Again, community does not seem to be consistent.
[First AGI (weak)?](https://www.metaculus.com/questions/3479/when-will-the-first-artificial-general-intelligence-system-be-devised-tested-and-publicly-known-of/) Apr 2033 (2026 - 2051)
[First AGI (strong)?](https://www.metaculus.com/questions/5121/when-will-the-first-artificial-general-intelligence-system-be-devised-tested-and-publicly-known-of-stronger-operationalization/) Aug 2044 (2032 - 2065)
[How long from weak AGI to superintelligence?](https://www.metaculus.com/questions/4123/after-an-agi-is-crea...
Everyone here should remember that on the world scale we are WEIRD - Western, Educated, Industrialized, Rich, and Democratic. The rest of the world is not like us and our condition does not define the humanity condition. Bring the whole world up to the standard of living of an average American, Dutchman or Finn would IMO be a fundamental change in human condition already.
We are living trough multiple revolution on a scale far greater than the agricultural and industrial revolutions simply because of how many people they impact. First revolution is the ...
For me the only problem with UX with the current design on mobile are the sliders. Especially on open questions. It's super cumbersome and fiddly to make precise predictions. But from what I can see it didn't change at all.
I also want to mention that like the current design, the night version, for it's unique, edgy and old school character. The new design is okeish, but it is bland.
The main positive I see in the new design is being able to see the current community prediction distribution on the list of the question. I also like that there is a label...
Supercentenarians and the oldest-old are concentrated into regions with no birth certificates and short lifespans "As such, relative poverty and short lifespan constitute unexpected predictors of centenarian and supercentenarian status, and support a primary role of fraud and error in generating remarkable human age records." https://www.biorxiv.org/content/10.1101/704080v1
I was looking at Metaculus community predictions looking for inconsistencies. I'm just paying with the idea that cross referencing predictions to create consistent narrative may improve predictions overall.
I don't think I can explain the following predictions:
- 55% for [Human-machine intelligence parity by 2040?](https://www.metaculus.com/questions/384/human-machine-intelligence-parity-by-2040/) resolved by testing AI and graduate students in each of physics, mathematics and computer science on an identical series of exam questions over a period of ...
I'm considering whether Metaculus could be leveraged as a tool to do serious science. Sort of predicting the future by actually making it.
My idea is to ask a well formulated research question on Metaculus that can be answered anytime just by carrying out an experiment. The question closes and resolves when the proof of carrying out the experiment is posted. The person who carried out the experiment gets to predict with full certainty, since she or he actually knows the result. The moderation team and community serve also as a peer-review system.
I have compiled list of some interesting resources:
[WHO situation reports.](https://www.who.int/emergencies/diseases/novel-coronavirus-2019/situation-reports) I guess it's the most reliable and official data source there is.
[BNO News: map and timeline](https://bnonews.com/index.php/2020/02/the-latest-coronavirus-cases/) It's very frequently updated and the time line is especially useful. They inform about first cases in a country and provide links to sources.
[GitHub repository with the data by Johns Hopkins CSSE.](https://github.com/CSSEGISandData/...
I tried this problems with OpenAI Davinci - Codex and it solves basically all of them with maybe minor issues here and there.
I have concerns that it may be memorizing solutions to the specific set of problems from the question. But it is solving pretty much all of them. It manages to solve quite a lot of them even if I modify them.
For example I tried this modified elementary problem 9:
Write a guessing game where the user iteratively has to guess a secret random letter. Indicate whether the secret letter is before or after the guessed letter in t...
@(moreheadm) I agree with @notany sentiment. In general recent tournament questions seem "made up" or forced e.g. road-to-recovery. Answering cookie cutter questions is not intellectually interesting.
This tournament is more interesting than road-to-recovery, but it could still use some more variety. Li Wenliang Forecasting Tournament had a lot of interesting questions and I think it was in big part due to crowdsourcing.
My suggestion to @Tamay and admins would be to give prizes for question writing as a part of tournaments. Good question writing is re...
Out of these 3 questions at least one has to be wrong by a lot:
- [What will be the doubling time of COVID-19 cases during the peak growth period in 2020?](https://www.metaculus.com/questions/3802/what-will-be-the-doubling-time-of-covid-19-cases-during-the-peak-growth-period-in-2020/) ~ 3.9-6.3 days
- [Which month of 2020 will see the biggest global increase of COVID-19 cases?](https://www.metaculus.com/questions/3781/which-month-of-2020-will-see-the-biggest-global-increase-of-covid-19-cases/) April-May-June
- [How many infections of COVID-19 confirmed ...
Seems like this resolves positive: [WebGPT](https://openai.com/blog/improving-factual-accuracy/)
> Previous work on question-answering such as REALM [Guu et al., 2020] and RAG [Lewis et al.,
2020a] has focused on improving document retrieval for a given query. Instead, we use a familiar
existing method for this: a modern search engine (Bing). This has two main advantages. First,
modern search engines are already very powerful, and index a large number of up-to-date documents.
Second, it allows us to focus on the higher-level task of using a search engin...
Indonesia seems to be underreporting cases.
From @ninjaedit [here](https://www.metaculus.com/questions/3711/how-many-total-confirmed-cases-of-novel-coronavirus-will-be-reported-in-the-who-south-east-asia-region-by-march-27/#comment-23634).
> People with travel history from Indonesia seem to be showing up as infected in other countries (https://www.thestar.com.my/news/regional/2020…). At least in Singapore, and possibly there was also a case in Australia. If all three examples are correctly reported, the undetected outbreak in Indonesia could well be abo...
@nagolinc It also appears that infections roughly follow Pareto principle - around 80% of infections are caused by around 20% of events. So, if Metaculus users are very good at avoiding this types of situations, the difference may be amplified in a nonintuitive way.
[Situation in South Korea is very encouraging](https://i.imgur.com/kXofqt4.png) with 0.6% case fatality rate.
WHO thinks that this virus can be controlled and contained with efforts similar to China or South Korea. Fundamentals of epidemiology work, tracking people and isolating works.
IMO testing, tracking down contacts and response of the general population will be the biggest factors determining the scale of the outbreak and whether it will grow out of any hope of control. This is true for rich countries, the question is also how countries poorer th...