🏅 Top traders

Canceling a bunch of personal markets due to the pivot. They don't really fit with real-money trading.

These should all NA, but I couldn't do some because of negative payouts so I did market.

Link your Twitter and Discord and make an LLM clone of yourself!

Here's what mine would probably sound like:

The average Manifolder is so boring it's taken until 2024 for LLMs to get powerful enough to imagine them as interesting characters. Nonetheless it is now possible to infer a character sheet for you from your posts and generate synthetic data to finetune or prompt an LLM to act as an exaggerated version of you.

I don't have enough money to make linear interpolation markets anymore, so I'll add cheap M25 options like this.

My LLM benchmarks idea is in this tweet:

@Mira This one's a keeper guys.

"This model has a 140 AIQ - Artificial Intelligence Quotient"

"AIQMetrics.ai"

Rolls right off the tongue.

Everybody HATES doing benchmarks. The average Manifolder can barely read, much less estimate the precise difficulty of 50 different problem types relative to others. So such problem is impossible for nearly anyone except me to solve.

"130 IQ with 15 standard unit" means 2 bits of selection bias on the uniform distribution to cause a 50% rejection rate.

Calculated over a population of problem instances(which are themselves calibrated) so that people don't upload 1000 dumb llama finetunes and give GPT-4 an infinity IQ.

Defining it as bits of rejection sampling naturally scales to groups/joint distributions too.

You guys are going to be begging to work for me in a year, and I'll have to reject you all because basic literacy is a requirement.

@Mira I'm a prodigy, ahead of the curve. I'm already begging to work for you, and I'm also already too illiterate to be accepted. I'm STREETS ahead of the other scrubs

@Tumbles That's the spirit! Keep the begging up and you'll one day earn an undeserved collar of your own.

Getting a job - https://manifold.markets/Mira/if-mira-gets-a-job-anywhere-how-muc

This option should be a lot lower. Like ⬇️20% or ⬇️30%. I have a goal for $1 million in 2024, and $500k is about what I could expect from a job search. So that's not going to be enough.

I know there's /Mira/if-mira-applies-to-openai-will-mira , but you should understand that's an exception and not a general principle that "Mira is someone that regularly works at and applies to jobs".

Trading Real Money Prediction markets

I guess it's reasonably priced since it only requires transferring the money. But there's not actually enough liquidity to make it worthwhile, especially since prediction market contracts are non-marginable. But maybe I transfer some in just to keep the account active in case there's another LK-99.

Custom AI Art Models - https://manifold.markets/Mira/if-mira-starts-selling-custom-stabl

Wouldn't rule it out. ⬇️40%

Virtual Staging for Real Estate - https://manifold.markets/Mira/if-mira-starts-a-company-virtually

⬇️30%.

AI Girlfriends - https://manifold.markets/Mira/if-mira-invents-ai-girlfriends-how

Extremely unlikely to start something. Someone did message me about a startup they were setting up after the Pixar guy got popular on Twitter. It's theoretically possible I could contribute something just for fun for them and it would qualify? ⬇️40%, which I guess it already is.

Smart Contract Auditing - https://manifold.markets/Mira/if-mira-builds-cryptocurrency-smart

A job is more likely than this. It would make okay money as a side gig, but filling my head with Ethereum VM proposals for the last couple years to get caught up again so I can do audits would not be useful for anything else. ⬇️10%

Deepfaking Platform - https://manifold.markets/Mira/if-mira-builds-a-deepfaking-platfor

Same thing as the custom AI art models, just marketed differently. ⬇️20%

AGI - https://manifold.markets/Mira/if-mira-works-on-agi-how-far-will-i

⬆️85%

Virtual shares in Startups - https://manifold.markets/Mira/if-mira-creates-a-market-for-virtua

⬇️5%. Same problem as the cryptocurrency auditing but worse. I would probably do a decent job at designing the market mechanism, but honestly this is mostly just sales.

Better AI art models - https://manifold.markets/Mira/if-mira-works-on-ai-art-models-with

⬇️15%. It would have to be paired with one of my other 2 AI art site ideas, such ideas would have to be very successful, and then it could plausibly make sense to do research projects on AI art models.

So it's not a bad idea, just conditional on a big conjunction.

AI Model API - https://manifold.markets/Mira/if-mira-starts-a-company-wrapping-c

⬆️25%. I know I said I wouldn't do it. But the prerequisites are so low and it naturally grows out from several other ideas(like if I'm serving Stable Diffusion generation services, maybe I also sell an API). So I wouldn't set out to do it as my primary thing; but it could happen.

Mirafold Prediction Markets - https://manifold.markets/Mira/if-mira-starts-a-new-prediction-mar

This seems fairly priced at 6%. Someone did ask, "Mira maybe Manifold would pay you to design market mechanisms and margin systems and things", but that wouldn't count for this option and also that sounds unlikely now and also the pay wouldn't be good.

BTE's got his thing too, which is an option. I'm very unlikely to actually do anything with it. Haven't looked at the code.

Market design

A lot of these markets aren't getting very many traders and I'm a bit short on mana now. I could print another million mana anytime if I needed it, but maybe I should've made binary threshold markets instead: "Will X idea by at least Y successful?" rather than "Given X, what is Y?"

The concern with the binary markets is they just have too much leverage at the endpoints. It feels weird even trading them because who cares if you have 999 users or 1000? They should be the same and thus the payouts should be the same. Hence my interpolation markets.

But I think these failed not because the descriptions are too long(my Sudoku market has a long description), but just because the mechanism asks a distribution rather than a direction and maybe Manifolders have trouble eliciting a liquidity curve from their mind to bet with.

Do you guys still want more options to bet on? I could add another dozen ideas.

@Mira

filling my head with Ethereum VM proposals for the last couple years to get caught up again so I can do audits would not be useful for anything else

😂😂😂

Here's something I'm doing right now that you can bet on:

This option is what I would work on with infinite money. Chances are low I get anything interesting, so there's a real risk of spending a year screwing around with evolutionary algorithms and having nothing to show for it at the end. But, when I look around at the rest of Earth and see our pathetic and ignorant species, there's a sense of "if I don't do it, who will? Nearly everyone else is simply too stupid that this is my comparative advantage. Even the average Manifolder realizes that if they try, they'll inevitably and kill everyone. Only the likes of @Mira are competent enough to do it properly."

The only reason I even consider the other options is GPU price inflation means I need more money than I have savings to compete with the OpenAIs of the world, and it'd be nice if it was something AI-adjacent so my skills don't atrophy while I work on it for a year.

So I feel like this option should be higher, because there's a good chance I say "money's not a problem yet, I can afford to spend another year doing something interesting before I have to get more".

Here's a poll market where you can vote YES/NO for your approval on each option separately from predicting anything.

/Mira/poll-which-of-mira-s-cool-ideas-do

@Mira And a bounty if you want to get paid to suggest ideas. You don't have to write a full description, I can elaborate it for you.

/Mira/get-paid-to-think-of-cool-ideas

All: I will not be trading on any of these markets, so it is safe to leave limit orders without worrying about insider trading. I also won't be giving select information to other Manifolders.

You might still lose money on a limit order if I post a comment with an update and aren't fast enough, but not from "insider trading".

what if some ideas are not cool in your judgement later

@firstuserhere Some of these are already starting to look less cool even just a day later. But they only have to meet the trigger condition to resolve.

For example, I see people buying down AGI. Reminder that this is not "Mira invents AGI", it's "Mira starts a project as described in that linked market, which is tangentially related to AGI".

I could spend the whole year playing with evolutionary algorithms and getting absolutely nothing done. Then all the options on the AGI market resolve NO, but AGI here would resolve YES.

@Mira some of these did strike me as not so cool and a bit mundane (especially knowing some of the work you've done prior)

Since you guys are bidding this one up, I should tell you I have not had a job in 2 years. (I think I left my last one October 2021)

@Mira how old are you? late 20s or early 30s

@Soli 34

@jacksonpolack Not a fan of Mirafold I see. Straight down to 9% below the "company that already exists" lol

just seems high effort low reward and people wouldn't use it

Is there a market for which of your terrible ideas you will actually do?

@ClubmasterTransparent This includes terrible ideas too, they're only cool because I own them. I added my own prediction market as an example of a terrible idea, so everyone knows terrible ideas may show up too.