Resolves as YES if Eliezer Yudkowsky, founder of the LessWrong forum, prominent advocate of the idea that AI will cause the extinction of humanity, and perhaps the founder of a new apocalyptic religious movement (https://web.archive.org/web/20230707061904/https://a16z.com/2023/06/06/ai-will-save-the-world/) is irrelevant by 2030.
Decision criteria: Go to Google Trends and search for 'Eliezer Yudkowsky'. Set the search window from 2004 to present. Set the area to be 'United States.' If the December 2029 trend score is less than or equal to three times the Feb 2022 trend score, then this market resolves as YES.
If Yudkowsky has been turned into paperclips by 2030 by an AI, claim resolves as NO on grounds of his martyrdom.
[If Google Trends ceases to exist by 2030, I will use the same algorithm on an equivalent service if it exists. If there is no equivalent service, then I will use my best judgment.]
[Modification to decision criteria: Google Trends doesn't seem like a reliable way to decide this. I'm selling all of my shares and opting out of this market because the decision criteria are going to become a whole lot more subjective. I will decide this and explain my judgment in 2029. For reference, in Spring of 2023 I think Yud was very influential and his ideas were relevant. In Spring of 2024 I think he's on the edge of irrelevance. As I write this, this market is trading at 7% -- which is way, way, way too low, because most semi-famous or influential people become irrelevant over any 7 year period.]
1,000
1.00Can you elaborate a bit more on the subjective piece and how you would define relevance? Also the original resolution criteria would have have been an indicator that he is less relevant than he used to be and not necessarily that he is irrelvant. He will probably be more relevant than the average person till he dies.
Ok, this is what irrelevance looks like to me.
The year is 2029.
A NYTimes writer is penning an article about AI safety in general. The piece will be 2000 words. The author splits his time 50/50 between NYC and the Bay Area, covering tech topics as his beat. He can send a text message to Elon Musk and he at least gets a "read" receipt under the blue bubble, even if he doesn't get a response. The writer contacts ten people for comment on the article. Yudkowski is not one of them.
I encounter a MIT grad, fresh out of school with a CS degree. ChatGPT4 came out when they were 17 years old. They can remember a time before LLMs, but it seems like that was a long time ago. The grad does not use Manifold or read lesswrong.org. However, they do read hackernews daily. They have a job lined up where they will do AI work. I ask them whether the name Eliezer Yudkowski rings a bell. They say, "Hmm. Sounds familiar. Who is that? Didn't he resign from OpenAI in 2024 because of safety concerns? Wait, no, that's Ilya Sutskever. No, I'm not familiar with ... Yanovski, was it? Why do you ask?"
A guest on Russ Roberts' EconTalk is discussing AI safety. The guest has written a book about the topic. The podcast is 120 minutes. But neither the host nor the guest mention Yudkowski by name during that time. They don't mention paperclips either, but in passing they do recall, with a laugh, that people suggested banning LLMs or attacking LLM data centers to protect humanity back when ChatGPT 3.5 came out.
Meanwhile, Yudkowski remains very relevant to people who move in the same circles -- Effective Altruism, LessWrong, ManifoldMarkets, the International Campaign to Prevent The Arrival of Roko's Basilisk, whatever. But that's not enough to get this to resolve as YES.
@OlegEterevsky Google changed their algorithm. The Feb 2022 trend score reads as 0 for me now if I limit the search to the United States -- which obviously doesn't work with my decision criteria. It wasn't 0 when I created this claim.
However, if I set the criteria to "Worldwide," the score is 2, which I can work with. This means that if the Dec 2029 trend score is 6 or less, then this resolves as YES.
I'm N/Aing this. There's not a good way to objectively measure whether Yud is irrelevant.
People remember the name "Alvin Toffler" even though he's irrelevant. Fame is not a good proxy for relevance.
Sorry to those who tied up their capital in this claim. I think N/A'ing this is the right thing to do.
@HarrisonLucas Wait what? You can't N/A a market anymore???
Well, I'm going to edit the decision criteria.
(1) I'm selling my shares at a loss.
(2) I won't participate in this market.
(3) I'll adjudicate this mostly subjectively and explain my reasoning.
Am I stupid or does this not make any sense "If the December 2029 trend score is less than or equal to three times the Feb 2022 trend score, then this market resolves as YES."
So if the interest in him grows by 3 he will be counted as irrelevant? I would say he is not irrelevant now, he has a ted talk and was in the Times, if interest grows by 3 he clearly would be relevant, no?
@Timothy Take a look at the Google Trends graph and it will make sense then. I consider 3x his Feb 2022 score to be 'Irrelevant.' It's a poor proxy for relevance but it's measurable and it's positively associated with relevance, so that's the best I can do.
@HarrisonLucas seems like 3x his relevance then is just his relevance now? (Far below his may 2023 relevance). So this question is equivalent to "will Yud have more relevance than today"
@Uaaar33 Sure, but you've made a sign error: purchasing NO means you are betting that Yud will have more relevance than he does as of May 15 2024.
Looking at the trend line, it looks like his 15 minutes are up. I'd be relieved, personally... fame must be awful.