46 Comments
User's avatar
Philosophy bear's avatar

I need to know how obsessed about Jaffa cakes before I answer the question. If they spend like 10% of their time talking, talking about Jaffa cakes, that's a cute quirk. If they spend 50%, that's starting to noticeably reduce the richness and variety of human life.

Expand full comment
dynomight's avatar

Maybe akin to a 75th percentile football/soccer fan?

Expand full comment
Fsmor;oid's avatar

...now I want some jaffa cakes...

Expand full comment
Sampsa's avatar

When I try to load the questionnaire (via the link https://cryptpad.fr/form/#/2/form/view/9u70EvBmvxs+wFOBk8yYWNLOTbpBzfZeDA+Eg1vuAZ4/ ), it flashes some form and then says: "This account was deleted because of inactivity"

Expand full comment
dynomight's avatar

I've gotten a bunch of these reports—seems to be an issue of cryptpad not coping well with weird browser configs. (Annoying because the reason for choosing cryptpad is to avoid such issues...)

Expand full comment
Victualis's avatar

I think for most of these questions the overwhelming majority answer would be "other" or "don't know". Are you intending for us to normalize the percentages so that the undecideds are excluded? I answered based on raw percentages so mostly close to 0. If I had normalized I would have given a completely different range of estimates.

Expand full comment
dynomight's avatar

I guess I'd like you to answer as if people were forced to pick one or the other. Among other reasons, even presenting an interface for you to specify a distribution over three options instead of 2 makes things much more complicated.

Expand full comment
Autumn Gale's avatar

FYI I answered some questions with 0% meaning "I think 0 is closer to the actual number than 10 is", rather than "I think there are literally 0 humans who agree". If I were allowed to answer freely I'd have put 1-3% for some things.

Expand full comment
dynomight's avatar

Great, that's as intended. (I'd have liked to give the ability to answer any number, but cryptpad apparently doesn't offer that.)

Expand full comment
Steve Newman's avatar

I'm very curious to learn where you're going with this.

And FWIW, I came close to abandoning the quiz on the Jaffa cakes question and then did drop out when I got to Gliese 65. I bounce off of questions like these because, like... exactly how obsessed is "really obsessed"? Do we understand the long-term effects of weirdly, artificially obsessed with Jaffa cakes? Is it even plausible that "there are no other effects"? I don't really understand how you could take a person who in general does not have obsessive special interests and make them (permanently?) obsessed with Jaffa cakes without altering them in any other way, so I either reject the premise of the question or recuse myself as unequipped to answer it.

As the questions get weirder, I feel this more strongly – the premises are (much) too far outside of my experience for me to engage with them as anything other than a Rorschach test.

Expand full comment
Alexander Kaplan's avatar

The Jaffa one was difficult for me too. At first I thought my answer was "no" because permanently changing someone's inner life without their consent goes against my moral intuitions, but then I figured, you know, that's just what memes (in Richard Dawkins' original definition) do, and ten percent more happiness/life is pretty substantial, so I threw my moral intuitions out the window.

Expand full comment
Isaac King's avatar

I suspect the results of this poll are highly context-dependent. Ask someone "is it unethical to torture animals" and most will say yes, then turn around and pay someone to do exactly that, and even oppose laws that prevent it.

This makes it difficult for me to answer these questions, since I don't know what you're actually asking. Are you asking "how do you expect other people to respond to this question exactly as I have presented it here", or are you asking "how many actually hold this belief"?

Expand full comment
Harold Godsoe's avatar

Something about the page formatting makes saving results quite hard. Printing (to PDF) doesn't capture beyond the first page and no circles are checked. Copy-pasting to a document doesn't preserve the checks. Screenshotted the thing in several steps, but this is less than ideal.

Expand full comment
cmart's avatar

Responded but had difficulty reasoning about "Western adults". I didn't know whether "Western" includes Latin American countries (which, combined, have a larger population than the US + Canada and somewhat different cultural values).

Expand full comment
Caro's avatar

Morally correct vs morally permissible had me frustrated.

Expand full comment
Jack Whitcomb's avatar

The probability-weighted bliss lost by the antiquarks is only 1.2x as large as the bliss gained by humans, so you don't actually have to show much favoritism toward humans to agree that the Dyson spheres should keep getting built. I'm assuming it's deliberately set up this way for people to notice if they do the math? Also, please don't put me in one of those machines!

Expand full comment
somervta's avatar

I thought this, but did inititally wonder if the % part was a mistake and it should have been 120x

Expand full comment
Julia D.'s avatar

I stand by showing a lot of favoritism to humans. But I also noted that the humans in this question a) explicitly instructed the AI to safeguard non-human conscious bliss, and b) aren't living very human lives themselves.

Expand full comment
Leo's avatar
Jun 17Edited

Sigh. Survey Making techniques... Maybe it is just me, but I had great difficulty with the "Do I personally agree" part. Probably because I was going fast, but I kept saying to myself "Agree with what?" Am I agreeing to match my guess at the percentage or agreeing to match my muddled memory of the moral given in the stem.

I would have appreciated it if question that explicitly restated the moral question: e.g.: "Do you personally agree that the mother should be legally allowed to abort?"

Cheers,

Expand full comment
dynomight's avatar

Yeah, unfortunately I've heard this from a LOT of people, so you're definitely not alone. I should test more before releasing!

Expand full comment
Chris Terrapin's avatar

So our agreement is with the original choice, not with the "percentage of adults", correct?

Expand full comment
dynomight's avatar

Yes, that's right.

Expand full comment
Chris Terrapin's avatar

So our agreement is with the original choice, not with the "percentage of adults", correct?

Expand full comment
Kenny DAmica's avatar

These are so weird. Answering them brightened my day therefore it was morally correct to create them.

Expand full comment
Vasquez's avatar

I chose to reject the cake addiction but once I saw what it looked like, I changed my mind.

Expand full comment
DavidI's avatar

For info, both my wife and I misunderstood at least one question and the answer we gave was, on reflection, the opposite of what we meant.

Expand full comment
lizzard's avatar

Can non-Westeners fill it in with our perceptions of what Western adults are like? Or will that skew your results?

Expand full comment
dynomight's avatar

Of course! (I assumed many people answering would be non-Westerners, but I made the questions about Westerners for reasons.) Just answer the questions literally—guess what percentage of Western adults would support and then answer for yourself.

Expand full comment