All of Solenoid_Entity's Comments + Replies

Under MWI of QM, anthropics gets weird.

In a single universe interpretation, we can posit biogenesis is rare, but we do know it happened at least once in ~two trillion galaxies worth of stars in ~13 billion years.

In MWI it could be even rarer - with unlimited branches for wild coincidences of chemistry to occur, we’re necessarily living in a branch where such did occur. Allow for argument’s sake that biogenesis is so rare that branches where life is found are tiny in measure. We find ourselves in such a branch, so anthropics and branching kind of gives us t... (read more)

2Vladimir_Nesov
In Transparent Newcomb's Problem, your decision determines whether you existed when you were making the decision. It's not valid to conclude that you exist merely from subjective observation of your own existence, because such observation can take place within counterfactuals, and you wouldn't be able to tell that you are within a counterfactual or actuality other than by reasoning about (or determining) your situation's actuality status. Like math, existence can't be perceived by looking at rocks. So this already doesn't follow, your conclusion is true, but doesn't require MWI. Even without MWI, observing our own existence doesn't tell us anything about the probability of biogenesis (and subsequent development of generally intelligent life).
3Dagon
Anthropics start out weird.    Trying to reason from a single datapoint out of an unknown distribution is always going to be low-information and low-predictive-power.  MWI expands the scope of the unknown distribution (or does it?  It all adds up to normal, right?), but doesn't change the underlying unknowns.
2RHollerith
I disagree. I think the fact that our reality branches a la Everett has no bearing on our probability of biogensis. Consider a second biogenesis that happened recently enough and far away enough that light (i.e., information, causal influence) has not had enough time to travel from it to us. We know such regions of spacetime "recent enough and far away enough" exist and in principle could host life, but since we cannot observe a sign of life or a sign of lack of life from them, they are not relevant to our probability of biogenesis whereas by your logic, they are relevant.

There are definitely some opportunities like that, but being a classical violinist with an orchestra is the first preference by far because it's so much more enjoyable to play the orchestral repertoire, and because having a full-time seat in an orchestra also puts you at the top of every booking agent's list for casual gigs too. Aim high, fail high, seems to be a good approach.

(To be fair, I would make anything sound this extreme, if I was writing about it while in the mood I was in when I wrote this. I love a rant.)

I guess any classical instrument is a device for torturing perfectionists, but violin has a particularly brutal drop-off in sound quality as you reduce your daily focused practice time. Between 'lapsed professional piano' and 'lapsed professional violin' I know which one I'd pick to listen to. You just can't do a few hours of practice a week and play the violin very nicely in tune, or at least I've never met anyone w... (read more)

Ahh good point, sorry I didn't notice that. I'll update the post shortly.

Thanks for checking! The Libsyn feed has been redirected, it's now hosted on BuzzSprout. All new episodes should still be going to all the platforms. Are you having trouble with any of the platforms, or just the Libsyn site itself? That one won't work anymore, unfortunately.

1Bojadła
My problem is that you link libsyn in the post but it doesn't contain all episodes. I was able to find the other episodes but I would still like for the post to be updated.

See here: https://www.lesswrong.com/posts/b9oockXDs2xMdYp66/announcement-ai-narrations-available-for-all-new-lesswrong 

Please share your feedback here or in the comments on that post, it's helpful for our decision-making on this :)

1Timeroutes
The LessWrong Curated Podcast used to be my regular and only check–in with the LessWrong world because it fit well into my schedule. I skip the episodes that are narrated by a robot because I find the emphasis unnatural, which makes the essays confusing for me. All the episodes of the past months have been narrated by the robot, so I have lost contact. I activated my LessWrong account just to ask what is going on. I realise probably not everyone is as picky as me and that the podcast is most likely a non–funded project, so I cannot ask for much. And that it is better to have a robot–narrated podcast than no podcast at all. I still appreciate the work that has been put into it. But if the plan is to keep the podcast robot–narrated, I will have to switch to reading the essays.

Big +1 to playing with others, especially others around the same level or slightly better or worse.
Motivation is one thing, but it's also just... healthier. One's musical 'practice' can't be totally inward-looking, that's when perfectionism starts to bite. Orchestra forces you to compromise and actually learn and perform music, gets you out of the practice room, and generally turbocharges your learning by exposing you to a more varied set of demands on your playing and musicality.
Super hard mode is forming a string quartet with others, since your playing is super exposed and it forces you to stay in time and balance your sound with others. 

Thanks for the feedback!

The audio reflecting updates to the text is relatively easily fixed, and that feature is in the pipeline (though for now user reports are helpful for this.)

There's some fairly complex logic we use for lists — trying to prevent having too many repetitive audio notes, but also keeping those notes when they're helpful. We're still experimenting with it, so thanks for pointing out those formatting issues!


 

You'd probably want to factor in some time for making basic corrections to pronunciation, too.
ElevenLabs is pretty awesome but in my experience can be a little unpredictable with specialist terminology, of which HPMOR has... a lot.
It wouldn't be crazy to do an ElevenLabs version of it with multiple voices etc., but you're looking at significant human time to get that all right.

1[comment deleted]

It's unlikely we'll ever actually GENERATE narrations for every post on LessWrong (distribution of listening time would be extremely long-tailed), but it's plausible if the service continues that we'll be able to enable the player on all LW posts above a certain Karma threshold, as well as certain important sequences.
If you have specific sequences or posts in mind, feel free to send them to us to be added to our list!

This is great to hear, and please feel free to contact us with any other features or improvements you'd find helpful :)

Ha, oops! Yeah, there's a lot of specialist terminology, we find feedback like this really helpful as often we're able to quickly fix this.

Currently we can trigger this if someone requests it, and we have a feature in the pipeline to detect significant changes automatically and re-narrate. 

Double-attrition perfectionism and the violin

An interesting thing about violin is that the learning process seems nearly designed to produce 'tortured perfectionists' as its output.

The first decade of learning operates as a two-pronged selection process that attrits students at different times in their learning journey, requiring perfectionism at some times and tolerance at others. 

You could be boring and argue that it always requires both attention to detail and tolerance of imperfection, simultaneously. You could also argue that there's a fractal, s... (read more)

1drossbucket
I keep thinking about this post! I've been trying to get back into playing violin on and off, and it does a good job of describing why I've found that so hard. I stopped early on in your fourth stage, and my ear is way ahead of my ability to play anything it actually wants to listen to. I guess once I join an orchestra I'll enjoy that enough to get some momentum, but solo playing is pretty unrewarding right now.
6Dagon
Interesting observation.  It matches my impression of MANY skills, from software development to carpentry - the seeking of perfection on many dimensions, along with the tolerance (and SEEKING) of variance and uniqueness in pursuit of a somewhat illegible goal.

I think there may be a typo in the table directly under the heading "Token probability shifts."
If it's not a typo, why are both coefficients positive? Aren't we meant to subtract the vector for ' '?
 

3TurnTrout
Yes, that's a typo. Fixed, thanks!

Edit: removed a bad point. 

If you object strongly to the use of the term UBI in the post, you can replace it with something else. 
Then I make a number of substantive arguments.

Your response so far is 'if it's a UBI it won't suffer from these issues by its very definition.'

My response is 'yes it will, because I believe any UBI policy proposal will degrade into something less than the ideal definition almost immediately when implemented at scale, or just emerge from existing welfare systems piecemeal rather than all at once. Then all the current concerning 'bad things that happen to people who depend on government money' will be issues to consider.

2ChristianKl
LessWrong is more about healthy epistemics than it is about political conclusions. Arguments about bad epistemic like redefining words matter independent of the conclusions. When it comes to taking Australia as an example for how political dissent is treated, it's worth noting that Australia takes actions like COVID-19 Quarantine camps that didn't happen in Europe or the US.  This year in Germany we changed our system in the direction of UBI. While it's still not UBI it does show the political viability of moving the system in that direction. If the FDP wouldn't have been in the government we likely would have moved more into the UBI direction. Apart from the change we see in Germany, how people who receive government money are treated also depends a lot on class. Various companies that get government subsidies are treated well. If you have a scenario where upper-class people think that they might receive UBI in the future you are likely to get laws that are a lot more friendly to UBI recipients.  You yourself said: This is evidence of political movement in the direction of real UBI, but somehow you take it as evidence against UBI. This journalism/civil society/activism is the political muscle pushing for UBI and its power is growing. 

I'm speaking about the policy that's going to be called UBI when it's implemented. You're allowed to discuss e.g. socialism without having to defer to a theoretical socialism that is by definition free of problems.

Anyway, it's a quibble, feel free to find and replace UBI with 'the policy we'll eventually call UBI', it doesn't change the argument I make.

2ChristianKl
Your whole post is about deferring to your idea of theoretical UBI. We do have real-world trials of UBI and there are policies that are used in those trials.  If you want something nontheoretical it makes sense to call the kind of policies that are in UBI trials UBI. Language is valuable. What you are doing is an attempt to remove the current meaning from the term UBI which makes it harder to talk about the underlying policies. 

Where do I call existing welfare systems UBI? That's a misunderstanding of my argument.

My point is that I don't think it's likely that future real-world policies will BE universal. They'll be touted as such, they might even be called UBI, but they won't be universal. I argue they're likely to emerge from existing social welfare systems, or absorb their infrastructure and institutions, or at least their cultural baggage.

I can see the confusion, and maybe I should have put 'UBI' in quotes to indicate that I meant 'the policy I think we'll actually get that people will describe as UBI or something equivalent.'

4ChristianKl
You use the term "UBI dystopia" in the title of the post. If you aren't speaking about UBI that's heavily misleading. 

My point is not to argue that existing welfare systems are UBI. I don't use any non-standard definitions. I don't call existing welfare systems UBI. 

My point is that the real-world policy we're likely to eventually call UBI probably won't actually be universal, and if it emerges as a consequence of more and more people relying on social welfare, or else is associated with social welfare culturally, bad things will likely happen. Then I give some examples of the sort of bad things I mean.

I frequently hear people saying something like "and this is why w

... (read more)
1M. Y. Zuo
Is your point then a 'true UBI' system is practically impossible? And any feasible implementable system shouldn't be called UBI?
1TAG
Why? Every universal healthcare system I have heard of was introduced at a stroke. And is "we" just the US?

I’m writing the original paragraph, and answering a bunch of questions designed to prompt me to reflect.

1M. Y. Zuo
They might not be your reflections though?  GPT-4 has a very large degree of freedom in choosing what questions to ask, and how they're worded. So phrasing it one way or another will prompt different thoughts from you.  And GPT-4, not you, is selecting those parameters. You would need to specify a very narrow range of possibilities, or the exact question for it to ask to avoid this, which would seem to be about as much effort as just doing it entirely yourself. EDIT: Changed wording a bit.

There are a few Obsidian plugins that do similar stuff using LLMs, (they purport to read your notes and help you something something).

I'm thinking of mocking something up over the next week or so that does this 'diary questions' thing in a more interactive way, via the API, from inside Obsidian. 

I also realise how much I sound like Chat-GPT in that comment... dammit

1awg
Disagree. There's still quite a bit of personal nuance to the way you write that wouldn't be present in the typical ChatGPT output. For now ;)

Yeah, I agree with a lot of this, and this privacy concern was actually my main reason to want to switch to Obsidian in the first place, ironically.

I remember in the book In the Age of Surveillance Capitalism there's a framework for thinking about privacy where users knowingly trade away their privacy in exchange for a service which becomes more useful for them as a direct consequence of the privacy tradeoff. So for example, a maps app that remembers where you parked your car. This is contrasted with platforms where the privacy violations aren't 'paid back... (read more)

1Christopher King
This is even stronger for something like LLaMA because you can actually fine-tune it on your personal info or fine-tune it for document retrieval.
1Solenoid_Entity
I also realise how much I sound like Chat-GPT in that comment... dammit

Currently just copy-pasting into GPT-4 via the web interface. I've got it working via the GPT-3 API as well today, but for now I prefer to suffer the inconvenience and get the better model. The questions it asks are MUCH more insightful.

The argument is:

1. You probably can't make it universal.

2. If people can be excluded from the program and depend on it, it creates a power differential that can be abused.

3. There are lots of present-day examples of such abuse, so absent a change, that abuse or similar will continue to exist even if we have a UBI.

1TAG
You can't make it universal at all, or you can't make it universal incrementally?
1Thomas Sepulchre
Thank you!

I'd just explicitly ask the teacher if they're happy with the instrument's setup. It's probably fine, but maybe they'll tell you it needs work. Generally 1/4 instruments aren't going to sound great anyway, but the setup is still very important.

3jefftk
Yes, the teacher is fine with it. When it's time for a larger one that will be a deeper look.

Thanks, great recommendation! I'll check it out for sure.

On the subject of jargon, there's one piece of jargon that I've long found troubling on LW, and that's the reference to 'tech' (for mental techniques/tools/psycho-technologies), which I've seen Duncan use a few times IIRC.

A few issues:
1. It's exactly the same usage as the word 'tech' in the fake scifi 'religion' that must not be named (lest you summon its demons to the forum through the Google portal). They do exercises to give them new mental tools, based on reading the lengthy writings of their founder on how to think, and those lessons/materials/techniq... (read more)

3Duncan Sabien (Deactivated)
To be clear, I'm blurring lines even further than you describe above; I use "tech" for all three of the following categories: * Actual technological development, such as LLMs or new kinds of steel * Mental technologies like TAPs and Gendlin's Focusing * Social tech like duels (obsoleted, but superior to their predecessors) and "I statements"

One question that occurred to me, reading the extended GPT-generated text. (Probably more a curiosity question than a contribution as such...)

To what extent does text generated by GPT-simulated 'agents', then published on the internet (where it may be used in a future dataset to train language models), create a feedback loop? 

Two questions that I see as intuition pumps on this point:

  1. Would it be a bad idea to recursively ask GPT-n "You're a misaligned agent simulated by a language model and your name is [unique identifier]. What would you like to say,
... (read more)
janusΩ4153

I think this is a legitimate problem which we might not be inclined to take as seriously as we should because it sounds absurd.

Would it be a bad idea to recursively ask GPT-n "You're a misaligned agent simulated by a language model (...) if training got really cheap and this process occurred billions of times?

Yes. I think it's likely this would be a very bad idea.

when the corpus of internet text begins to include more text generated only by simulated writers. Does this potentially degrade the ability of future language models to model agents, perform logic

... (read more)

Yeah this is a super useful method and increasingly my go-to for esp. Chinese 

1nnmc
This is a great strategy, Viliam, thank you! Stumbled upon it while trying to pronounce Ali Maow Maalin.

Some websites are great, but I've found they're wrong often enough I usually want to corroborate them with something else.

The correlation between "bothers to have an opinion on correctness of others' writing" and "knows what the correct answer actually is" seems too high.

(Edit: I'm reading between the lines and assuming you're saying you think the cohort of people who actually care enough about faze/phase to be judgemental about it, but don't themselves know the correct spelling is 'faze', is small.)

This is very interesting. I certainly agree this is our point of difference – I think there's a big cohort out there with strong, judgey opinions about 'correctness' and an active... (read more)

Just pitching in on the last two: there's an abbreviated register of speech in English called 'note-taking register' that has crept its way into a lot of parts of speech and writing, including website navigation. Dropping the definite article (or most articles in general) is a core part of that register.

Note taking = abbreviated English register. Has crept into parts of speech, writing inc. website nav. Dropping definite article core part of register.

I suspect dropping the definite article in 'refresh page' is not related to definiteness, it's a linguistic... (read more)

Software: Newsfeed Eradicator + Leechblock NG

Need: Resilient self-control/anti-akrasia for web browsing.

Other programs I've tried: Stayfocusd, Forest

The problem with Stayfocusd and any website blocker is that, invariably, you have to navigate to a given tweet or youtube video or facebook profile, for legitimate reasons, and it means you have to go and deactivate the plugin. This is bad because 1. it trains you to do this action and 2. It incentivises you to avoid making deactivating the plugin too tricky.

Newsfeed Eradicator kills only the problem parts of ... (read more)

I do the SSC Podcast; one of my patreon supporters said he'd be really keen to have this as an audiobook. I'd certainly be keen to get an idea of the demand for that and could potentially make it happen if it seemed like it would be useful. If you wanted to chat about it you can get me on slatestarpodcast@gmail.com. Thanks!