I may be exposing my own subjectivity here, but I, personally, did it because Randall asked nicely and I like the things that he does.

Or as an alternative theory, Randall Munroe wants to know if it's OK to tell everyone when he throws away all his socks and buys a bunch of identical ones.

This seems strangely plausible.

It is the one way to eliminate the problem of matching socks after laundry.

I kept going because the questions were entertaining.

That's what I though, at least while reading the "type 'cat'" one.

I thought about it for a while, and then it occurred to me, that there are actually many ways one could possibly understand that -- thus replying e.g.:

- cat

- "cat"

- 'cat'

- panther (not exactly a cat, but kinda -- that's why it was in quotes, no?)

- [one could type "cat" in one's terminal]

- "Whiskers", etc. (i.e. name of one's cat)

then, one could make a typo (thus e.g. "ca" or "catt"), or finally something more or less totally random, given that it's an open question (thus e.g. "no", or "How are you?", or whatever).

So, not that stupid a question I thought initially :)

In a bit of "don't tell me what to do" pique, I typed the cat emoji, .

Turns out something in Google Forms uses UCS-2 internally, so it rejected my whole survey.

EDIT: I guess Arc isn't very good at this either.

> Turns out something in Google Forms uses UCS-2 internally, so it rejected my whole survey.

Google love Java to bits, and it's UCS-2ish.

My phone was going put 'Cat', but I corrected it to match 'cat'. I'd be curious to know how many people on smartphones would do the same.

Took it on a smartphone. Also overrode to match case. So, at least 2.

I just pasted a picture to heavy machinery.

I included the quotes, just to be cute.

I included the quotes and then typed here with a colon after it.

not to mention the obvious ones

- Cat

- Cat.

That's probably a trap question, to let you filter out people who aren't paying attention. Similarly, in the 'recognize words' section, I'm fairly sure some of those words aren't real and so you could increase data quality by throwing out anyone who claims to recognize fake words.

I took my time to answer everything, but didn't type in 'cat'.

Here are some obvious signals that I think can be pulled from the data set:

• gender

• age (since they just straight up ask for it)

• rough geo data (if the survey records IP geo information, so assume they are doing that)

• economic status

• rough tech literacy level

As far as I know, it's not possible to record respondents' IP addresses with Google Surveys, so I think that one's out. I suppose you could make rough inferences based on the food, weather, and snow questions.

The keyboard part will help quite a bit, at least. Being from Norway, the occasional æ-ø-ås are quite revealing.

I type on Dvorak, so lots of "oeuieu" (clusters of vowels) in there. Midway through the gibberish I typed "this is a dvorak keyboard" just for grins.

I would've stopped if the questions were boring.