Will George3d6's 'Increasing IQ is trivial' replicate?

See this post for context: https://www.lesswrong.com/posts/h6kChrecznGD4ikqv/increasing-iq-is-trivial


Resolves YES if there are independent reports from people who were in contact with George3d6 (I'm thinking of johnswentworth and jacobjacob from the comments from the post) and used his methodology that they found a significant increase in IQ. This will resolve YES if at least THREE people report a significant (not specifying exactly what significant means, but if is unclear, I will wait until further replication) increase using this method.

Resolves NO if George3d6's experiments indicate this is not replicable AND at LEAST three people report they have not found a significant increase using this method.

I'm not exactly sure if these resolution criteria are good, so if you have any suggestions, please put them in the comments and if I think there are better, I will update the resolution criteria. I will try to resolve this in the spirit of the answer rather than EXACTLY adhering to the resolution criteria if things seem unclear.

Edit: updated resolves no to be more strict. If it is unclear, I will just keep the question open until parties agree. If it seems that no one is going to agree (and people arrive at a sort of stalemate), I will resolve N/A


This market's resolution will be decided by site moderators. This overrides any resolution criteria above.

Get Ṁ600 play money
Sort by:

big limit order! take it!

I'm normally not on the hopium snorting side around here, but in this case there seems to be a pretty straightforward hypothesis for YES (full disclosure: I have NOT done any research on this). Namely, people in daily life are simply not at their best, so why would it be surprising that getting some nice massages and good sleep or whatever would increase your IQ by 5 points? AFAIU, the buff can't be stacked and runs out as soon as you stop actively maintaining it.

Compare to the well-documented negative effects of even mild sleep deprivation.

I think it would be surprising if some nice massages increased your IQ (in the sense of, how generally capable you are at intellectual tasks) by 5 points, tbh. I would expect 'getting better sleep' to help a bit, but I don't think enough people are perpetually sleep deprived to lead to a 5 point increase in general.

…targeted NIR interference therapy, short UV during the morning, a lot of inversion-based exercises where I focused on contracting/relaxing neck and face muscles, a few customized breathing exercises (think wim hof), figuring out the correct levels for a bunch of cholinergic vaso[dilators/modulators] (think noopept), massage therapies to reduce tension on the spine, some proprioception-heavy movement practices, a niche tibetan metta meditation series… and about 5 other things that are even harder to compress.

Doesn't seem right to me.

That's a large percentage, for something that would be so important and neglected.

@DavidSartor markets are unfortunately inefficient somewhat often.

I'm actually curious which mods will resolve the market here and how that will be decided. I have found more people who want to run versions of my self-experiment and I could have them report their findings & chat with someone over DMs or email.

Otherwise, all I have is a spreadsheet with that, but for all intents and purposes I could be faking that to gain fake internet points, people have done worse for such (:

I was hoping the Lighthaven team would try and replicate this, but it looks unlikely due to the time commitment involved -- so best I can do is provide people that have tried your way, and they could vouch for effects & for the controls, since people are bringing their own controls too

@GeorgeHosu3131 Good point. I would maybe ask on the manifold discord server, the mods seem to be active there.

@GeorgeHosu3131 I've not followed this at all but my understanding of the second edit is that a "Cheaty-Win" would not count because the original criteria were overridden.


This market's resolution will be decided by site moderators. This overrides any resolution criteria above.

Maybe those original criteria should all be crossed out in the description to make this more clear?

@GeorgeHosu3131 Speaking not as a moderator but just a dude, my thought is that any chance of your experiment actually working consistently is a much huger deal than this manifold market and you should just pursue that without any thought to play money. You could make a lot of real money with this!

@GeorgeHosu3131 im in christchurch too if u need someone to experiment on

@Joshua done!

@Joshua I agree that it's a bigger deal and at this point I am pursuing that but "how to convince the external world that this works" vs "what is the proof manifold need to know that it works" are related at like r=0.2, so it might be worth figuring that out.

@jim I'm no longer in Christchurch, I'm in SF at present, but if this sort of stuff interest you then you should hit me up the next time I'm there, or hit up the one employee I have that's in chch ^^ (Does manifold have DMs ? If so DM me your signal and I'll create a group) -- I feel like there's a 50/50 chance we know each other vaguely already if you're on manifold & live in chch, limited crowd : p

bought Ṁ40 NO

This will resolve YES if at least THREE people report a significant (not specifying exactly what significant means, but if is unclear, I will wait until further replication) increase using this method.

It matters how many people try it, right? Like if 20 people try it and 3 people report success, I would not consider that a successful replication. Or if 20 people try it, and 17 of them never comment because it didn't work for them, so it looks like a 3/3 success rate.

@MichaelDickens Read the update. The mods will now resolve. These criteria do not matter at all (the mods will just review all the data and make a decision irrespective of the criteria).

@MichaelDickens This is precisely the issue I see too, see my comment above. I have no good solution here myself and I'm not quite sure what "mods will resolve" means.

I'd love for this market to be resolved as honestly and independently of myself as possible.

opened a Ṁ20,000 NO at 10% order

50k no order @ 10%

@jacksonpolack These sorts of experiments are mainly bottlenecked on large numbers of volunteers, which is specifically what George was asking for in the post.

I think that bets like these risk FUD-ing these kinds of projects enough to be potentially considerable self-fulfilling prophecies enacted by elite forecasters. But elite forecasters also normally need to bet freely.

A solution might have been to have a separate market that predicts whether enough high-quality volunteers will sign up (I.e. predicting if the experiment would run out of steam before even starting) and see if that market crashes from no stimuli other than the elite forecaster making the bet.

There's still confounding variables but it's better than nothing (too late here).

opened a Ṁ1,000 YES at 6% order

@jacksonpolack added 1k yes at 8%.

My latest update on this: https://open.substack.com/pub/morelucid/p/increasing-iq-by-10-points-is-possible?r=c5fr0&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Intervention mean increases: (11.2 [9%], 9.6 [8%], 12.6 [10%]) (mean of means: 11.1) - Average increase: 9.3% Control mean increase: (14.2 [12%], 4.4 [3%], 8.8 [7%], 7.6 [6%], 5.2 [4%], 5.6 [5%], 3.2 [2%]) (mean of means: 7.0) - Average increase: 5.9% Controlled mean increase: 4.1 Related T-test between the before/after means for the intervention: -12.846 (p=0.006) Related T-test between the before/after means for the control: -5.015 (p=0.002) Independent T-test between the before/after difference between intervention and control: -2.46 (p=0.04)

If this data looks sufficient to solve the market I'm ok sharing it + my analysis code with whomever wants to resolve it.

Otherwise there will be more people self-experimenting with my approach soon and more data should come in.

The change is indeed smaller than the one I got initially and the control looks better (though see disclaimers about control advantages)

can you please post the raw data of everything you collected, including everything that you don't think is relevant

@jacksonpolack I kinda wanted to keep the data private until all my controls came through but for the sake of you being the biggest "No" holder, here: https://docs.google.com/spreadsheets/d/13VCkiqrcidIEbe7xDDI0Sl_3qoHT6Z23shdS56pvS5U/edit?usp=sharing (note as per spreadsheet I've paid more people to participate as controls since one of my original controls bailed + the controls had very high variance and I wanted more data points)

I don't think this is enough information to "resolve" the market, but I think it's a good indicator towards "yes" (see my full explanation above, I basically used worst-case-scenario people for both control and intervention)

If some credible 3rd party is interested in doing this (I'm hoping the LH team) or if you can suggest a 3rd party that's interested I'd say let's resolve with them.

The self-experimenters from which the data above derives are all my friends in the "I will try this crazy self-experiment for 4 hrs a day for 2 weeks" sense so if you want to argue bias you certainly can. C0, 1,3, and 9 are also friends of mine.

Thanks! Even if it turns out you made a mistake (which is likely, the first time I tried doing a self-experiment I made a series of very funny mistakes), posting your data is how you learn how to do it properly

bought Ṁ50 YES

Did George ever share the specifics oft bis Intervention? I'd try too but i'll propably won't be able to get someone to be my control.

@Schwabilismus Nope he hasn't given specifics.