All of dumky's Comments + Replies

Answer by dumkyDec 31, 202210

I subscribe to the Jaynes/Laplace view of probabilities, namely that they exist in the mind and result from changes in information rather than changes in the world, let alone multi-worlds.

Imagine I tell you about an urn with black and red balls, without an additional detail. You can provide a probability of getting a black or red ball (50/50). 
As I provide more information ("there's 5 red balls but 50 black balls", "the red balls are sitting on top of the pile", ...) your probability assignment will change without the physical urn having changed at all. 
As your knowledge of the urn and the selection mechanism becomes more complete, your uncertainty decreases and your confidence/probability levels grow.

Again, the existence of in-the-mind probability doesn't imply the non existence of objective probability.
Answer by dumkyDec 14, 202270

One possible incentive for corporations to promote wokism is that it's a distraction from other forms of activism which might their hurt bottom lines more. 

Also as Bryan Caplan and other comments pointed out, regulations may also drive this shift: the regulations leave companies vulnerable to lawsuits with vague criteria and a defense strategy is to 1. not be the slowest target (this results in a race/escalation) and 2. prepare a vague signaling defense (look at all these employee training programs we do, surely that proves your vague and non-falsifiable accusations of discrimination must be wrong).

The bank transfer example seems silly to me. In the situation where the exchange results in a net gift to both A and B, it seems easy to get consent from both of them.
More generally, if you can produce a Right, then you can get consent for the actions you need to aggregate and they are no longer deontological wrongs.

Thanks for the write-up!
I find the nomenclature "target network" confusing, but I understand that comes from the literature. I like to think of it as "slightly out-of-date reward prediction net".
Also, it's still not clear what "Q" stands for in DQN. Do you know?

Looking forward to your future deep learning write-ups :-)

1Jay Bailey8mo
Thanks for the feedback - glad you liked the post! "Slightly out-of-date reward prediction net" is a good way of thinking about it. It's called the "target network" because it's used to produce targets for the loss function - the main neural network is trying to get as close to the target prediction as possible, and the target prediction is built by the target network. Something like the "stable network" or "prior network" might have been a better term though. "Q" in DQN is a reference to Q-learning. I'm not 100% sure, but I believe the Q term in Q-learning is supposed to be short for "quality" - a Q-function calculates the quality of a state-action combination.

There's a chapter on the evolution of sexes in Nick Lane's The Vital Question ("Sex and the origin of death"). It explores the benefits and costs of sex.
If I recall the big benefit of sex is gene shuffling (allows for recombinations, ie. faster experiments). Also it apparently evolved very early (didn't need much new machinery, benefits significant).

Pretty cool experience, but I have to say this rang some alarm bells for me: "My goal for covid lockdown was to train Brittany to cook.".

What did she want?
But it sounds like she was onboard, or at least ended up appreciating the training and assistance :-)

Brittany wanted tasty food and not to be sick all the time. She was enthusiastic about learning to cook every step of the way.

Whenever discussing the effects of increasing the money supply, we tend to highlight the first order effects: if there's twice as much money, then prices are expected to double after a while.
This effect is already hard for the public to notice compared to straight-up taxation, but the effects during the transition are even more insidious (so called Cantillon effect): some people receive the fresh money first while prices haven't adjusted, and some people receive the fresh money last (after prices have already almost doubled). The folks who are the furthest... (read more)

1Phil Scadden2y
Those reviews are very helpful, but looking at "Derek" comments in the Cochrane blog, it is not at all clear whether these are critiques of MMT or critiques of Kelton ( or more specifically what Kelton believes that governments can do given that MMT is an accurate representation). I am also very suspicious of how this would work within a democratic system. Many countries do not let governments set interest rates - that is the role of the independent central bank to control inflation. I would feel happier about governments printing money if the central bank was also dictating the level of taxation (the amount of money to be destroyed but not how that tax is distributed).

Sorry to hear this has been difficult for you. It sounds like you really want to find a way to make durable change and tried many techniques already.
I think it's admirable and impressive to see how passionate and driven you are. But if it's having some negative consequences or making you unhappy maybe a bit of tuning could be useful.
How do you feel when you have trouble to go to bed? Taking a wild guess I'd say you might be feeling a bit anxious, inadequate, frustrated or maybe embarrassed. What thoughts do you have? Is there a deep-seated value that drive... (read more)

Sometimes this is true, but over time 1) I realised that having these thoughts aren't helpful 2) I probably did it too many times that these feelings don't happen as much Worthiness is definitely something that I've been contemplating a lot recently. Here's my current brain dump: []

It can also be good to list the positives (advantages and beautiful things) in self-criticism. It shows honesty, accountability, high standards, ...

On the other hand, we often use a double standard (harder on oneself) which can imply a sense of superiority (it'd okay for others to make mistakes, just not for us). Another downside of self-criticism is that spots that we beat down become sensitive and so we may be less receptive to feedback (being defensive).

I apologize if it's a bit tangential, but I want to point out that "should" statements are a common source of distortions (in cognitive behavioral therapy). It is often good to clarify the meaning of "should" in that context to see if it's valid (is it a law of the universe? is it a human law? ...). It often just means "I would prefer if", which doesn't bear as much weight...

David Burns explains this clearly and I was struck when he pointed out the linguistic heritage of "should" and how it connects to "scold". Here's one podcast episode on the topic (ther... (read more)

Nice parallel to CBT! On a meta level, I really like comments which take the idea from a post and show a parallel idea from some other area.

Exposure is the most effective remedy for most forms of anxiety. It is also helpful to understand the thoughts behind that anxiety and consider to what extent they are valid or beneficial (there are good reasons to be anxious, but they may only justify it to a mild extent). 
Many techniques can help such examination. One that is particularly appealing to me is to imagine what I would tell a close friend who is very much like me ("double standard" technique).

Castify was a crowd-funded project that recorded "Rationality: From AI to Zombies" among other things. They closed in 2016, but I made a backup as they suggested. But I don't think the audiobook ended up being hosted on as they announced...

"To all Castify customers old and new,

It's been a fun project but it's time to bring Castify to a close.  This means all podcast feeds and audiobook files will cease to be available for download at the end of September.  

IMPORTANT: If you need to download a fresh copy of your purchases&nbs... (read more)

1Jeffrey Ladish3y
Was the quality not up to par? If the reading / recording was good, it's surprising that it's not up on

I've only had a few interactions, but was less than impressed. Some articles that were critical of authors I have much respect for did not give a fair reading in my opinion.

Specifically, does the prevalence of progeny in different groups affect the average time preference of said groups? I would say it raises an empirical question, but the prior seems reasonable. I do not consider such thought horribly offensive or bigoted as suggested, even if you consider that sexual orientation would affect prevalence of progeny.

The thread seemed more woke than rationalist in tone.

1snog toddgrass3y
Yeah I observed that as well. Why have a rationalist community if you accept motivated reasoning or have an ideological standard. Like being woke is still an ideology. You did not become woke by evaluating every assumption your ideology supports and finding all of them true. So why use that to write answers at the bottom of the page. Whatever, Dagon is right. This does update me toward thinking rationality has many many failure modes which are hard to avoid. Tsuyoku Naratai is important. Having external goals to our rationalism is important.

Another possibility (not sure how likely): maybe treatment capacity and protocols have improved?

Capacity is definitely higher, since this is much less concentrated than the first round. Even if protocols have improved, deaths staying essentially flat since mid April while cases have grown ~5x is very strange.

James Burke mentioned such a graph being worked on in a talk he gave in 2009. Not sure how it progressed.

Note: if you’re interested in this topic, check out Burke’s book and TV show, Connections. It’s fascinating (a lot of clever repurposing and also happenstance).

Thanks! this seems to be the project: [] with this graph being the beta version: []

Forget money for a minute, since it’s just a proxy for real resources. Assuming that people involved have a decent understanding of what resources are helpful, then Charity is trying to provide those resources. For the question of whether we know what resources would be useful and whether they have the desired impact in the long-term, that’s a hard question. GiveWell is an attempt at evaluating the impact of charities. There are others as well.

That presented with a bet that essentially hinges on that thing being true, you take that bet.

Note that actions involve uncertainty and costs, so they involve a form of informal betting.

You haven't said anything about belief in that answer.

Sure, but that means there possibly is no answer (the problem is under-specified). Maybe the answer depends on preferences, not universal ethical or rational principles (such as deontological or utilitarian principles).

There is a third approach to the trolley problem, which I have not often seen discussed: whose property are the trolley tracks?

In other words, does this require a universal answer, or can this allow for diversity based on property and agreement? When you board a ship, you know that the captain has the last word when it comes to life-or-death situations, and different captains may have different judgement. The same goes for trolley tracks. The principles can be explained beforehand (such as "the mission comes first"). Another question that seems relevant to the person in charge (owner) is whether the 5 people are to blame for putting themselves at risk.

The trolley problem is a hypothetical situation designed to explore the clash between following deontological principles and allowing a great deal of harm to occur. So, in setting up this problem, we should be trying to limit these extraneous factors as much as possible. So we'll say that you are the person in charge with absolute authority over what call to make. We will say that none of this is your fault or anyone's fault - these people just mysteriously appeared on the tracks as a result of quantum fluctuations.

If I remember correctly, Jaynes discusses this in Probability Theory and arrives at the conclusion that if a reasoning robot assigned a probability to its changing its mind a certain way, then it should update its belief now. Of course, the general caveat here: humans are not robots, they don't perfectly adhere to either formal logical or plausible reasoning.

He does, it's in the chapter about Ap distribution, which are basically meta-probability, or better, Ap is the probability assigned to receive a future evidence that will put the probability of A at p. Formally P(A|Ap) = p. From this you can show that P(A) is the expected value of the Ap distribution.
Ideally, your current probability should include the probabilility-weighted average of all possible future evidence. This is required for consistency of probability across those evidence-producing timelines. Collectively, the set of probabilities of future experiences is your prior. But this article isn't talking about belief or decision-making, it's talking about communication (and perhaps encoding in a limited storage mechanism like a brain). You really don't have the power to do that calculation well, nor to communicate in this level of detail. The idea of a probability range or probability curve is one reasonable (IMO) way to summarize a large set of partly-correlated future evidence.

There is a difference between P("Heads came up") and P("Heads came up" given that "I was just woken up"). Since you will be woken up (memory-less) multiple times if tails came up, the fact that you are just getting woken up gives you information and increases the probability that tails came up.

Let's consider P(H | JustWoken) = P(H and Monday | JustWoken) + P(H and Tuesday | JustWoken) Because I have no information about the scientist's behavior (when he chooses to ask the question), I have to assign equal probabilities (one th... (read more)

I think the solution to the problem depends on what you want to measure. The probability of being tails per wakening is not the same as the probability of being tails per flip or per day.

It would be helpful to define terms first. Capitalism is the system of private ownership of scarce objects. People are scarce, land is scarce, resources are scarce. Even if energy becomes plentiful, the means of production of that energy will remain scarce (solar panels and other capital goods). Even if you can 3D print any object, the resources used as inputs are still scarce.

In short, scarcity will not go away, although it will continue to get significantly better, as more valuable and useful things can be produced with less (less labor, less energy, less resources, etc). Therefore private ownership and voluntary association, aka capitalism, will remain.

I agree with you that the post seems like an ad.

That said, the question "why aren't people doing this more already?" is not very informative to figure if this is a good idea or not.

As economists joke, the twenty dollar bill on the sidewalk is probably too good to be true, otherwise someone would have picked it up already. In reality, new solutions and arrangements are not easy to discover, or to bootstrap, and they take time to become adopted.

What the post describes sounds a lot like fraternal societies, which were a common model for healthcare ... (read more)

Backed. Thanks for starting this project.

What will be the recommended software to consume this?

I'm an Audible member anymore. I'm using a regular podcast iOS app (called RSSRadio) for HPMOR at the moment, but it's a bit manual (have to download each episode from archive). It seems like there would be a way to package an audio book (mp3 files + some XML metadata) and have an app that can consume that. Basically, like Audible but with a simple/open format. But I haven't found any so far.

You'll be able to subscribe to the books in podcast form by adding it to iTunes or your favorite podcatcher. We'll also make it available in .m4b format which can be used with many audiobook players, including Apple devices.