The Wuhan Coronavirus is looking like a pandemic event that could be serious enough to threaten this outcome.

If @datscilly doesn't get bored and lazy, probably some time in the next year. There are a few factors here. The question volume is up and I haven't been keeping up. Many questions now are range questions, which are harder to just 'wing it' for a good prediction.

Also my usual access pattern is on my phone when I have a spare moment, and the UI is broken to near unusability for range questions from the android browser. If @max.wainwright would fix the UI for android use it might give me some extra months.

@(Tamay) I outlined my thinking as of January 24 a little more thoroughly in a thread at https://www.metaculus.com/questions/124/will-the-world-population-increase-every-year-for-the-next-decade/#comment-20441. I had become concerned about the novel Coronavirus a few days earlier and read everything I could get my hands on, particularly the epidemiology and virology scientific preprints that were starting to come out. On January 24 I wrote: "The Wuhan Coronavirus spread is still in early stages, so it is hard to get reliable estimates. One preprint fr...

I think the community prediction is way too low on this. We've seen now that the virus can be transmitted by an infected patient who shows no or only very mild symptoms. [1] This means that a SARS containment strategy of quarantine the sick and trace contacts cannot work. Also in the absence of strong intervention the number of infections doubles every roughly 6.4 days. [2] Also based on travel patterns out of Wuhan there are statistically very likely undetected cases in countries like Indonesia, Thailand and Saudi Arabia. Such clusters will in ...

I spent 50 tachyons and tested my influence on the metaculus prediction. (I don't much like tachyons anyway.) If I predict at 1% the metaculus prediction is 27%; if I predict at 99% the metaculus prediction is 60%. So my influence can swing the odds ratio by a factor of 4. And I guess I did just swing it by that factor, albeit temporarily, so depending on how exactly the question is interpretted maybe now it just depends on the community prediction. (Does the period of tripling ratio have to be the same for the two predictions? Should there ne som...

@admins This is a request for improvement of the "current points" and "total available points" for questions like this. The current displayed points are (if I understand right) for if the resolution date stayed at the end of the period even though the event date is graphed on the x-axis. But the resolution date will be set to the date when the event occurs, so the points on the graph are not correct for what will happen. If the points calculation used the resolution date as the x-axis date then it would be correct (if I understand right).

@(notany) I hope you are right. The back of the envelope calculation is that world population growth rate is about 1.2%, so a death rate of 1.2% would cause a decline. The Spanish influenza infected about 1/3 of the world population and killed about 10% of those infected (some wide variance on those estimates however.) Spanish flu had a transmission rate r0 of about 2.0: on average each person spread it to 2 others. The Wuhan Coronavirus spread is still in early stages, so it is hard to get reliable estimates. One preprint from today is giving an e...

Okay I am back off a bit, to 98% here. The 2019 nCov epidemic still threatens to become a global pandemic, but evidence is growing that the fatality rate is comparatively low.

A coronavirus case was confirmed at my workplace today.

@Converse it should be true, just no one knows how to prove it. The "KAM torii" that make up the understood bounded solutions are a Cantor-like set. Read <https://en.m.wikipedia.org/wiki/Smith–Volterra–Cantor_set> about how a Cantor set can be constructed with positive measure, and what that means. I picture it sort of like baklava with infinitely thin pastry layers seperated by goo. The pastry layers are the KAM torii; the goo is everything else. The total thickness of pastry is positive, as is the total thickness of goo. There are individual la...

My contribution to the graph: exaggerated weekend bumps due to lower admission rate on Saturday and Sunday.

As the local techno-pessamist (or at least thats how I feel; really I aim for *realist*), I think this is at least two orders of magnitude below 1% likely. The interesting bit which has a chance of positive resolution in that timeframe is whether it will be conclusively shown to be physically possible to build a space elevator. Basically the only candidate material to have sufficient tensile strength-to-weight ratio is a ribbon of carbon nanotube fibers. To date I don't believe macroscopic samples have been generated which test at sufficient strengt...

I would prefer a slightly less specific resolution, something like "uses either the word 'communist' or 'commie' in reference to Bernie Sanders."

It should also include retweets and tweeted videos that contain the words "communist" or "commie".

The problem with asking for specific phrasing is what if you miss the one that is used. As weitten, for instance, Trump could adopt "Commie Bernie" as his nickname for Sanders and use it repeatedly, but as long as the last name was omitted it would resolve negative.

The European Commission's model has Nigeria's population at 379 million in 2050 for their middle scenario [1]. My feeling is their model, which takes into account the effects of education driving the demographic transition, is a bit better than the UN's simpler model.

[1] Lutz et al 2018 : https://ec.europa.eu/jrc/sites/jrcsh/files/lu…

Can we please fix the UI for continuous question prediction to work on android mobile? Currently it is almost impossible to get the sliders to slide. Very frustrating!

@(Linch) @(randallburns) I am at least partly responsible for the low and possibly overconfident metaculus prediction. My thinking is that the first disease peak will put us to the end of summer with 65k to 85k deaths and a low rate of active disease at that point. Given the proven ability to control the disease by scoial distancing and lockdowns I expect a governmental response that controls the second peak to no more than half the size of the first. Implicit in this are some assumptions that I think are probable but uncertain. The total size of t...

There are a few competing important facts that make this hard: (1) China and South Korea have demonstrated it is possible to stop an advanced outbreak of Covid-19; Taiwan, Hong Kong and Singapore have demonstrated it is possible to head off an outbreak at the earliest stage. (2) Europe (including UK), Iran and the US have advanced outbreaks but are not attempting to stop them. The strategy seems to be ignore it as long as you can, then try to mitigate the worst by efforts to slow transmission and "flatten the curve". But the final outcome will be clo...

As this question examines potentially large numbers of cases, I would suggest to be clear that it should resolve based on scientific estimate of actual number of infections, rather than laboratory-confirmed counts. If the disease becomes so widespread only a small fraction of cases will be confirmed.

@(Pablo) @juancambeiro I hadn't noticed Juan's update. I think it is bad form materially changing the resolution criteria close to the resolution date, particularly since scoring depends on your prediction at earlier times with the original criteria. Also I think it is bad form to change the resolution criteria in a way that contradicts the original text : clarification is fine; change is not. Disclaimer: as resolved I received +57 points. My mid date was Jan 24; for what I think is a correct resolution date of early January by the original criteria I...

A new academic preprint puts the all-infection fatality rate at about 1%, though the 95-percentile range is still 0.5% - 4%.

https://www.imperial.ac.uk/media/imperial-col…

— edited by traviswfisher