Thread by @notionparallax

We developed a method to work out how #toilets really get used. This sounds silly, but it's got some big consequences. Here's a thread about how to do it, and what it means. Perfect for reading with your breakfast!

Our paper was just published in Architectural Science Review, so if you want all the details, go there. tandfonline.com/doi/full/10.10…
@nicolelgh @BaptisteHiggs @aidenkray @ishaan_varshney did so much important work on this. There is NO WAY that this could have happened without them.

There's 2 parts to this, how people use urinals, and how people use cubicles. (Or whatever you call the enclosed bit of a public toilet, stalls, private office...)
Let's start with urinals:

Lots of people have talked about how people pick a urinal. Some people (e.g. @xkcd) have even modelled it with... MATHS! But as far as we can tell there's barely any measured data behind it.

In 1975 Edward Reid & Patricia Novak, as undergrads, did a study where they hid in a cubicle, looked through the crack in the door and made notes. (This might have been easier in the USA because the door cracks are HUGE!)

As far as we can tell, that's it for urinal scholarship at this level of granularity.

The whole measuring people going to the toilet thing really stems from Sandra Rawls' PhD, ahd her method is still being used. ⏱ Stand outside the bathroom with a stopwatch. There was a paper in 2019 where the big innovation was more stopwatches >⏱⏱⏱

There's a little bit that was collected by bathroom god, Alexander Kira, but it's all measured from outside the whole room and it's all about time, not human interactions or other behaviour

So we wanted to see if we could answer the questions about how people pick urinals.

Turns out that when you have a nutbag who asks strange questions (me) and three incredibly gifted nerds: @aidenkray on the electronic eng, @ishaan_varshney doing the computer science, and @BaptisteHiggs doing the all round boy genius stuff, those questions get answered!

Using a GridEye sensor meant that we could get a signal that was low resolution enough to preserve privacy (which is SUPER IMPORTANT) but had enough detail to do some useful processing with. This is what we get out of the sensor:

L–R @aidenkray @BaptisteHiggs @ishaan_varshney did some very important calibration work with hot cups as targets

And it turns out that the mathematical models are basically right, at least in the context of a 3 urinal setup.

After some hard maths, @aidenkray processed the data into this state change/markov/state machine graph. It's a bit hard to understand. The numbers on the arrows are probabilities and always sum to 100%

(Some of these middle slot times are me using the middle urinal just to see what happened)

But why would you want to know this? Well, knowing things is pretty good for it's own sake, but a big enough dataset of this stuff would have a really profound impact on how urinals are designed and laid out. (Assuming we keep using them at all)

The work happened while @BaptisteHiggs @aidenkray @ishaan_varshney and I were working at @ArchitectureBVN. So thanks to @Mcfblair for letting us do something that looks so silly on the surface, but is ultimately so important.

This is already a mega thread, so I'm going to start a new one for part two, how to understand the cubicle. Urinal users are less than half of toilet users, so it's way more important to understand cubicles.