Written by Mordecai
on 

Tool Unification & Mental Overhead

Recently, after many frustrations with kitchen timer devices, I caved and bought a cheap, single-purpose tool: A digital display timer. It has six buttons, manages two timers, and does nothing more than that. And I love it, it’s fantastic. I have a much easier time using it than futzing with my phone to enter timers, deal with trying to see which ones belong to what, what I need to pay attention to.

There’s an interesting scale here, and let’s stick with the “things that can time kitchen tasks” for now. It roughly goes, “how much can your tool do besides what you want to use it for?”.

The simplest tool to use for this would be a dedicated hourglass for the interval you want, you could buy a bunch of cheap ones for common durations and use those in the kitchen, the simplest tool that could possibly fulfil your criteria. The next level up would be the classic kitchen timer, a mechanical, spring-wound timer from one to 60 minutes. You gain the feature of variable time length, and lose some ease of use. You now have to invest more effort into using the tool, winding it up to the right amount of force, and dealing with potentially adjusting if you miss the initial target, something that can’t happen with the hourglass.

Next up the scale is what I bought, a single-task device that now has two timers, visual indication (there’s a flashing light for timers that are up), keeps track of how much your current timer has overrun what you initially wanted, and can repeat timers without additional input. The downside is that I now have to contend with buttons, reading digital displays, and mode errors: Depending on which timer the clock selects between, I might be changing the one I didn’t want to modify, so I have to check for that as well.

Next up the list is some form of portal computer, probably a phone, on the upper end, a laptop. Here, you can do arguably whatever you want with the timers, in terms of functionality. Schedule them, spawn them dynamically, use voice controls, have them play Danger Zone once up, whatever.

The downside is that you can do whatever you want with your timers now, and that means you now have to choose. The space of potential actions has grown exponentially, and that’s a space you have to deal with if you want to time the baking time of your cake. It increases the cognitive overhead in dealing with the tool. This overhead is something we can usually afford, but certainly not always.1

Realistically, everyone will have arbitrary defaults they always go to, regardless of what options they might have in theory. The default ringtone, at default volumes, while adjusting the timer, the thing they actually care about. But the larger amount of capability the chosen device has can negatively impede you; for an arbitrary example, if you have a do-not-disturb mode enabled on your phone because you don’t want it to light up like a Christmas tree every time someone sends you a message, it might now also muffle your kitchen timer2, you miss it expiring, the cake will bake for too long and be dry, and everyone will be sad. Nobody wants that.

A tool with less capability, has, consequently, also less potential for errors like this. Nothing about my simple digital kitchen timer can accidentally be configured to not alert once the timer expires. I can mute the noise, but it will still blink a LED at me. My cake timer will not suppress phone call audio, or music, when going off because it does not run on my phone and does not share an audio channel with it.

But clearly, having a million kitchen timers for a million tasks is also bad. The modern phone consolidated and unified roughly a billion (scientific guess) individual doodads and gadgets, from an actual phone, to music players3, to pieces of scrap paper used for notes, to tamagotchis.

Having a tool with you that can help you with what you want to do is better than having no tool with you. It might not be the best tool you could use, but it will still be better than glancing at the clock, trying to maths out when you need to pull out your cake, and then trying to not forget when exactly it was and also not forgetting that you have a cake in the oven in the first place. The individual, single task of timers in the kitchen was valued enough by me to go back and buy a better tool than the default choice for it, my phone.

If you have a lot of simple tools, you have to put in a certain amount of effort first, in picking your tool, before you can go ahead and do the thing you intended to do, whereas I can grab my phone and be pretty sure it will solve my problem in some way, and I can figure it out when it comes to it.

A lot of tools are on scales like this, and it is, for the most part, absolutely fine, as long as the problem gets solved. However, when tools that are easier to use by default get used for things that have a really high floor to them being solved well, we start getting into tricky territory.

The poster child for this is Tesla’s big touch screen as the centre console, rather than having one with tactile buttons, like the rest of car manufacturers. It’s sleek, it’s pretty, it’s elegant, it’s endlessly malleable, it’s easy to update, and it also gifts nightmares to safety engineers for free in its spare time. Even aside the fact that the display controlling a large chunk of your car is buggy as hell4, it now means that you have introduced the same sort of failure mode that we talked about above, where things entirely unrelated to the task of timing cake baking times end up interfering with that.

That aside, this scale offers an interesting space for experimentation: If you find yourself not wanting to use a certain tool for a certain purpose, part of that aversion might be the cognitive interference/noise from being on the wrong level of tool unification. It might be too simple and you now have to make a guess as to what tool to grab for your task at hand, or it might be too unified, and you now have to remember all the things that have nothing to do with your task potentially interfering and correcting for that.

Simple, directed, custom tools are best for repeated actions and known domains, things you know you will need to do time and time again. I’m a big fan of making these specific, sharp tools sooner than later, because they remove cognitive load as soon as I have identified the thing I need them to do. Complex, unified tools are best for when you’re not quite sure what you’ll need yet5, or that the tools required might change quickly. But if you’re not quite sure where your task falls yet, experiment with unification. It might remove some aversion from activities you didn’t know you had.6

  1. Behind Human Error gives the example of an anaesthetic device in an operating theatre: It had too many buttons, too many ways of dealing with things, that the most common way it was used was with as few of its features as possible, because that made it comprehensible under the stress and pressure of the operating theatre. ↩

  2. depending on OS, settings, etc, etc, you see that even in this, there’s a problem here. ↩

  3. Rest in Peace, iPod. You were my first fancy thing. ↩

  4. For a recent example from just a few days ago, look at Tesla recalling 130,000 cars cars because the screen could bug out, and show a blank display. This is, as they say, not ideal. ↩

  5. An example from a realm where “the simple tool” is of nightmarish complexity: Field-programmable Gate Arrays (FPGAs) are used to experiment with integrated circuits and CPUs, things that take months to make and iterate on because each individual example needs an insane amount of resources. I encourage you to read this list of steps on how to make a modern CPU because it is fascinating, and also holy shit we managed to to make this the basis of modern information processing. But now also imagine needing to run through all of these steps when you aren’t even sure yet of what you need. FPGAs are neat as hell. ↩

  6. For example, I avoided setting timers for some stuff when cooking when exhausted, because it meant looking at my phone, which meant dealing with my message notifications. ↩