patrick is in london

btw, twitter pals, I am spending the summer attempting to cross the USA by bike! true to form I have a prediction market on it here: manifold.markets/draaglom/what-…

@uPrice11 let's go

Offer accepted wooooo

@NathanpmYoung @metaculus great work! I'm ashamed to admit I seriously doubted you at first

The tournament is finished. I beat the @metaculus community. And came 2nd overall! I spent 80 hours over the last 4 months forecasting this.

@bananamelter @eigenrobot I wouldn't duck into a doorway personally, but it's generally not a good idea to have your phone out and do something that compromises your awareness for a while, e.g. I would not walk along the street video calling someone

@bananamelter @eigenrobot it's a real issue, although the police are tackling it a bit more now than the peak which (subjectively) was 2-3 years ago

paul graham loving "travel" makes more sense now

@BenShindel @NunoSempere (it's me I'm guy)

@draaglom @NunoSempere So true

We will likely never have definitive evidence one way or the other around the origins of covid. 26% doesn't seem like a terrible probability for it coming from a lab as things stand.

@BenShindel @NunoSempere thanks! To your last point: I don't think it's particularly crazy or a "trick"; the modal forecasting/rat subculture guy is a software engineer who likes to debate things after all

Just look at the Bayes factors here. We saw the first COVID cases in *many* cities in China in wet markets, yet Peter assigns this a ratio of 0.00025, lol. The "highest case density near wildlife stalls" is just completely bs spotlight effect, lol, just a complete artifact of where they measured in the wet market. Peter gives this a 33x Bayes factor (0.03). Just incredible grift all across this idek where to start. The "DEFUSE not happening" sub-debate was a total missed opportunity from Rootclaim. None of the ppl involved in this debate appear to understand how grant applications work in STEM bc they've never worked in a lab. Generally, about 50% of the work on a grant application is already completed prior to even submitting it, lol. And a lot of work happens anyway even without funding, on the money/resources from other grants! Everyone knows this. The entire debate seemed predicated on a very specific version of lab leak that involved heavy manipulation of the virus, while I think by far the most likely scenario that could result in lab leak just requires culturing or improper storing of viral samples. Tons of random mutations can happen without sophisticated methodologies that both sides in the debate seemed to assume was a precondition. Idk man, I'm sure Peter is a nice guy but it's kind of crazy how he tricked our entire subculture into viewing him as some expert on this topic because he's good at debating. He's a software engineer.

@BenShindel @NunoSempere (I also don't want to dive into an object level debate about the top level question btw, but I do want to know what informs this aspect of your view re: Peter)

@BenShindel @NunoSempere what's a specific example of something Peter is confidently wrong about, in your view?

@BenShindel @NunoSempere sure, we're agreed that the market format isn't good.

@draaglom @NunoSempere Hard to say, but generally questions with <5% chance of resolving in the next decade strike me as a terrible use of Prediction Markets and a better use of crowdsourced forecasting or the kinds of methodologies that intelligence agencies use.

@BenShindel @NunoSempere oops that's the wrong user ss, I mean:

@BenShindel @NunoSempere the specific claim re: 60-70 seems unlikely. here's peter's trades over time - note that he directly accounts for very little of the price action in the last year / since 50%. you could argue that the last year is still his aura pushing away yes bettors but this seems p tenuous

@izaakearl @aliceisplaying (hand waving this a bit because obviously you can have one household victimized N times, but it still doesn't really make the maths work out well)

@izaakearl @aliceisplaying This can be true and be a problem and yet not add up to a 4.5x relative risk, given that both cities are starting from a high baseline of petty crime. US surveyed petty crime victimisation is ~10%/household/yr, so 4.5x relative risk is 45%/yr, which is obviously not true

someone in the replies said that surely london has lower crime rates than new york to which i say lol. lmao even

@patio11 @moseskagan In Britain, non citizen permanent residents can vote in local (but not national) elections. It works fine, AFAICT.

@moseskagan This was floated in another nation and I slapped myself so hard to almost get a concussion, while of course offering no opinion on the merits of the matter, as the citizens of a nation dear to me are per se entitled to argue for which laws their government should pass.

First article in today's LA Times

@NunoSempere @Daaanvdb Oh, I misunderstood you, sorry. I believe Nino 3.4 is the "main" one (in the sense of being referred to in mainstream weather reporting for El Nino)

@draaglom @Daaanvdb I wasn't confused about the different lines! I was confused about the different charts, which are from different initial conditions and for different locations

And the latest El Niño ensemble is even warmer. Even the coolest models now peak well over 2°C. Ensemble mean reaching above 3°C.🤯

@NunoSempere @Daaanvdb AFAIK, from trying to interpret these before, each member of the ensemble is typically the same program with slightly varied input parameters

@Daaanvdb What are we looking at here? I see that there are many such plots here, but what's the difference? cpc.ncep.noaa.gov/products/CFSv2…

@vraiqx @goblinodds i picked blue. if 49% of people get permabanned that will be very funny

@goblinodds lmfao. i picked blue so i can trick shot my way into survival. i'm a real daredevil, a toxoplasmosis infected little freak if you will

@realTomBayes @predictionbench variance

On @predictionbench, Claude Opus 4.6 grew its starting balance by 7x while (paper) trading on Polymarket. While doing the same thing on Kalshi, the same model lost $2,000. Very interesting mechanics. Why do you think this is?