a knight's helmet, you can hammer a nail with it, or you can put it on a gas stove. You can simply do more. By using it mistakenly, you might even discover new things you can do with it.
If you think that "If all you have is a hammer, everything looks like a nail" is scary – considering how it reframes your thinking – imagine what a world of rice cookers' thinking will look like.