I have a story to tell that I have to get off my chest.

Two nights ago I was falling asleep and I heard faint screaming. I wasn't sure where it was coming from, but I decided to look out my window. Across the street is the parking lot of a different apartment complex. There was a car with a woman on top of the windshield. She was screaming and periodically would punch and kick the windshield of the car.

Then I saw the car accelerate and decelerate around the parking lot in an attempt to get her off, like a bull trying to get a bull rider off. I was appalled. I immediately called the police and gave a quick yell out the window saying "Hey, stop!" or something. She very well could have fell off, been run over, and died. At the end of it my girlfriend also was screaming out the window to tell her to get off and stop. She was able to get the girl to start talking, then the man inside the car started talking. Turns out she cheated on him, the man wanted to break up and get away, and she didn't want to let their relationship end. When my girlfriend said we were calling the police the woman finally got off the car, the man opened the door, let her in, and they drove away.

The next day I called the local police department because after having hung up with 911 that night, I learned that the man in the car's name was Alex. I figured that providing that information to the police might be useful. Hopefully there'd only be one male resident named Alex in the apartment complex who is living with a girl and owns a hatchback car.

The lady I talked to sounded annoyed that I called and thought that this was nothing. When I described what happened she immediately blamed the woman, saying that she shouldn't have been on top of the car kicking and banging in the first place. Of course. But two wrongs don't make a right. That doesn't give the man the right to w recklessly endanger her life like that.

Then she got to asking me what my new information was. I said how I learned that the guy's name is Alex and that maybe they could track him down. She said there's nothing to track down and to call back if I see anything violent happen. But she also asked me what my thinking is. So I explained that a) what the man did seems illegal and probably worth pursuing criminally, and b) he might be a danger. The lady looked at me like I had two heads. She asked me: did you see a gun? A knife? Punches? Shoving? Anything violent? No, but the guy could have ran her over or seriously injured her. Seems like a dangerous person. Domestic violence seems very plausible. Seems worth following up on.

Here's where we get into rationality a little bit. After hearing me out, she explained that all of this is just my opinion. I don't have any "facts", like seeing a gun or a punch being thrown. To her, "facts" like those would be a sign that the man was dangerous and maybe worth following up on, but here it's merely my opinion that a guy who tries to whip someone off the roof of his car is dangerous.

I am seeing three (related) errors in her logic. The first is having an arbitrary cutoff for what "counts" as evidence. As explained in Scientific Evidence, Legal Evidence, Rational Evidence, rational evidence is anything that "moves the needle" but in various social institutions they may have arbitrary cutoffs for what "counts" for whatever reasons, like how science often has the cutoff of p < 0.05. It's possible that there is a good reason why she is drawing the cutoff here. It seems highly unlikely though. But more to the point, she seemed to be saying that because it is merely my opinion, it doesn't even count as rational evidence. That "I have nothing". I would have thought that someone working in law enforcement would be better at assessing danger than this.

But this brings me to the second error: belief in belief. I don't think that she actually would predict this man in the car to be an average threat. I don't think she would be happy to find herself crossing paths with him in a back alley. Yet she professed as if all of this is merely my opinion which I'm not basing on any "hard facts" and thus there's no reason to suspect anything of this man.

Third error: fallacy of grey. Even if you want to draw an arbitrary line between opinion and fact, all opinions aren't created equally. Some shades of grey are darker than others. Some opinions are stronger indicators than others.

It's sad because there really does seem to be a practical path forward here. Maybe there were cameras that captured it. Maybe there were other witnesses. I'd imagine that there would be. They were screaming pretty loudly, the car was screeching, and there were a lot of other apartments in the vicinity. Most of which were closer than mine. Mine was across the street and my window was closed. And even if you don't want to prosecute the man for some sort of reckless endangerment it seems worth following up with the woman to see if there is any domestic abuse going on and if she feels safe.

Maybe this just doesn't relate to me and so I should stop talking about it? Maybe. Or maybe the police officer's rationality is my business, given that I am a member of the community that she is policing and trying to keep safe. It's certainly the business of the girl who could have been ran over.

Another sad thought: there is a security patrol that is usually stationed maybe 200 feet away from where this all happened. I couldn't see if they were out there from my window, but last night I was out for a bike ride, saw a security car stationed in that usual spot. I told him what happened the previous night and asked if he thinks there might have been someone stationed there. He said there probably was because they're supposed to be out all night, but that they wouldn't have been able to do anything. They are contracted by some factory nearby and if he said if he left his post for something unrelated to the factory like that that he'd be fired on the spot. These experiences have all shifted my beliefs about just how inflexible and inhumane bureaucracies can be.

I am experimenting with holding office hours where people can talk to me about really anything they want to talk to me about. First iteration, next week Wednesday at noon (PT) at this link: 

http://garden.lesswrong.com?code=bSu7&event=habryka-s-office-hours

Some more event description: 

Come by if you want to talk to me! I am a man of many hats, so many topics are up for discussion. Some topics that seem natural:

+ LessWrong
+ AI Alignment Forum
+ LTFF Grantmaking
+ Survival and Flourishing Fund grantmaking

Some non-institutional topics that seem interesting to discuss:

+ Is any of the stuff around Moral Uncertainty real? I think it's probably all fake, but if you disagree, let's debate!

+ Should people move somewhere else than the Bay? I think they probably shouldn't, but am pretty open to changing my mind, and good arguments are appreciated.

+ Is there any way we can make it more likely that we get access to the vaccine soon and can get back to life? If you have any plans, let me know.

+ What digital infrastructure other than the EA-Forum, LessWrong, and the AI Alignmnent Forum do you want the LessWrong team to build? Should we revive reciprocity.io?

+ Do you think that we aren't at the hinge of history because of Will's arguments? Debate me, because I disagree, and I would really like someone to defend this position, because I feel confused about what's happening discourse wise.

Almost anything else is also fair game. Feel free to come by and tell me about any books you recently read, or respond to any of the many things I've written in comments and posts over the years.

I'm trying out Bayes Theorem with a simple example and getting really strange results.

p(disease A given that a patient has disease B) = p(b|a)p(a) / p(b)

p(disease B given existing diagnosis of disease A) = 0.21

p(A) = 0.07

p(B) = 0.01

I get 1.47 or 147%. I know that the answer can't be >=100% because there are patients with A and not B.

Where am I going wrong?

Hey LessWrong, new member here. I'm a Psych major from Australia who's still figuring out things out. I discovered LW through Effective Altruism through 80 000 hours through.. I've forgotten but most likely Internet stumbling. 

I'd been feeling the itch to write as a way to express and iron out my thoughts, but it's taken me somewhere near a decade to get around to writing anywhere publicly online. So here we are. Looking forward to engaging, discussing and chiming in more with the LW community.

I've just looked at scoring functions for predictions. There's the Brier score, which measures the squared distance from probabilities ($p_1,...,p_n$ with $p_i\in [0,1]$) to outcomes ($x_1,...,x_n$ with $x_i\in \{0,1\}$), i.e.,

$$-\sum _{i=1}^n(p_i-x_i{)}^2$$

(Maybe scaled by $\frac{1}{n}$.) Then there's logarithmic scoring, which sums up the logarithms of the probabilities that came to pass, i.e.,

$$\sum _{i=1}^n{x}_i\ln \left(p_i\right)+(1-x_i)\ln (1-p_i)$$

Both of these have the property that, if $k$ out of $n$ predictions come true, then the probability that maximizes your score (provided you have to choose the same probability for all predictions) is $\frac{k}{n}$. That's good. However, logarithmic scoring also has the property that, for a prediction that came true, as your probability for that prediction approaches 0, your score approaches $-\infty$. This feels like a property that any system should have; for an event that came true, predicting (1:100) odds is much less bad than (1:1000) odds, which is much less bad than (1:10000) odds and so forth. The penalty shouldn't be bounded.

Brier score has bounded penalties. For $x_i=1$, the predictions $p_i=\frac{1}{1000}$ and $p_i=\frac{1}{{10}^{1000}}$ receive almost identical scores. This seems deeply philosophically wrong. Why is anyone using Brier scoring? Do people disagree with the intuition that penalties should be unbounded?

Age 4-7ish learning resources recommendations! Needless to say, different kids will take to different things. But here's my experience fwiw.

MATH - DragonBox - A+, whole series of games running from basic numeracy through geometry and algebra. Excellent gameplay, well-crafted, kid loves them.

Number Blocks (Netflix) - A+, basic numeracy, addition, concept of multiplication. Kid must have watched each episode 10 times and enthused about it endlessly.

Counting Kingdom - A+, Mastering mental addition. Excellent gameplay; fun for adults too. Note: Not currently available on ipad; I got it on PC Steam.

Slice Fractions 1 & 2 - A+, Teaches fractions. Great gameplay, great pedagogy.

An old-fashioned pocket calculator - A+, an underrated toy.

LITERACY: Explode The Code book - A, been around since at least the 1980s, still good.

Teach Your Monster To Read - B+, gameplay is a bit repetitive & difficulty progresses too quickly, but got a few hours of great learning in there before he lost interest

Poio - A-, Good gameplay, kid really liked it. Limited scope but great for what it is.

For reading, no individual thing seemed to make a huge difference and none of them kept his interest too long. But it all added up, bit by bit, and now he's over the hump, reading unprompted. Yay!

PROGRAMMING - Scratch Jr - A+ , duh

So... it looks like the second AI-Box experiment was technically a loss. 

Not sure what to make of it, since it certainly imparts the intended lesson anyway. Was it a little misleading that this detail wasn't mentioned? Possibly. Although the bet was likely conceded, a little disclaimer of "overtime" would have been nice when Eliezer discussed it.

GPT-1: *sentiment neuron*

Skeptics : Cute

GPT-2: *writes poems* 

Skeptics: Meh 

 

GPT-3: *writes code for a simple but functioning app* 

Skeptics: Gimmick. 

 

GPT-4: *proves simple but novel math theorems* 

Skeptics: Interesting but not useful. 

 

GPT-5: *creates GPT-6* 

Skeptics: Wait! What? 

 

GPT-6: *FOOM* 

Skeptics: *dead*

-- XiXiDu (I added a reaction to GPT1).

I found 2 bugs in the Less Wrong website. Where do they go? (this is the first bug; I couldn't find a place to report problems after looking through the FAQ and home page)

Does anybody have recommended resources that explain the timeline of clinical trials of interventions? Specifically why they take so long and whether that is because of practical necessity or regulatory burden. Bonus points if Covid-19 is included as a context. 

So I've read Pearl's The Book of Why and although it is really well written I don't understand some things.

Say we have two variables, and variable X could 'listen' to variable Y, Y--->X. But we don't know if it is qualitative or quantitative. I would have appreciated it if the book included a case study or two on how people plot their studies around this thing.

For example, we want to know what features of an experimental system can influence the readout of our measuring equipment. Say, Y (feature) is the variety of fungi species inhabiting the root system of a plant, and X is the % of cases in which we register specific mycorrhyzal structures on slides we view through a microscope (readout). And our 'measuring equipment' is a staining/viewing procedure.

Conceivably, if there are several species of fungi present, the mycorrhizal one(s) might form fewer (or more numerous) specific structures. This would be what I mean by a quantitative effect. Also conceivably, only some species or combinations of them have this effect on X. This would be qualitative.

Measuring both Y and X is more or less impossible, since you either stain a root or try to obtain a mycorrhizal culture from it (which is expensive.)· Even if we do try out some number of combinations of fungal inoculum, who knows how it compares against the diversity in the wild.

So... does this mean that we should split Y into Y-->Y1-->X and Y-->Y2-->X... or what?

· we don't consider some stains which maybe allow both.

In the interest of making people's experiences less surprising/frustrating: I am aware of a bug that seems to happen on some posts, where you can't vote on comments that are deeply nested. (It happens on the latest Zvi post for me, for example). I am working on fixing it. In the meantime, you can still vote on them, but your votes will only display when you refresh (which of course isn't an acceptable state of affairs, but might make things slightly less frustrating).

I wanted to make two points as feedback for the LW designers, but I realized there's actually only one point: Personal Blog feature makes me very sad.

1. I wanted to find this Open Thread page, but it didn't show up on my front page as it usually does. 
2. I was wondering why Zvi hasn't posted about covid in a while. I went to his user page and turns out he has been posting, but because they are Personal Blog posts, I haven't seen them.