Author, YouTuber, Script Writer for Rational Animations. A.B. in Math (Harvard 2020)
aggliu
Karma: 472
Bears are wild animals. I think it would take way too much effort to get a large enough consistent supply of autumn bear fat even for a food truck, especially given that people probably wouldn’t pay for one cracker’s worth at a time.
Fine then, let’s use beef tallow. We could sell jars of beef tallow mixed with honey and salt as some kind of paleo peanut butter alternative and branch out from there. I think plenty of people would enjoy it, though I think it would be hard to convince the kind of people who love beef tallow to buy it in a jar from us rather than make it themselves.
The base rate of packaged food startups failing within the first year is 90%.[1] If I had to bet based on taste alone, I would expect a honeyed tallow startup to have a better shot than average, but my guess is that marketing would have to be heroic.
There is some bear fat left should you happen to be at Lighthaven before Thanksgiving, or you can buy some on your own from Borderline Tree Works—the smallest size is $8 plus shipping. I would also really love to see your opinion—oh glorious leader ;)
The people who liked the bear fat snack the most really seemed earnest, including going back for more and raving about how good it was (particularly Person A). Gwern compared it to pemmican in that it grows on you as you eat more, and did start eating it with a spoon.
In case it helps, here are the full results of the survey:
Person A, who requested their name be removed: Bear fat snack 10, Ice cream 10, comment: “yum”
Anonymous: Bear fat snack 9, Ice cream 10
Skyler: Bear fat snack 10, Ice cream 7, comment: “The salt was the best component”
Anonymous: Bear fat snack 7, Ice cream 9
Jenn: Bear fat snack 8, Ice cream 6
Eneasz: Bear fat snack 7, Ice cream 9, comment: “i love unique experiences, thank you!”The main reasons I would still doubt the broader applicability of these results are:
The people who took the survey are all people who decided they wanted to try this AND stuck around long enough after tasting it to fill out the survey, so it’s a super biased sample. Even then, 3 preferred ice cream and only 2 preferred bear fat, with 1 tie
People may have enjoyed the bear fat snack for its novelty, and might not choose it over ice cream if it were commonplace
I was asking people to compare the actual taste of bear fat to their memory of how ice cream tastes, and maybe ice cream (or food generally) tastes better in the moment
The bear fat snack was set up as a fancy treat with Maldon salt flakes and local raw honey, and the survey takers may have been comparing it to run-of-the-mill ice cream instead of the best or fanciest ice cream
To that last point: I had a mass produced ice cream sandwich about an hour later and it was maybe worse than the bear fat snack, but hand made ice cream from my local scoop shop tastes much better to me.
Gwern started eating it with a spoon, so I’d say yes
Can do. Do you think I should I edit it into this post, or should I make a separate one?
I ate bear fat with honey and salt flakes, to prove a point
This comment confuses me:
Is the spoiler block supposed to be empty, as it appears to be?
Just an idle thought, “design@intelligence.org” sounds like a creationist mailing list.
When will AI automate all mental work, and how fast?
The biggest piece of evidence that I’ve seen is the Emergent Misalignment paper, linked in Gurkenglas’s parallel comment to this one.
This is probably well-known at this point, but the $6 million item is not the duct-taped banana itself, but the right to display the duct-taped banana. That may not make it any more or less sensible, though.
Yep, that’s why I mentioned evil numbers specifically.
I am a bit worried that making an explicit persona for the AI (e.g. using a special token) could magnify the Waluigi effect. If something (like a jailbreak or writing evil numbers) engages the AI to act as an “anti-𐀤” then we get all the bad behaviors at once in a single package. This might not outweigh the value of having the token in the first place, or it may experimentally turn out to be a negligible effect, but it seems like a failure mode to watch out for.
Here’s a link: https://borderlinetreeworks.com/bear-grease